Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Influence of social housing conditions on mouse mammary tumor growth and behavior Grimm, Michele S. 1993

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1993_fall_grimm_michele.pdf [ 4.39MB ]
JSON: 831-1.0098799.json
JSON-LD: 831-1.0098799-ld.json
RDF/XML (Pretty): 831-1.0098799-rdf.xml
RDF/JSON: 831-1.0098799-rdf.json
Turtle: 831-1.0098799-turtle.txt
N-Triples: 831-1.0098799-rdf-ntriples.txt
Original Record: 831-1.0098799-source.json
Full Text

Full Text

to the required standardINFLUENCE OF SOCIAL HOUSING CONDITIONS ONMOUSE MAMMARY TUMOR GROWTH AND BEHAVIORbyMICHELE SAULT GRIMMB.Sc., Bates College, 1989A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinTHE FACULTY OF GRADUATE STUDIES(Program in Neurosciences)We accept this thesis as conformingTHE UNIVERSITY OF BRITISH COLUMBIAAugust 1993© Michele Sault Grimm, 1993In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of /Vdct ro Sci eric e_The University of British ColumbiaVancouver, CanadaDate DE-6 (2/88)iiABSTRACTEpidemiological evidence suggests that human breast cancermay be influenced by a number of psychosocial factors. In thepresent study we investigated psychosocial variables that mayinfluence growth of the Shionogi mouse mammary carcinoma (SC115).Change in social housing condition, direction of change[individual to group (IG) or group to individual (GI)), groupsize, sibling relationship and dominance were the major variablesexamined. Male mice were housed from weaning in groups (G) orindividually housed (I), and at age 2-4 months mice were injectedwith 3 x 106 tumor cells/mouse and placed into four treatmentconditions. In the first two treatments, animals remained intheir rearing conditions, group (GG) or individual (II); in theremaining treatments, animals were rehoused from group toindividual (GI) or individual to group (IG). All animals wereexposed to daily novelty stress in the post-injection period. Thefrequency of aggressive, defensive, social, nonsocial, sleep andfighting behaviors were measured three times per week in the homecage for about 10 days before and 17-18 days after tumor cellinjection and rehousing. Behavioral frequencies were alsomeasured immediately after each exposure to daily novelty stress.Rehousing from group to individual (GI) produced the highesttumor growth rates, and rehousing from individual to group (IG)produced the lowest tumor growth rates. Animals remaining intheir rearing conditions (GG and II) showed tumor growth ratesintermediate to the rehoused animals. Thus, changes in housingiiirather than individual or group housing per se, were found to bemodulators of tumor growth rate. In the IG treatment, dominantmice showed faster tumor growth than subordinate mice, whereas inthe GG treatment, dominant mice showed slower tumor growth thansubordinate mice. Within the IG treatment, nonsibling groups ofthree (IN3) showed the highest tumor growth rates. Pre-injection,individually housed animals showed more nonsocial behavior andless sleep than group housed animals, and groups of three showedmore social behaviors than groups of five. In both the pre- andpost-injection groups, dominant animals displayed a more activerole than subordinate animals in agonistic, social and nonsocialbehaviors. Post-injection, overall behavioral activity washighest on the day of injection and rehousing. GI mice showedmore nonsocial behavior and less sleep than IG mice. For grouphoused animals, IG mice showed more defensive and fightingbehaviors than GG mice. Active, passive and total socialbehaviors were highest in the IS5 group, and fights were highestin groups of five. Differences among groups in tumor growth rateand behavior were discussed in terms of arousal, social dominance,and stability of social status.ivTABLE OF CONTENTSAbstract ^  iiTable of Contents ^  ivList of Tables  viList of Figures ^  viiAcknowledgements  xINTRODUCTION ^1A. Psychosocial factors and human breast cancer ^11. Stress Definitions ^12. Adulthood stress ^23. Childhood stress ^24. Personality and coping: Cancer onset ^35. Personality and coping: Cancer progression ^46. Social support ^5B. Stress and cancer studies in animals ^61. Overview of the classical stress response ^72. Physical and environmental stressors and cancer 73. Stressor controllability and predictability ^94. Stressor chronicity^  10D. Social housing conditions: Physiological andbehavioral effects^  111. GI housing conditions  112. II housing conditions^  123. Social housing conditions and cancer^ 124. Social housing as a modulator of stress^ 155. Crowding^  16E. Agonistic behavior and social dominance^ 161. Isolation-induced fighting^  162. Effects of fighting on cancer  173. Immunological effects of fighting^ 18F. Background to the present study  19MATERIALS AND METHODS^  22A. Animal-tumor model^  221. Tumor propagation  222. Experimental animals^  22B. Behavioral observations  25C. Dominance^  26D. Data analysis  291. Tumor growth rate^  292. Pre-injection behavior  303. Post-injection behavior^  30RESULTS^  32A. Tumor growth rate^  32B. Pre-injection behavior  351. Nonsocial behavior^  35V2. Sleep/rest^  353. Aggressive and defensive behavior^ 424. Social behavior  425. Fights/attacks  42C. Post-injection behavior: 1900 hr observations,days 0 and 1^  511. Nonsocial behavior^  512. Sleep/rest  513. Aggressive and defensive behavior^ 544. Social behavior  545. Fights/attacks^  59D. Post-injection behavior: Post-novelty stressobservations, day 1  591. Nonsocial behavior  592. Sleep/rest^  643. Aggressive and defensive behavior^ 644. Social behavior  645. Fights/attacks  71E. Post-injection behavior: 1900 hr observations,days 1-7^  711. Nonsocial behavior^  712. Sleep/rest^  743. Aggressive and defensive behavior^ 814. Social behavior  865. Fights/attacks  86F. Post-injection behavior: Post-novelty stressobservations, days 1-12^  911. Nonsocial behavior  912. Sleep/rest^  913. Aggressive and defensive behavior^ 1004. Social behavior  1035. Fights/attacks  106DISCUSSION^  107A. Tumor growth rate^  107B. Pre-injection behavior  110C. Post-injection behavior^  1131. Overall effects  1132. Nonsocial behavior and sleep/rest^ 1143. Dominance^  1154. Aggressive behavior^  1155. Fights/attacks  1166. Defensive behavior  1177. Social behaviors  1208. Conclusion^  122REFERENCES^  126viLIST OF TABLESTable^ Page1. Housing Conditions ^242. Behaviors and Definitions ^27vi iLIST OF FIGURESFigure^ Page1. Tumor weights for all mice by group ^332. Tumor weights for mice in the IG treatment bysize and sibling ^363. Tumor weights for mice in the IG and GGtreatments by treatment and dominance^ 384. Pre-injection nonsocial behavior for all miceby size (A), for group housed mice by dominance(B), and for group housed mice by size andsibling (C)  ^405. Pre-injection sleep for all mice by size (A),for group housed mice by dominance (B), andfor group housed mice by size and sibling (C)....^436. Pre-injection aggressive, defensive, activesocial and passive social behavior for grouphoused mice by dominance^  457. Pre-injection active (A), passive (B) and total(C) social behaviors for group housed mice bysize ^478. Pre-injection fights and attacks for grouphoused mice by size and days^  499. Post-injection nonsocial behavior in the firsttwo observations at 1900 hr for group housedmice (A) and individually housed mice (B) bygroup; and for mice which experienced a changein housing, by change, sibling and dominance (C).. 5210. Post-injection sleep in the first twoobservations at 1900 hr for mice whichexperienced a change in housing, by change,sibling and days^  5511. Post-injection defensive behaviors in the firsttwo observations at 1900 hr for group housedmice by group (A); and aggressive behaviors inthe first two observations at 1900 hr for micein the IG treatment by dominance, sibling anddays^  5712. Post-injection passive (A) and total (B) socialbehaviors in the first two observations at 1900hr for group housed mice by group^  60viii13. Post-injection fights and attacks in the firsttwo observations at 1900 hr for group housedmice by group ^6214. Post-injection nonsocial behavior in the firstpost-novelty stress observation for group housedmice (A) and individually housed mice (B) bygroup ^6515. Post-injection sleep in the first post-noveltystress observation for group housed mice (A)and individually housed mice (B) by group^ 6716. Post-injection defensive behavior in the firstpost-novelty stress observation for group housedmice by group ^6917. Post-injection nonsocial behavior in the eightobservations at 1900 hr for group housed mice(A) and individually housed mice (B) by group^ 7218. Post-injection nonsocial behavior in the eightobservations at 1900 hr for mice which experienceda change in housing; by change and dominance(A), and by change, dominance and days (B) ^ 7519. Post-injection sleep in the eight observationsat 1900 hr for group housed mice (A) andindividually housed mice (B) by group^ 7720. Post-injection sleep in the eight observationsat 1900 hr for mice which experienced a changein housing; by change and dominance and by changeand size (B) ^  7921. Post-injection aggressive and defensive behaviorsin the eight observations at 1900 hr for grouphoused mice by days^  8222. Post-injection aggressive behavior in the eightobservations at 1900 hr for mice in the IGtreatment by size, dominance and days^ 8423. Post-injection active (A) and passive (B) socialbehaviors in the eight observations at 1900 hrfor group housed mice by dominance^ 8724. Post-injection active, passive and total socialbehaviors in the eight observations at 1900 hrfor group housed mice by days^  8925.^Post-injection fights and attacks over the eightobservations at 1900 hr for mice in the IGixtreatment by size and days^  9226. Post-injection nonsocial behavior in the 12post-novelty stress observations at 1900 hr forgroup housed mice (A) and individually housedmice (B) by group ^9427. Post-injection nonsocial behavior in the 12post-novelty stress observations at 1900 hr formice which experienced a change in housing bychange and days^  9628. Post-injection sleep in the 12 post-noveltystress observations at 1900 hr for group housedgroup housed mice (A) and individually housedmice (B) by group ^9829. Post-injection defensive behavior in the 12post-novelty stress observations for grouphoused mice by group and dominance^ 10130.^Post-injection active, passive and total socialbehaviors over the 12 post-novelty stressobservations for group housed mice by days^ 104xACKNOWLEDGEMENTSI would like to extend my sincere thanks to Dr. JoanneWeinberg and Dr. Joanne Emerman for their guidance and supportthrough every stage of this project. In addition, I greatlyappreciate the financial support I received through Dr. Weinberg'slaboratory. I would also like to thank Shannon Wilson and GerryRowse, who introduced me to the laboratory and contributed theirexpert technical assistance to this study. Finally, many thanksto my husband, Barrett, who provided much-needed encouragementduring my graduate studies.1INTRODUCTIONPsychosocial factors and human breast cancerA role of psychosocial factors (PF) in the onset andprogression of breast cancer has long been speculated, but theevidence has been mainly anecdotal until recently (Dorian &Garfinkel, 1987). Recent evidence in the field of psychosocialoncology indicates that stress, personality, coping and socialsupport may modulate breast cancer incidence and growth.Stress definitions. Hans Selye (1936) originally described stressas a nonspecific syndrome of adaptive responses to a wide range ofphysical factors or "stressors," characterized by activation ofthe hypothalamic- pituitary-adrenal (HPA) axis, and resulting ina specific pattern of physiological responses.It is now known that responses to stress are not uniformacross situations. Behavioral coping is an important modulator ofthe stress response in humans and animals. According toGreenberg, Dyck and Sandler (1984), the response to a stressorinitially consists of active behavioral control processes, but ifthe behavioral responses are ineffective, physiological adaptationprocesses are activated. Contemporary writers often define stressin terms of the organism's inability to cope with environmentaldemands (Maes, Vingerhoets & Van Heck, 1987; Puglisi-Allegra,Cabib & Mele, 1989). The phrase "psychosocial factors"encompasses the concepts of stress and coping, as well asindividual and interpersonal factors; thus it is a useful term forthe variety of factors which may be relevant to disease states.2Adulthood Stress. Highly stressful life events have beenassociated with increased breast cancer risk. Questionnaires aretypically administered before diagnosis (semiprospective design),in an attempt to eliminate the knowledge of health status. Breastcancer incidence has been found to be associated with death of afamily member or close friend during the past two to three years(Cooper, Cooper & Faragher, 1989; Scherg, Cramer & Blohmke, 1981).Although one study found a positive association of divorce withcancer in women (Scherg & Blohmke, 1988), an extensive studycomparing 1782 women with breast cancer to healthy women revealedno association between presence of breast cancer and maritalstatus, length of marriage, divorce, or widowhood (Ewertz, 1986).Adulthood stressors could produce physiological changes which maypredispose the individual to a neoplastic change, or mayaccelerate the clinical manifestation of cancer which is alreadypresent. For example, events such as bereavement may lead toreduced natural killer (NK) cell activity (Irwin et al., 1987) orreduded T-lymphocyte activity (Bartrop et al., 1977) in women. NKcells, which comprise a class of non-antigen-dependentT-lymphocytes, have been recognized during the past decade to beimportant immunological defenses against tumors (Riley,Fitzmaurice & Spackman, 1981).Childhood Stress. Traumatic life events during childhood andadolescence, particularly the loss of a close relative, may beassociated with later diagnosis of cancer in women (Scherg, 1987;Scherg & Blohmke 1988). Women who develop breast or gynecologicalcancers have also reported a higher incidence of traumatically3experienced psychosocial and socioeconomic strains in theirparental homes compared to age-matched controls (Gehde &Baltrusch, 1990).In addition, the nature of early family relationships mayinfluence cancer risk. A major prospective study in male medicalstudents revealed that, compared to healthy subjects or subjectswho developed essential hypertension or coronary heart disease,those who later developed cancer reported lower closeness toparents during childhood and youth (Thomas & Duszynski, 1974). Arecent study on a female population revealed similar differencesin relationships with parents between cancer-prone and healthyindividuals (Gehde & Baltrusch 1990). Experience of severe orchronic stress during childhood could generate a predispositiontoward certain coping styles in response to future life stressors.Personality and coping: Cancer onset. Coping styles modulateindividual physiological responses to stress (Cohen, 1984; Miller,1987). The "Type A" or coronary prone behavior pattern, marked byhostility, expression of anger, and competitiveness, has been usedto identify persons at risk for cardiovascular disease (Baltrusch,Stangel & Titze, 1991). Scherg (1987) found that women withbreast cancer tended to show less Type A behavior than healthywomen, as well as higher committment to social and religiousnorms. Emotionally, breast cancer patients have been shown toexhibit high rationality (Wirsching et al. 1982), extremesuppression of anger (Greer & Morris, 1975), and low anxiety(Scherg, 1987). However, some studies have not supported thesefindings (Greer, Morris & Pettingale, 1979; Hislop et al., 1987).4A retrospective study of 100 women with breast cancer indicated noassociation between breast cancer and extraversion, neuroticism,or number and severity of life events (Priestman, Priestman &Bradshaw, 1985). Other features associated with breast cancer areavoidance of conflicts and harmonizing behavior (Wirsching et al.,1982). Many of these charactereistics have recently been groupedinto a "Type C" or psychosocial cancer risk pattern (Baltrusch,Stangel, & Titze, 1991). It has also been found that these womenperceive life events as more severe than healthy women or womenwith benign breast disease (Cooper, Cooper & Faragher, 1989),which could result in a greater need to suppress the emotionalresponse to stressful events. In addition, a prospective study of2,264 subjects, revealed a strong association of depressed moodwith subsequent cancer, but only among cigarrette smokers (Linkins& Comstock 1990).Personality and coping: Cancer progression. Cancer itself is amajor stressor, and treatments such as surgery, chemotherapy andradiation place additional demands on the patient. Ramirez andcolleagues (1989) demonstrated that severely threatening lifeevents after surgical treatment were associated with the firstrecurrence of breast cancer. Other studies have shown that howone responds to having cancer can influence the progression of thedisease. Greer, Morris & Pettingale (1979), in a study of 69early stage breast cancer patients, found that denial or a"fighting spirit" in the initial response to diagnosis wasassociated with recurrence-free survival at five years aftersurgical treatment. Conversely, reactions of stoic acceptance or5helplessness were negatively associated with disease-freesurvival. In another study, low cognitive disturbance wasassociated with disease-free survival four years after diagnosis(Hislop et al., 1987). In addition, extreme life stressors maydetermine coping style; for example, a history of traumaticexperiences has been shown to produce extreme psychologicaldistress in response to cancer (Baider, Peretz & Kaplan De-Nour,1992).Social support. Social support, which can be described asemotional or instrumental aid (Maes, Vingerhoets & Van Heck,1987),is sometimes viewed as a way to reduce the negative impactof life stressors (Levy et al., 1990). A prospective study ofapproximately 7000 adults showed that people with few social orcommunity ties were more likely to die prematurely from variousdiseases than others (Berkman & Syme, 1979). Conversely, in womenwith breast cancer, frequent involvement in social activities maybe related to longer disease-free survival (Hislop et al., 1987).Levy and colleagues have demonstrated that in women with breastcancer, lack of social support and certain coping styles isassociated with low NK cell activity, which in turn predicts poorprognosis (Levy, 1985; Levy et al., 1985, 1990).In a study of geriatric subjects, one month of relaxationtraining resulted in significant increases in NK cell activity aswell as decreases in self-rated distress (Kiecolt-Glaser et al.,1985a). A recent study indicated that one year of weekly supportgroup therapy with self-hypnosis for pain significantly increasedsurvival in breast cancer patients (Spiegel et al., 1989).6Psychotherapy appears to increase survival of breast cancerpatients by eight to 18 months on the average, and may improveresponses to chemotherapy and quality of life (Barinaga 1989;Grossarth-Maticek & Eysenck, 1989).Stress and cancer studies in animalsIn human studies, issues such as inability to determine whenthe disease was initiated relative to reported stressful events,unreliability of subjects' memory or self-reports, and, inretrospective studies, subjects' knowledge that they have cancer,may complicate interpretation of data (Sklar & Anisman, 1981).Animal models provide a way to investigate the relationshipbetween stress and tumor growth under more controlled conditions.A large number of investigators have found tumor-enhancing orsuppressing effects using physical stressors such as electricshock, cold, or restraint. Others utilize psychosocial stressorssuch as changes in housing as models of psychosocial stressors inhumans. For example, separating an animal from its usual socialgroup and placing it in isolation can be used as a model ofseparation stress in humans; e.g. death of a family member or lossof social support. In animal models, however, the relationshipbetween stress and cancer is also complex. Both stress-inducesincreased and stress-induced decreases in tumor incuction andgrowth have been reported in experimental animals. Numerousexperimental variables may influence the resulting physiologicaleffects (Newberry, Liebelt & Boyle, 1984; Sklar & Anisman, 1981).Relevant stressor-related factors include the type of stressor7(Rodriguez-Enchandia et al., 1988), intensity (Orr et al., 1990),time of application (Burchfield et al. 1978), chronicity, andcontrollability. Additional variables which modulate experimentaloutcomes are species and strain of animal (Chiueh & McCarty, 1981;Jurcovicova et al., 1984; Shanks et al., 1990), and thephysiological effects measured, such as type of tumor (Justice,1985). In addition, fighting and social dominance among maleshave been shown to affect the neuroendocrine and immune systemsand to modulate cancer development.Overview of the classical stress response. One of the mostconsistent indicators of stress is glucocorticoid secretion fromthe adrenal cortex; the primary glucocorticoid in humans iscortisol, and in rodents, corticosterone (CS) (Monjan, 1981).High circulating CS levels during stress cause functionalinhibition of macrophages, reduction in lymphocyte number,supression of T-cell growth factor (TCGF), and suppression oflymphokines (Riley, Fitzmaurice & Spackman, 1981).Physical and environmental stressors and cancer. Bothcarcinogen-induced and transplantable tumors have been shown to beinfluenced by physical stressors. Neiburgs and colleagues (1979)demonstrated that 90 or 150 days of electric shock, cold airtemperature, or handling increased the incidence and size ofmammary tumors induced by the carcinogen dimethylbenzanthracene(DMBA) in rats, and depressed lymphocyte function. However, asubstantial number of studies have shown that stress can delay theappearance of carcinogen-induced tumors (Fox, 1981).Repeated daily footshock (four or 14 days) reduced survival8time in rats bearing a mammary ascites tumor, whether administeredbefore or after tumor cell injection (Lewis et al., 1983/84).Cage-shaking stress (30 min/day for 35 days) led to earlierappearance of transplanted melanoma in hamsters (Temoshok et al.,1987). Three days of stress administered twice daily led todecreased rejections of leukemia in mice (Teshima & Kubo, 1984).Endogenous opioids (endorphins and enkephalins) released from theadrenal medulla during stress may perform a mediating role inreduced NK activity and defense against tumors (Greenberg, Dyckand Sandler, 1984; Lewis et al., 1983/84; Tejwani et al., 1991).Steplewski and colleagues (1985) showed that chronic orrepeated stress over a number of days increased the growth oftransplantable tumors, but recovery from stress suppressed tumorgrowth. Growth of a transplanted mammary adenocarcinoma wassignificantly enhanced in female rats stressed by daily restraint(three hr/day) for 11 days. A recovery period of 12 days afterstress markedly reduced tumor burden relative to unstressedcontrols, and increased numbers of circulating lymphocytes and NKcell activity.Environmental stressors which are not physically severe mayalso influence cancer. Riley, Fitzmaurice and Spackman (1981)have demonstrated that standard animal colony conditions representan undetected source of low-level stress in mice, which maydramatically interfere with presumably baseline physiologicmeasurements such as plasma CS levels. Greenman, Kodell & Sheldon(1984) conducted an extensive study of the effects of cage shelflevel on tumor growth, using over 20,000 female BALB/c mice. Mice9housed on the top shelf exhibited longer survival, as well assignificantly fewer liver and bladder tumors induced by2-acetylaminofluorene (2-AAF) and significantly lower incidence offive out of six spontaneous neoplasms tested.Stressor controllability and predictability. Studies indicatethat an organism's control over a stressor, i.e. the ability toescape, alleviates the tumor-enhancing and immunosuppressiveeffects of stress. It is well established that inescapable butnot escapable shock leads to a variety of behavioral andphysiological deficits in rodents including increased freezingbehavior in the open field (Holson et al., 1988) and suppressionof lymphocyte proliferative responses (Laudenslager et al., 1988).Acute inescapable shock following tumor injection markedlyincreased growth of the P815 mastocytoma in mice, while the sameamount of escapable shock had no effect (Sklar & Anisman, 1979).Moreover, tumor rejection in rats bearing Walker 256 sarcoma wasdecreased in rats exposed to inescapable but not escapable shock(Visintainer, Volpicelli & Seligman, 1982). Interestingly, Marsh,Miller & Lamson (1959) demonstrated early on that active shockavoidance conditioning in a shuttlebox resulted in inhibition ofthe growth of Ehrlich carcinoma in mice.Repeated inescapable shock (15 days) results in increasedhypothalamic norepinephrine (NE) (Irwin et al., 1986). Weiss etal. (1981) have shown that escapable shock results in reductionsin regional brain concentrations of NE, including hypothalamic NE;as well as increases in dopamine (DA) in the anterior cortex andserotonin (5-HT) in the brain stem. Investigators have failed to10find differences in plasma CS or ACTH between rats exposed toescapable or inescapable shock (Laudenslager et al., 1973; Maier& Seligman, 1976).Two studies indicated that the physiological effects ofstress can be alleviated by allowing animals to anticipate thestressor. With inescapable shock, rats receiving a warning signalbefore each upcoming shock had only 20% as many stomach lesions asrats that did not receive a signal, and most showed lower levelsof brain NE (Weiss, 1970). Aarstad, Thiele & Seljelid (1991)showed that if mice were stressed by cold water immersion twicedaily at variable time points, the Con-A response to mitogen wassuppressed for 14 days, but in mice stressed at fixed time pointsthe immune response was only suppressed for eight days. It isclear that coping with a stressor by escape or predictability mayreduce or eliminate the deleterious physiological effects normallyassociated with stressors.Stressor chronicity. Acute and chronic (repeated) exposure tostressors often produce different physiological responses. Forexample, a single session of shock following mastocytoma cellinjection caused increased tumor growth in mice, but five to 10days of shock sessions had no effect (Sklar, Bruto & Anisman,1981). Elevations of plasma CS occur in response to one sessionof stress (Monjan and Collector, 1977), with a return to baselineCS by the end of about two weeks of stress. Recently, Aarstad andcolleagues (1991) compared several acute and chronic regimens ofdaily cold water immersion stress in mice. Spleen cellstimulation by Concanavalin A (Con-A) was decreased after one day11of stress in mice stressed once daily as opposed to after repeated(eight or 14 days) stress in mice stressed twice daily. After 14days of stress, the mitogen-stimulated proliferation was increasedif mice were stresed once daily but decreased if stressed twicedaily.Social housing conditions: Physiological and behavioral effectsSocial housing conditions influence a wide variety ofphysiological systems as well as behavior. Isolation (individualhousing), group housing, crowding, and changes in housingconditions have also been investigated for their influence oncancer. Changes in housing condition have been utilized tocompare animals housed continuously in groups (GG), housedcontinuously in individual cages (II), or rehoused from group toindividual (GI) or individual to group (IG).GI housing conditions. In adult mice previously housed in groups,a GI change in housing followed by a long period of isolation(usually several weeks) results in endocrine changes such asincreased adrenal weight and plasma CS, and increased gonadalweight and activity compared to group housed mice (GG) (Brain,1975, Sayegh et al., 1990). Elevated plasma CS levels followacute exposure to physical (Holson, 1988; Irwin et al., 1986; Kantet al., 1983a, 1983b) and psychological (Gentsch et al., 1981;Kant et al., 1983a) stressors, indicating that adulthood isolationrepresents a significant source of stress. Alterations in brainneurotransmitter activity, including dopamine, serotonin (5-HT),and acetylcholine, have been shown to accompany the GI condition12in the mouse (Valzelli, 1973). Behaviorally, GI mice exhibithyperreactivity to stimuli (Brain, 1975), decreased locomotion andexploration, and learning deficits (Valzelli, 1969, 1973).Reductions in brain 5-HT activity have been implicated in some ofthe behavioral deficits (Brain, 1975), and restoration of brain 5-HT activity by chlordiazepoxide, an anti-anxiety drug, alleviatesthe locomotor and exploration deficits (Valzelli, 1969). Thesedata provide further evidence that GI housing is stressful.II housing conditions. Extensive studies of individual housing inrats and mice indicate that the effects of isolation rearing(isolation from weaning into adulthood; II condition) aredifferent from those of adult isolation (GI).^II rats alsoexhibit learning deficits, but show increased rather thandecreased locomotor behavior (Holson, 1988; Morgan, 1973). IIrats do not show increased behavioral timidity or emotionalitycompared to socially-reared rats (GG) (Morgan, 1973; Holson,1988).Social housing conditions and cancer. Steplewski, Goldman & Vogel(1987) investigated the effect of a GI change in housing onmammary carcinoma growth in rats. Animals were rehoused on thefirst day of DMBA injections or on the day of tumor cellinjection. Animals that experienced a change in housing fromgroup to individual (GI) exhibited faster growth of transplantedor DMBA-induced mammary carcinomas than did rats which remained ingroups (GG) or remained individually housed (II). In contrast,mice injected with cells of the Shionogi mammary carcinoma andexperiencing a GI change in housing, or mice remaining13individually housed (II), showed faster tumor growth than did GGmice (Emerman & Weinberg, 1989; Weinberg & Emerman, 1989). In onestudy, GI rehousing (10/cage, grouped for seven weeks) was shownto have a beneficial effect; GI rehousing prolonged survival inmice with melanoma, leukemia, or ascites tumors (Dechambre &Gosse, 1973).^The large group size may have affected thedirection of the results, since social structures are much lessstable in large than small groups of mice (Ebbesen et al., 1991).Transplanted mammary tumors grew more slowly in mice changedfrom individual to group housing (IG) than in mice remaining intheir rearing groups (GG) (Emerman & Weinberg, 1989; Weinberg &Emerman, 1989). In contrast, Sklar and Anisman (1980) foundincreased mastocytoma growth in IG mice (five per cage) comparedto GG mice. The IG mice in the latter study were isolated foronly two weeks prior to tumor transplantation; they actuallyexperienced a GI change in housing followed by an IG change. Itis likely that two changes in housing conditions result in morestress than a single change in housing. In addition, female miceinjected with lymphosarcoma cells, and housed in a "populationcage" (consisting of three standard cages connected by tubes)showed reduced survival times compared to individually housedmice, despite equal population density (Riley, Fitzmaurice &Spackman, 1981). It is possible that social interactions maystimulate cancer in female mice but inhibit cancer in males. Insummary, GI and IG changes in social housing have both been shownto enhance or suppress tumor growth, and II housing conditions may14or may not enhance tumor growth. The discrepancies among studiesmay depend on type of cancer, group size, duration of grouphousing prior to tumor cell injection, or sex, species or strainof animals.Changes in housing condition may influence tumor growth rateby altering immune functions. Stress-induced immunosuppressionhas been well documented in humans (Glaser et al., 1986; Workman& LaVia, 1987) and experimental animals (Kandil & Borysenko 1987,1988; Laudenslager et al., 1988; Teshima & Kubo, 1984). Despitethe indicators that a GI change in housing is stressful, thiscondition tends to stimulate immune system activity (Jessop, Gale& Bayer, 1987,1988; Rabin et al., 1987). However, other studiesfound no immunological effects of GI housing (Plaut, Friedman &Grota, 1971; Rabin, Lyte & Hammill, 1987).Male rats in a GI condition showed changes in mitogen-inducedlymphocyte proliferation and plasma CS, dependent upon theduration of isolation (Jessop & Bayer 1989). During the firstweek of isolation, T and B lymphocyte responses were depressed,but within two weeks, at least a two-fold immunoenhancement wasobserved, still present at 35 days after isolation.A study in mice indicated trends toward higher splenic NKcell activity housed for three weeks in isolation, followed by oneweek in groups of 5 (IG), and lower NK activity in mice exposed tothe reverse treatment (GI) (Hoffman-Goetz, Simpson & Arumugam,1991). However, in a study using the Shionogi mouse mammarytumor, differences in NK cell activity did not appear to mediatedifferences in tumor growth rate induced by GI or IG changes in15housing condition (Rowse et al., 1990).Social housing as a modulator of stress. Social housingconditions modulate the effects of other stressors on physicalfunctioning and cancer development. For example, mastocytomagrowth following acute shock was decreased in mice isolated fortwo weeks before tumor cell injection compared to grouped mice(Sklar &Anisman 1980), which indicates a stress-protective effectof adulthood isolation. In another study, mice were immunizedwith the bacterium Corneybacterium parvum (C. parvum), whichsuppresses P815 mastocytoma growth, and were exposed to acutefootshock. Shock was found to eliminate the tumor-suppressiveeffect of C. parvum in mice housed in small groups, but had noeffect on mice rehoused in groups from isolation (Turney, Harmsen&Jarpe, 1986). Interestingly, isolation rearing protects againststress-induced leukocytopenia (reduction in leukocyte number)(Simmel, Wright & Smith 1974). Cold water immersion stressreduced spleen cell mitogenesis in group-housed mice, and theonset of these reductions varied with number of mice per cage. Ingroups of two mice, decreased mitogenesis occurred after 14 daysof stress, but in groups of four mice, the reductions inmitogenesis began at eight days of stress and continued to 14 days(Aarstad et al., 1991). Finally, in individually housed mice,immunosuppression began one day after stress.Suppression of secondary antibody activity occured in GI mice(Emerman & Weinberg, 1989; Weinberg & Emerman, 1989), in bothtumor bearing and non-tumor bearing mice exposed to daily noveltystress. High secondary antibody responses were associated with16slow tumor growth rates. Changes in immune functions may be morelikely to occur in animals exposed to stress; one study showed nodifferences in spleen cell responses to mitogens in unstressed GImice (Aarstad, Thiele & Sejelid, 1991).In animals exposed to daily novelty stress, splenic T cellactivity was reduced in IG mice following 3 weeks of grouphousing, but only in those without tumors (Emerman & Weinberg,1989; Weinberg & Emerman, 1989).Crowding. The effects of crowding on HPA activity arecontradictory. Gamallo and colleagues (1986) found that isolationor crowding of rats for 6 weeks led to higher CS levels thanhousing in small groups. However, other studies in male ratsfound no changes in plasma CS (Armario, Garcia-Marquez & John,1987) or other physiological variables (Riley, Fitzmaurice &Spackman, 1981) after several weeks of crowding. Crowding of malemice significantly reduced survival compared to males crowded withfemales or housed individually (cf. Ebbesen, 1991).Agonistic behavior and social dominanceIsolation-induced fighting. Increased aggression toward othermale mice also occurs following several weeks of adult isolation,in many but not all strains of mice (Valzelli, 1969; 1973).Increasing the period of isolation leads to increasedaggressiveness and social interaction (Brain, 1975; Lister &Hilakivi, 1988). When pairs of male outbred mice were tested overa nine-day period, isolated mice showed decreased fighting overdays, and group-housed mice (four per cage) showed increased17fighting over days (Burright et al., 1988).A unique experiment found effects of familiarity onaggressive behavior in male deer mice (Dewsbury, 1988). Male micewere raised in groups or pairs. For at least two months beforetesting, each male was either housed in a pair with another malefrom its rearing group, or isolated. They were tested in a largeenclosure in groups of two males and two females. When the twomales were familiar; i.e., from the same cage, they showed lessaggression than unfamiliar males. However, since the unfamiliarmales were also isolated, their increased aggression could be dueto individual housing rather than lack of familiarity per se.Sibling relationship was also examined, but had no effects onaggressive behavior.Effects of fighting on cancer. Agonistic behavior is a term usedto indicate aggressive and defensive behaviors. Agonisticbehavior influences physiological processes related to cancer andcan also secondarily modulate physiological responses to otherstressors. Spontaneous home-cage fighting in mice was associatedwith reduced growth of transplanted P815 mastocytoma in male mice(Sklar & Anisman, 1980), and Moloney virus-induced sarcoma infemale mice (Amkraut & Solomon, 1972).In a study by Weiss and colleagues (1976), rats shocked inpairs and thus given the opportunity to fight developed fewergastric lesions that rats shocked alone, even when a translucentbarrier prevented actual contact between the animals, indicatingthat the psychological component of fighting reduced the effectsof shock. In female hamsters exposed to daily cage-shaking stress18(30 min/day for 35 days), delayed melanoma appearance waspositively correlated with fighting, display of social behavior,and dominance (Temoshok et al., 1987). Conversely, early tumorappearance was correlated with passive or "downtrodden" behavior.Subordinance and lack of social heirarchy have been shown toresult in a higher incidence of virus-induced leukemia in malemice of the DBA/2 strain. When caged in groups of three, dominantmale mice survived longer than subordinate group members, andoverall mice in groups of three survived more than mice in groupsof nine, among whom no dominance order was established. Afterchallenge with Moloney virus, leukemia developed among subordinatemice in groups of three and in members of groups of nine, but notin dominant animals in groups of three (Ebbesen et al., 1991).Immunological effects of fighting. Social defeat and subordinancegenerally produce immunosuppression in rats and mice. Micedefeated in a single attack by a dominant mouse have been found toexhibit lower T-cell proliferation and IL-2 production compared todominant or nonfought mice (Hardy et al., 1990). Similar resultswere obtained for rats defeated by a trained fighter rat.Following social defeat, rats showed reduced production ofimmunoglobulin E compared to nonfought controls (Ito et al.,1983), and showed reduced serum antibody responses to antigen(Fleshner et al., 1989). Subordinate male mice living in groupsshow characteristic signs of stress; enhanced adrenocorticalactivity and thymic involution (McKinney & Pasley, 1973). Whenthe subordinates were moved into a new group cage each day, theiradrenocortical activity was increased further, indicating that19repeated changes in housing conditions can produce HPA activation.Another study showed that for male mice experiencing a change inhousing from individual housing to a group of eight (IG), dominantmales exhibited decreased antibody responses, while subordinatemales exhibited increased antibody responses (Fauman, 1987). Inpaired mice, the magnitude of antibody responsiveness waspositively associated with submissive behavior and negativelyassociated with dominant behavior (Fauman, 1987). In male guineapigs as well, the performance of submissive behavior bysubordinates in the home cage increased serum antibody responses,but aggressive behavior in dominant males was also related to highantibody responses (Stefanski, Hendrichs & Ruppell, 1989).Background to the present studyOur laboratory has developed an animal-tumor model to examinethe effects of psychosocial stressors on mammary tumor growth rateutilizing the transplantable androgen-responsive Shionogi mousemammary carcinoma (SC115) (Emerman & Weinberg, 1989; Weinberg &Emerman, 1989). This tumor arose spontaneously in a female mouseof the DD/S strain. After 19 passages in male mice, an androgen-responsive variant arose that grows more rapidly in males than infemales (Bruchovsky & Rennie, 1978; King & Yates, 1980). Thismouse mammary tumor was selected for study as it is similar tohuman breast cancers and to other endocrine-responsive tumors suchas prostate cancer (Ekman et al., 1979; Geller et al., 1979) inits sensitivity to different classes of steroid hormones,including androgens (King & Yates, 1980), estrogens (Noguchi et20al., 1987) and glucocorticoids (Wanatabe et al., 1982). Usingthis animal-tumor model, we have demonstrated marked effects ofsocial housing condition and exposure to novel environments ontumor growth rate (Emerman & Weinberg, 1989; Weinberg & Emerman,1989). Being reared individually housed and remainingindividually housed (II) or being reared in a sibling group andthen singly housed (GI, n=3 siblings per group) following tumorcell injection significantly increased tumor growth rate comparedto that in mice remaining in their standard sibling rearing groups(GG, n=3 siblings per group) (p < 0.05). In contrast, beingreared individually and then moved to a larger social group (IG,n=5 nonsiblings per group) significantly reduced tumor growth rate(p < 0.05).The present study was designed to examine psychosocialvariables that might modulate the differential tumor growth ratesobserved in our model. These variables include the factors ofgroup size, sibling relationship, change in housing (change or nochange), and direction of change (IG versus GI). Thus, inparallel with our original model, animals were reared eitherindividually (I) or in groups of three or five, consisting ofeither nonsiblings (N3, N5) or siblings (S3, S5). On the day oftumor cell injection animals either remained in their rearingconditions or were rehoused, individual to group (IG) or group toindividual (GI). For mice in all conditions, a detailedexamination was made of home cage behavior. In addition, for allanimals housed in groups, dominance status was noted. Wepostulated that fighting and other social behaviors, as well as21social dominance status, might be related to the changes in tumorgrowth rate observed in our model. Therefore, nonsocial,aggressive, defensive and social behaviors prior to and followingtumor cell injection and rehousing/group formation of animals wereexamined.22MATERIALS AND METHODSAnimal-tumor modelTumor propagation. The androgen-responsive mouse mammarycarcinoma subline designated SC115 Class A (Bruchovsky & Rennie,1978) is maintained by serial transplantation in male mice of theDD/S strain. Tumors weighing approximately 2 g are dissociated in0.05% trypsin and 0.025% EDTA (Sigma Chemical Co., St. Louis, MO)in Ca2+- and Mg2+-free Saline A (pH 7.3) and the cell suspensioncentrifuged at 80 x g for 4 min to enrich the epithelial cellpopulation. Pellets are resuspended in Dulbecco's ModifiedEagle's Medium (DME; Terry Fox Laboratory, Vancouver, BC), andpassed through a 150 micrometer Nitex filter (Tetko, Inc.,Elmsford, NY) to collect single cells and small cell aggregates.Viable cells, determined by trypan blue exclusion, are counted ona hemacytometer. Suspensions of 3 x 106 cells/mouse in 0.1 mlDulbecco's Modified Eagle's Medium (DME) were injectedsubcutaneously (sc) into the interscapular region of male mice 2-4months old. Tumor weights are calculated twice weekly (Emerman &Weinberg, 1989) according to the formula (Simpson-Herren & Lloyd,1970):length (cm) x [width (cm)]2= g.2Experimental animals. Test animals for this study were adult malemice (n=160) of the DD/S strain.^Mice were housed under23conditions of controlled temperature (22° C) and lighting (12 hrlight, 12 hr dark) in a colony room that was protected fromextraneous lab and building noise. Food (Purina Mouse Chow) andwater were available ad libitum. Following weaning at 3 weeks ofage, mice were housed either individually (I) or in groups of 3 or5, consisting of either nonsiblings (N3 and N5) or siblings (S3and S5). At 2-4 months of age, animals in each rearing conditionwere injected sc in the interscapular region with tumor cells (3x 106 cells in 0.1 ml DME). Experimental groups were formedimmediately following injection (see Table 1). Those rearedindividually housed either remained individually housed (II) orwere rehoused in groups of 3 or 5 nonsiblings or siblings (IN3,IS3, IN5 and IS5); those reared in groups either remained in theirgroups (N3N3 and S3S3) or were rehoused individually (N3I, S3I,N5I, and S5I). Due to the large number of groups in theexperimental design, the GG conditions with a group size of 5(N5N5 and S5S5) were not included in the study. Since a limitednumber of animals could be studied at any one time, the study wasconducted in 6 replications; housing conditions were balancedacross replication.Beginning on the day after tumor cell injection, all animalswere exposed to the acute daily stressor (15 min/day, 5 days/week)of being placed in a novel environment, a treatment that we haveshown to enhance the differences in tumor growth rate among groups(Emerman & Weinberg, 1989; Weinberg & Emerman, 1989).^Fivedifferent novel environments were used:^1) a clear plasticcontainer, 9 cm in diameter x 7 cm in height; 2) a polypropylene24TABLE 1HOUSING CONDITIONSTreatmentNo Change^ ChangeGroup Composition^II^GG^GI^IGIndividualNonsibling 3/cageSibling 3/cageNonsibling 5/cageSibling 5/cageN3N3 (9)S3S3 (9)N3I (18)S3I (12)N5I (15)S5I (10)IN3 (24)IS3 (9)IN5 (20)IS5 (10)A total of 4 treatments (II, GG, IG and GI) were used, with 5unique pre-injection housing groups (I, N3, S3, N5, S5) and 11unique post-injection housing groups.25container, 12 x 10 x 4 cm; 3) a cardboard box divided intocompartments, 7 x 7 x 14 cm; 4) a polyethylene container, 6 cm indiameter x 10 cm in height; 5) a standard rodent cage, 18 x 29 x13 cm, empty of bedding, food and water. Mice were exposed to 1of the 5 novel environments each day between 0800 and 1200 hr.Times were randomly varied to increase the unpredictability of thestressors.Mice were palpated twice weekly; once tumors were palpable,caliper measurements were taken twice weekly as described above.Tumors are typically palpable within 6-8 days after injection andthe fastest growing tumors reach a weight of about 3 g inapproximately 17-21 days. The day when tumors were first palpablewas designated as measurement day 1, and the following 3measurement days were designated as days 2, 3, and 4.Behavioral ObservationsThe behavior of all animals was monitored in the home cagefrom about 10 days before tumor cell injection/rehousing to about17-18 days after injection/rehousing. Observations were conducted3 times per week, at 1900 hr during the lights-off period, whenmice are most active. Sessions were begun 5 min after lights-off,because a sudden increase in behavior occurs at that time. Duringthe 18 day post-injection period, in addition to the 1900 hrobservations, animals were also observed immediately after eachexposure to novelty stress. There were 4 pre-injection and 8post-injection observation periods for the 1900 hr observationtime, and 12 post-injection observation periods for the post26novelty stress observations.All mice were tailmarked for identification. During the 1900hr observations, red lights sufficient to illuminate all observedcages were used. The observer sat approximately 0.5-1.0 m fromthe cages during observation sessions.A single observer performed all behavioral observationsThebehaviors chosen (Table 2) were selected following a pilotexperiment, using previous studies of rodent behavior as a guide(Cutler & Piper, 1970; Sorenson, 1987). Behaviors were classifiedinto 5 categories: nonsocial, sleep/rest, aggressive, defensive/receiving aggressive, active social, and passive social. Inaddition, the total social behavior score was calculated (sum ofactive and passive social behavior).Each mouse was observed as a focal animal for four 30-secintervals. The number of times each individual behavior occurredduring each interval was recorded. The intervals were thensummed, giving a total score for each behavior (total time 2 min).Finally, the scores for individual behaviors were summed withineach behavioral category to give a composite score.The total incidence of spontaneous fights and attacks (fordefinitions see Table 2) was also recorded for each cage ofanimals for each session.DominanceThe dominance status (dominant or subordinate) of each animalwas noted. Although wounds resulting from fighting were minimaland limited in extent, it was typically possible to determine an27TABLE 2CATEGORIES AND DEFINITIONS OF BEHAVIORSBehavior^ DefinitionHanging from bars of cage with all pawsoff floorMoving bedding with paws or muzzleDrinking waterEating foodChewing on a piece of beddingJumping off floorBoth forepaws off floorRapid locomotion with all 4 pawsScratching or lickingQuick movement; a small, sudden jumpLocomotion with all 4 pawsLying down, eyes openLying down, eyes closedSudden movement of head and forepaws orbody toward another mouseBiting another mouseRunning after another mouseSeries of lunging, biting, vocalizing,etc. with physical contact between miceLifting both forepaws off ground to faceanother mouseNonsocial behaviorsClimbDig/move beddingDrinkEatGnawJumpRearRunSelf-groomStartleWalkSleep/rest behaviorsRestSleepAggressive behaviorsAttackBiteChaseFightOffensive rearDefensive/receiving aggressive behaviorsDefensive movement^Cringing or backing away from anothermouse, often with eyes closed and/orone forepaw raisedDefensive rear^Lifting both forepaws off ground inresponse to social or aggressivebehavior by another mouseReceive attack lunge^See corresponding definitionsReceive bite^under Aggressive behaviorRetreat^Walking or running away from anothermouseActive social behaviorsAnogenital sniff^Sniffing the anogenital region of anothermouse28Approach^Moving toward another mouseFollow Walking behind and toward another walkingor running mouseGroom Licking another mouse or rubbing mousewith pawsMount^ Clasping another mouse from behindNose Pushing another mouse with muzzleSniff Sniffing another mouseSolicit groom^Pushing head and/or body under anothermouse; second mouse usually groomsfirst mouseTouch^ Contacting another mouse with forepaw(s)Passive social behaviorsReceive anogenital sniffReceive groomReceive mountReceive noseReceive sniffReceive touchSee corresponding definitionsunder Active social behaviorAdapted from Cutler & Piper (1970) and Sorenson (1987).29animal's dominance status by the number of small marks on its tailand rump, with dominant animals having the least number of marks.Dominance status was recorded when the mice were tailmarked, 3times per week. In cases where animals in a cage did not appearto differ in number of wounds, dominance status was determined bythe total number of aggressive and defensive behaviors recordedduring the study. A single mouse consistently emerged as thedominant male in each cage, and no changes in dominance werefound.Data AnalysisAll analyses of variance (ANOVAs) using observation days asa variable were conducted with repeated measures over days.Neuman-Keuls post hoc tests were conducted for all significanteffects found in the ANOVAs. For all analyses, the term treatmentrefers to whether mice were housed individually or in groupsduring pre- and post-injection periods. Thus there were 4treatments: II, GG, IG and GI. The term group is used to denotethe 11 unique post-injection housing conditions: II, IN3, IS3,IN5, IS5, N3N3, S3S3, N3I, S3I, N5I, and S5I.Tumor growth rate. Tumor weights were recorded for animals in 5of the 6 replications. Two ANOVAs were conducted to examine theeffect of treatment on tumor weights. A treatment x days ANOVAwas run to compare animals in the four treatments, GI, GG, II, andIG. In addition, for animals housed in groups post-injection, atreatment x dominance x days ANOVA was conducted. For animals inthe 11 post-injection groups, an overall group x days ANOVA was30performed to analyze tumor growth rate. Finally, an ANOVA wasconducted for animals in the IG treatment (IN3, IS3, IN5, and IS5)to examine the effects of group size (3 or 5), siblingrelationship (nonsibling or sibling), dominance status (dominantor subordinate) and days on tumor growth.Pre-injection behavior. ANOVAs were conducted for the 1900 hrobservations for the 4 pre-injection observation days. Group sizex days ANOVAs were conducted for nonsocial behaviors and sleep tocompare groups of 1, 3 and 5 mice per cage.Aggressive, defensive and social (active, passive and totalsocial) behaviors could only be analyzed for group housed animals(N3, N5, S3, and S5). Thus, for group housed animals, size xsibling x dominance x days ANOVAs were conducted for aggressive,defensive, active social, passive social and total socialcategories of behavior. Nonsocial behaviors and sleep were alsofurther analyzed for group housed animals using size x sibling xdominance x days ANOVAs. In addition, the number offights/attacks per cage of mice was analyzed using size x siblingx days ANOVAs.Post-injection behavior. The greatest amount of behavioralactivity occurred on the injection day and within the first twodays post-injection; in contrast, behavior was relatively low onobservation days 2-7. Therefore all analyses for the 1900 hrobservation period were conducted for the first two observationdays; i.e. observation day 0 (injection day) and observation day1 (24 or 48 hr following injection), and all analyses for thepost-novelty stress observation period were conducted for31observation day 1 data only (the first day on which novelty stresswas imposed).In addition, for both the 1900 and post-novelty stressobservation times, ANOVAs were run including all observation daysover the 18 days post-injection (8 sessions for the 1900 hrobservations, 12 sessions for the post-novelty stressobservations).Nonsocial behavior and sleep were analyzed using group x daysANOVAs. In addition, ANOVAs were conducted to examine the factorsof direction of change in housing condition (GI or IG), groupsize, sibling relationship, dominance and days (change x size xsibling x dominance x days).For mice housed in groups post-injection, 2 sets of ANOVAswere performed. For all group housed animals, aggressive,defensive, and active, passive and total social behaviors wereanalyzed using group x dominance x days ANOVAs. In addition, forgroup housed animals which experienced a change in housingcondition (IN3, IS3, IN5 and IS5), size x sibling x dominance xdays ANOVAs were conducted for aggressive, defensive and socialbehavior in order to examine the effects of these factors.Finally, as with the pre-injection data, the numbers offights/attacks per cage of mice were analyzed using size x siblingx days ANOVAs.32RESULTSTumor growth rateThe overall group x days ANOVA (including all 11 groups)revealed a significant main effect of group (p < 0.001, Fig. 1 A& B). Post-hoc tests indicated that mice in the GI groups (N3I,S3I, and N5I) showed higher tumor growth rates than mice in the IGgroups (IN3, IS3, IN5, and IS5; p's < 0.05), with the singleexception of mice in the S5I group, which showed greater tumorgrowth rates than IN5 mice only (p < 0.01). In addition, S3I micedisplayed the fastest tumor growth rates overall, and weresignificantly faster than II mice (p < 0.05) and N3N3 mice (p <0.05).Tumor growth rates did not differ significantly among any ofthe mice that were group-housed pre-injection (GI or GG); i.e.,for mice group housed pre-injection, there were no effects ofgroup size, sibling relationship or dominance on tumor growth.Tumor growth rates in the GG and II housing groups were generallyintermediate to, but not significantly different from, those shownin the GI and IG groups, with the exceptions of the S3I group,which showed faster tumor growth than II (p < 0.05) and N3N3 (p <0.05) mice, and the S3S3 group, which showed faster tumor growththan the IN5 group (p's < 0.05).The size x sibling x dominance x days ANOVA on animals in theIG treatment revealed a size x sibling interaction (p < 0.01).Averaged across days, IN3 mice had greater tumor weights than micein the other three IG treatments (IS3, IN5 and IS5) (p's < 0.05,33Figure 1. Tumor weights over the four measurement days for micegroup housed post-injection (Graph A) and mice individually housedpost-injection (Graph B) . ( ) = number of subjects per group.Main effect of group, F (10,122) = 10.11, p < 0.001; N3I, S3I, N5I> IN3, IS3, IN5, IS5, p's < 0.05; S3I > II, p = 0.012; S3I > N3N3,p < 0.05; S5I > IN5, p < 0.01; S3S3 > IN5, p < 0.05; N3N3 > IN5,p = 0.104; N3I > II, p = 0.094..--,to....4.504.003.50E—iZ 3.00C.,Wi 2.502.00C:40 1.501.00E-40.500.00A-NI-- IN3 (18)1 2 3 44.50  4.00 -3.50 -3.00 -2.50 -2.00 -1.50 -1.00 -0.50 -0.001^2^3^40^ IS3 (9)-40- INS (20)-0- IS5 (10)- -A- N3N3 (9)- -6- S3S3 (6)/Ar////i/ '// // //// //,///_-,AMEASUREMENT DAYS^MEASUREMENT DAYS35Fig. 2).Interestingly, there was also a treatment x dominanceinteraction for group housed mice (p < 0.001). In the IGtreatment, dominant mice had higher tumor growth rates thansubordinate mice (p = 0.05, Fig. 3A). Conversely, in the GGtreatment, subordinate mice exhibited faster tumor growth ratesthan dominant mice (p < 0.05, Fig. 3B).Pre-injection behaviorNonsocial behavior. The size x days ANOVA comparing groups of 1(II), 3 (N3 and S3), and 5 (N5 and S5) revealed a main effect ofgroup size (p < 0.01). As expected, individually housed miceexhibited more nonsocial behavior than mice in groups of 3 (p <0.05), or groups of 5 (p < 0.01, Fig. 4A).In addition, the size x sibling x dominance x days ANOVA formice housed in groups revealed a main effect of dominance (p <0.001). Dominant animals showed more nonsocial behavior thansubordinate animals (Fig. 4B). The size x sibling interaction wasalso significant (p < 0.001). More nonsocial behavior occurred inthe N3 and S5 groups than in the N5 group; the S3 group displayedan intermediate level of behavior (Fig. 4C).Sleep/rest. The overall size x days ANOVA revealed a main effectof group size (p < 0.001). Mice in groups of 5 slept more thanmice in groups of 3 and individually housed mice (p's < 0.001,Fig. 5A). For group housed mice, a significant main effect ofdominance (p < 0.05) indicated that subordinate mice slept morethan dominant mice (Fig. 5B). A size x sibling interaction (p <36Figure 2. Mean tumor weights averaged across the four measurementdays for mice in the IG treatment. ( ) = number of subjects pergroup.Size x sibling interaction, F(1,149) = 9.88, p < 0.01; IN3 > IS3,IN5, IS5, p's < 0.05.3 738Figure 3. Tumor weights over the four measurement days for micein the IG treatment (Graph A) and mice in the GG treatment (GraphB). D = dominant, S = subordinate, ( ) = number of subjects pergroup.Treatment x dominance interaction, F (1,68) = 14.89, p < 0.001;IG-D > IG-S, p < 0.05; GG-S > GG-D, p < 0.05.1^2^3^4MEASUREMENT DAYS4• IG-D (15)0^ IG-S (42)1^2^3^4MEASUREMENT DAYS40Figure 4. Mean frequency of nonsocial behavior averaged acrossthe four pre-injection observations for all mice (Graph A) and formice group housed pre-injection (Graphs B & C). (SZ=1) = 1 mouseper cage, (SZ=3) = 3 mice per cage, (SZ=5) = 5 mice per cage, DOM= dominant, SUB = subordinate, ( ) = number of subjects per group.Graph A: Main effect of group size, F (2,145) = 5.06, p < 0.01;SZ=1 > SZ=5, p < 0.01; SZ=1 > SZ=3, p = 0.087.Graph B: Main effect of dominance, F (1,63) = 17.24, p < 0.001;DOM > SUB.Graph C: Size x sibling interaction, F (1,63) = 13.86, p < 0.001;N3 > N5, p = 0.002; S5 > N5, p < 0.05.252044150=I 10I^ I DOM (21)20101525525201510SZ= 1 (77)SZ=3 (47)I SZ=5 (24)[FM SUB (50)M N3 (26).111. S3 (21)N5 (14)S5 (10)420.05) indicated that N5 animals slept more than those in all othergroups (p's < 0.005, Fig. 5C).Aggressive and defensive behavior. The size x sibling xdominance x days ANOVAs revealed main effects of dominance forboth aggressive (p < 0.001) and defensive (p < 0.05) behavior. Asexpected, dominant mice displayed more aggressive behavior thansubordinate mice (Fig. 6A) and subordinate mice displayed moredefensive behavior than dominant mice (Fig. 6B).Social behavior. Main effects of dominance were also found foractive (p < 0.001) and passive social behavior (p = 0.053).Parallelling the differences in aggressive and defensive behav-ior, dominant animals displayed more active social behavior thansubordinate animals (Fig. 6C) and subordinate animals showed morepassive social behavior than dominant animals (Fig. 6D).The size x sibling x dominance x days ANOVAs revealed maineffects of group size for passive (p < 0.01) and total (p < 0.05)social behavior, and a trend for group size for active socialbehavior (p = 0.085). Animals housed in groups of 3 showed moreactive, passive and total social behavior than those housed ingroups of 5 (Fig. 7 A, B, C).Fights/attacks. The analysis revealed a main effect of group size(p < 0.001), as well as a size x days interaction (p < 0.05, Fig.8). Post-hoc tests indicated that groups of 3 consistently showeda low level of fighting on all 4 observation days, but groups of5 showed a high level of fighting on day 1 and a decrease infighting over days. However, groups of 5 had more fights thangroups of 3 on each observation day.43Figure 5. Mean frequency of sleep averaged across the four pre-injection observations for for all mice (Graph A) and for micegroup housed pre-injection (Graphs B & C). (SZ=1) = 1 mouse percage, (SZ=3) = 3 mice per cage, (SZ=5) = 5 mice per cage, DON =dominant, SUB = subordinate, ( ) = number of subjects per group.Graph A: Main effect of group size, F (2,145) = 12.65, p < 0.001;SZ=5 > SZ=1, SZ = 3, p's < 0.001.Graph B: Main effect of dominance, F (1,63) = 5.23, p < 0.05; SUB> DOM.Graph C: Size x sibling interaction, F(1,63) = 5.17, p < 0.05; N5> N3, S3, S5, p's < 1.000.500.00SZ■■ 1 (77)SZ=3 (47)SZ=5 (24)AI^ IDOM (21)M SUB (50)I^I2222 N3 (211)ME 83 (21)NS (14)SS (10)45Figure 6. Mean frequency of aggressive (Graph A), defensive(Graph B), active social (Graph C) and passive social (Graph D)behaviors averaged across the four pre-injection observations formice group housed pre-injection. DOM = dominant, SUB =subordinate. ( ) = number of subjects per group.Graph A: Main effect of dominance for aggressive behavior, F(1,63) = 38.36, p < 0.001; DOM > SUB.Graph B: Main effect of dominance for defensive behavior, F(1,63) = 35.71, p < 0.001; SUB > DOM.Graph C: Main effect of dominance for active social behavior, F(1,63) = 19.05, p < 0.001; DON > SUB.Graph D: Main effect of dominance for passive social behavior, F(1,63) = 3.90, p = 0.053; SUB > DOM.4 647Figure 7. Mean frequency of active, passive and total socialbehavior averaged across the four pre-injection observations formice group housed pre-injection. (SZ=3) = 3 mice per cage, (SZ=5)= 5 mice per cage, ( ) = number of subjects per group.Graph A: Main effect of group size for active social, F (1,63) =3.07, p = 0.085; SZ=3 > SZ=5.Graph B: Main effect of group size for passive social, F (1,63)= 8.27, p < 0.01; SZ=3 > SZ=5.Graph C: Main effect of group size for total social, F (1,63) =5.89, p < 0.05; SZ=3 > SZ=5.202015 1510 105 5o oTOTAL SOCIALACTIVE SOCIAL20oPASSIVE SOCIAL49Figure 8. Mean frequency of fights and attacks over the four pre-injection observations for mice group housed pre-injection. (SZ=3)= mice per cage, (SZ=5) = 5 mice per cage, ( ) = number ofsubjects per group.Main effect of size, F (1,17) = 27.803, p < 0.001;Size x days interaction, F (3,51)= 2.85, p < 0.05; days 1, 2, 3 &4: SZ=5 > SZ=3, p's < 0.05.o1 2 3 420OBSERVATION DAYS51Post-injection behavior: 1900 hr observations, days 0 and 1Nonsocial behavior. The overall group x dominance x days ANOVA(including all 11 groups) revealed a main effect of group (p <0.001). Generally, animals in the GI groups (N3I, N5I and S5I)showed the most nonsocial behavior, and animals in the IG (IN3,IS3, IN5, and IS5) group showed the least nonsocial behavior (Fig.9 A & B). Animals in the S3S3 group also showed low nonsocialbehavior (S3S3 < N3I, N5I, and S5I, p's < 0.05). In addition, amain effect of days (p < 0.001) indicated that more nonsocialbehavior occurred on observation day 0 than on day 1.The change x size x sibling x dominance x days ANOVA foranimals experiencing a change in housing condition (GI and IGtreatments) revealed a change x sibling x dominance interaction (p< 0.05). For mice from GI sibling groups (S3I and S5I), dominantmice showed the most nonsocial behavior, whereas for mice from GInonsibling groups (N3I and N5I) subordinate mice showed the mostnonsocial behavior (p's < 0.01) (Fig 9C).A change x sibling x days interaction was also noted fornonsocial behavior (p < 0.01). Animals from GI sibling groups(S3I and S5I) showed a marked decline in nonsocial behavior,falling to the level of the IG groups (IN3, IS3, IN5, and IS5) byday 1.Sleep/rest. There was a main effect of group (p < 0.001) forsleep. Overall, mice in the IN5 group slept more than mice in theN3I, N5I, and S5I groups (p's < 0.05).The change x size x sibling x dominance x days ANOVA revealeda change x sibling x days interaction (p = 0.002). On day 1 the52Figure 9. Mean frequency of nonsocial behavior averaged acrossthe first two post-injection observations at 1900 hr for micegroup housed post injection (Graph A), mice individually housedpost-injection (Graph B), and mice which experienced a change inhousing condition (Graph C). DOM = dominant, SUB = subordinate,NS = nonsibling, SIB = sibling, ( ) = number of subjects pergroup.Graphs A & B: Main effect of group, F (10,137) = 8.65, p < 0.001;N3I, N5I, S5I > IN3, IS3, IN5, IS5, S3S3, p's < 0.05; S3I > IN3,INS, p's < 0.05.Graph C: Change x sibling x dominance interaction, F (1,100) =6.10, p < 0.05; GI-SUB-NS, GI-DOM-SIB > IG-DOM-NS, IG-SUB-NS, IG-DOM-SIB, IG-SUB-SIB, p's < 0.01; GI-SUB-SIB > IG-SUB-NS, IG-SUB-SIB, p's < 0.05; GI-DOM-NS > IG-SUB-NS, p = 0.005.DOM-NS(9, 12)SUB-NS(22. 32)DOM-8(6, 5)81311-8(16, 14)GI IGGROUP INDIVIDUAL54IG nonsibling conditions exhibited large increases in amount ofsleep while the GI conditions (both nonsibling and sibling)remained stable (Fig. 10).Aggressive and defensive behavior. The group x dominance x daysANOVAs indicated a main effect of group for defensive behavior (p< 0.001) and main effects of dominance for both aggressive (p <0.001) and defensive (p < 0.05) behaviors. Mice in the IG groups(IN3, IN5, and IS5) showed more defensive behavior than those inthe GG groups (N3N3 and S3S3; Fig. 11A). Further, as noted forthe pre-injection analysis, dominant mice performed moreaggressive behavior than subordinate mice, and subordinate miceperformed more defensive behavior than dominant mice. Inaddition, there was a main effect of days for defensive behavior(p < 0.001), and a similar trend for aggressive behavior (p =0.104). More aggressive and defensive behavior occurred on day 0than on day 1.For aggressive behavior, a dominance x sibling x daysinteraction (p < 0.05) was also present. On day 0, moreaggressive behavior was shown by dominant mice housed withnonsiblings (IN3 and IN5) than by dominant mice housed withsiblings (IS3 and I55) or any subordinate mice (p's < 0.011; Fig.11B). Dominant mice housed with nonsiblings were the majorcontributor to the main effect of days, exhibiting a decrease inaggressive behavior from day 0 to day 1 of observation. Allgroups of animals were similar by day 1.Social behavior. The group x dominance x days ANOVAs revealedmain effects of group for passive (p < 0.001) and total (p < 0.05)55Figure 10. Mean frequency of sleep averaged across the first twopost-injection observations at 1900 hr for mice which experienceda change in housing conditions. ( ) = number of subjects pergroup.Change x sibling x days interaction, F (1,100) = 9.68, p =0.002;day 1: GI-NS > GI-SIB, IG-NS, IG-SIB, p's < 0.01.o^1OBSERVATION DAYS57Figure 11. Mean frequency of defensive behavior averaged acrossthe first two post-injection observations at 1900 hr for micegroup housed post-injection (Graph A) , and frequency of aggressivebehavior over the first two post-injection observations at 1900hre for mice in the IG treatment. DOM = dominant, SUB =subordinate, NS = nonsiblings, SIB = siblings, ( ) = number ofsubjects per group.Graph A: Main effect of group for defensive behavior, F (5,69) =4.22, p = 0.002; IN3, IN5, IS5 > N3N3, S3S3, p's s 0.054.Graph B: Dominance x sibling x days interaction for aggressivebehavior, F (1,55) = 4.24, p < 0.05; day 0: DOM-NS > SUB-SIB,SUB-NS, p's s 0.011.o 1OBSERVATION DAYS-II- DOM-NS (12)^0^ DOM-SIB (5)--e^ SUB-NS (32)--6^ SUB-SIB (14)432I0__59social behavior. The IS5 group showed more passive and totalsocial behavior than all other groups (p's < 0.05) (Figs. 12A &12B). The S3S3 group generally showed the lowest level of passivesocial behavior (S3S3 < IN5, IS5 and N3N3, p's < 0.05; S33 < IN3,p = 0.082). Main effects of days were present for active (p <0.05), passive (p < 0.01) and total (p < 0.01) social behavior.Animals displayed more active, passive, and total social behavioron observation day 0 than day 1.In the size x sibling x dominance x days ANOVAs for totalsocial behavior, there was a sibling x dominance x daysinteraction (p < 0.05). On observation day 0, all groups weresimilar and showed a high level of total social behavior;subordinate mice in sibling groups maintained a high level ofsocial behavior on observation day 1 while all other groups showeda marked decrease in social behavior (p's < 0.05).Fights/attacks. In the group x days ANOVA, there was a maineffect of group (p < 0.001, Fig. 13), as well as a group x daysinteraction (p < 0.01). Significantly more fighting occurred inthe IN5 and IS5 groups than in all other groups (p's < 0.05). Inaddition, the pattern over days differed among groups. The N3N3and S3S3 groups maintained the lowest level of fighting; theyshowed no change over days. All other groups showed a decrease infighting from observation days 0 to 1.Post-injection behavior: Post-novelty stress observations, day 1Nonsocial behavior. The ANOVA comparing all 11 groups revealed asignificant main effect of group for nonsocial behavior (p <60Figure 12. Mean frequency of passive (Graph A) and total (GraphB) social behavior averaged across the first two post-injectionobservations at 1900 hr for mice group housed post-injection.( ) = number of subjects per group.Graph A: Main effect of group for passive social, F (5,69) =8.02; p < 0.001; IS5 > all others, p's < 0.01; IN5, IS5, N3N3 >S3S3, p's < 0.05; IN3 > S3S3, p = 0.082.Graph B: Main effect of group for total social, F (5,69) = 2.81;p < 0.05; IS5 > all others, p's < 0.05.30 70605040302010IN3 (24)ctun IS3 (9)INS (20)135 (10)N3N3 (9)    S383 (9):C•XI^IPASSIVE SOCIAL^TOTAL SOCIAL62Figure 13. Mean frequency of fights and attacks averaged acrossthe first two post-injection observations at 1900 hr for micegroup housed post-injection. ( ) = number of subjects per group.Main effect of group, F (5,17) = 10.57, p < 0.001; IN5, IS5 > IN3,IS3, N3N3, S3S3, p's < 0.05.6 3640.001) (Fig. 14). Overall, the GI groups showed more nonsocialbehavior than the IG groups (N3I, S3I and N5I > IN3, IN5 and IS3,p's < 0.05). Within the group-housed conditions, S3S3 was thehighest (S3S3 > IN3, IS3 and IN5, p's < 0.05), whereas within theindividually housed conditions, the S5I group was the lowest (S5I< N3I and N5I, p's 0.01).Sleep/rest. The group ANOVA revealed a significant effect ofgroup (p < 0.001, Fig. 15). Generally, mice in the IG groups(IN3, IS3, and IN5) slept the most, and mice in the GI nonsiblinggroups (N3I and N5I) and the S3S3 group slept the least (p's <0.05). In addition, the S3I and II groups tended to show lowsleep, but only the comparison with tht IN5 group reachedsignificance (p < 0.01). Within group-housed conditions, S3S3mice slept least (S3S3 < IN3, IS3 and IN5, p's < 0.05), and withinindividually housed conditions, S5I mice generally slept most (S5I> N5I, p < 0.05, S5I > N3I, p = 0.104)Aggressive and defensive behavior. In the group x dominanceANOVA, there was a significant main effect of dominance foraggressive behavior [F(1,69) = 4.57, p < 0.05]. Dominant miceshowed more aggressive behavior than subordinate mice [dominant:0.174 ± 0.102; subordinate: 0.000 ± 0.000 (means ± SEM)].Interestingly, the subordinate mice displayed no aggressionwhatsoever during the first post-novelty stress observation. Inaddition, there was a significant main effect of group fordefensive behavior (p = 0.012). Post-hoc tests indicated that IS5mice showed the most defensive behavior (p's < 0.05, Fig. 16).Social Behavior. There were main effects of group for active (p65Figure 14. Frequency of nonsocial behavior during the first post-novelty stress observation for mice group housed post-injection(Graph A) and mice individually housed post-injection (Graph B).( ) = number of subjects per group.Main effect of group, F (10,137) = 8.29, p < 0.001;N3I, S3I, N5I > IN3, IS3, IN5, p's < 0.05;N5I > S5I, IS5, N3N3, p's < 0.01;N3I > S5I, p = 0.012;N3I > IS5, p = 0.065;N3I > N3N3, p = 0.079;S3S3 > IN3, IN5, p's < 0.05;S3S3 > IS3, p =0.057.2520150 10I^10110 IS3 (9)IN3 (24) 135 (10)N3N3 (9)■:•M INS (20) M 8333 (9)141,:b.*"""473  2520151050S5I (10)N31 (18);OM N5I (14)MILO S31 (12)^H (14)GROUP^ INDIVIDUAL67Figure 15. Frequency of sleep during the first post-noveltystress observation for mice group housed post-injection (Graph A)and mice individually housed post-injection (Graph B). ( ) =number of subjects per group.Main effect of group, F (10,137) = 6.98, p <0.001;IN3, IS3, IN5, S5I > N5I, S3S3, p's < 0.05;IN3, IS3, IN5 > N3I, p's < 0.01;IN5 > S3I, p = 0.01;IN5 > II, p = 0.094;S5I > N3I, p = 0.104.GROUP^ INDIVIDUAL69Figure 16. Frequency of aggressive (Graph A) and defensive (GraphB) behavior during the first post-novelty stress observation formice group housed post-injection. ( ) = number of subjects pergroup.Main effect of group for defensive behavior, F (3,59) = 4.00, p =0.012; IS5 > IN3, IS3, N3N3, S3S3, p's < 0.05; IS5 > IN5, p =0.072.7 071< 0.05), passive (p < 0.01) and total (p < 0.01) social behavior.Consistent with the 1900 hr observations, the IS5 group showed themost social behavior on all 3 measures, although for passivesocial behavior, only the comparisons with the IN3 (p = 0.001) andIS3 (p = 0.058) groups reached significance, and for active andtotal social behaviors, only the comparison with the IN3 groupreached significance (p's < 0.05). In addition, there were maineffects of group size for passive (p = 0.003), active (p = 0.002),and total (p = 0.002) social behaviors. Groups of 5 exhibitedmore social behavior than groups of 3. Finally, there were maineffects of sibling relationship for all measures of socialbehavior (p's < 0.05), with siblings consistently higher thannonsiblings.Fights/attacks.^There were no significant differences amonggroups in the first post-novelty stress observation.Post-Injection Behavior: 1900 hr Observations, Days 0-7Nonsocial behavior. The group x days ANOVA for all 11 groupsrevealed a main effect of group (p < 0.001, Fig. 17).Individually housed animals generally exhibited more nonsocialbehavior than the IN3 and IN5 groups (p's < 0.05). The IS3 groupalso showed low nonsocial behavior (IS3 < N3I, S3I and S5I, p's <0.05), and the GG groups (N3N3 and S3S3) tended to show lownonsocial behavior, but only the comparison with the N3I groupreached significance.The change x size x sibling x dominance x days ANOVA for micewhich experienced a change in housing conditions revealed a change72Figure 17. Frequency of nonsocial behavior over the eight post-injection observations at 1900 hr for mice group housed post-injection (Graph A) and mice individually housed post-injection(Graph B). ( ) = number of subjects per group.Main effect of group, F (10,137) = 10.69, p < 0.001;N3I, S3I, N5I, S5I, II > IN3, IN5, p's < 0.05;N3I, S5I > IS3, IS5, p's < 0.05;N3I > N3N3, S3S3, p's < 0.05;S3I > IS3, p < 0.05.N5I (14)g.444INDIVIDUALGROUP0 2520150g 1 0135 (10)N3N3 (9)S3S3 (9)302520151050B 10:11 S3I (12)^II (14), N3I (18) I^I 351 (10)74x dominance interaction (p < 0.001). Averaged across observationdays, IG subordinate mice displayed the least nonsocial behaviorcompared to all other groups (p's < 0.001), and GI subordinatemice showed more nonsocial behavior than IG dominant mice (p <0.01) (Fig. 18A). Finally, there was a change x dominance x daysinteraction (p < 0.01, Fig. 18B). On the first three observationdays, IG mice were consistently lower than GI mice in nonsocialbehavior. Over the eight observations, the GI and IG subordinatemice maintained low and high levels of nonsocial behavior,respectively. Interestingly, IG dominant mice increased from lowto high nonsocial behavior, whereas in contrast, GI dominant micedecreased from high to moderate nonsocial behavior. Thus, whilethe GI and IG mice were initially different in nonsocial behavior,the dominant mice from these treatments were no longer differentfrom each other by day 4.Sleep/rest. A main effect of group was found in the overall groupx days ANOVA (p < 0.001, Fig. 19). The IN5 group slept the moston average, significantly more than all other groups except IS3and N3N3 (p's < 0.05).The change x size x sibling x dominance x days ANOVA revealeda change x dominance interaction (p < 0.01). IG subordinate miceslept significantly more than any other subgroup (p's < 0.001,Fig. 20A). There was a change x size interaction as well (p <0.05, Fig. 20B). On average, IG mice slept more than GI mice,regardless of group size (p's < 0.05). In addition, IG mice ingroups of 5 slept more than those in groups of 3 (p < 0.01), butthis effect was due to high sleep in the IN5 group.75Figure 18. Frequency of nonsocial behavior over the eight post-injection observations at 1900 hr for mice experiencing a changein housing conditions. GI-D = GI dominant mice, GI-S = GIsubordinate mice, IG-D = IG dominant mice, IG-S = IG subordinatemice; ( ) = number of subjects per group.Graph A: Change x dominance interaction, F (1,100) = 18.25, p <0.001; GI-D, GI-S, IG-D > IG-S, p's < 0.001; GI-S > IG-D, p <0.01.Graph B: Change x dominance x days interaction, F (7,700) = 2.91,p < 0.01;days 0-3: GI-D, GI-S > IG-D, IG-S, p's < 0.05;day 3: GI-S > IG-D, p = 0.052;day 4: GI-D, GI-S, IG-D > IG-S, p's^0.01;day 5: GI-S > GI-D, p = 0.084;days 5, 6, 7: GI-S, IG-D > IG-S, p's < 0.05.OBSERVATION DAYS2520r1.1 150110 100^1^2^3^4^5^6^7I^IA GI-D (15)GI-S (38):CO:4 IG-D (17)^3040IG-S (46)2010077Figure 19. Mean frequency of sleep averaged across the eightpost-injection observations at 1900 hr for mice group housed post-injection (Graph A) and mice individually housed post-injection(Graph B). ( ) = number of subjects per group.Main effect of group, F (10,137) = 7.56, p < 0.001; IN5 > IN3,IS5, S3S3, N3I, N5I, S3I, S5I, II, p's < 0.05.n•:•N N5I (14)3.002.502.001.501.000.500.00N3I (18) I^I S5I (10)tniu S3I (12)^II (14)3.00ri2.502.0044^1.500c4^1.000.500.00GROUP^ INDIVIDUAL79Figure 20. Mean frequency of sleep averaged across the eightpost-injection observations at 1900 hr for mice experiencing achange in housing conditions. GI-D = GI dominant mice, GI-S = GIsubordinate mice, IG-D = IG dominant mice, IG-S = IG subordinatemice; GI-3 = GI mice from groups of 3, GI-5 = GI mice from groupsof 5, IG-3 = IG mice from groups of 3, IG-5 = IG mice from groupsof 5; ( ) = number of subjects per group.Graph A: Change x dominance interaction, F (1,100) = 9.23, p <0.01; IG-S > GI-D, GI-S, IG-D, p's < 0.001.Graph B: Change x size interaction, F (1,100) = 4.62, p < 0.05;IG-3, IG-5 > GI-3, GI-5, p's < 0.05; IG-5 > IG-3, p < 0.01.I^I GI-S (3 8)^IG-S (46)2.000.500.001.501.002.000.00GI-D (15)^IG-D (17)81Aggressive and defensive behavior. The group x dominance x daysANOVAs revealed main effects of dominance for both aggressive anddefensive behavior (p's < 0.001). Dominant mice exhibitedsignificantly more aggressive and less defensive behavior thansubordinate mice. A main effect of group was present fordefensive behavior (p < 0.001). The results were similar to thosefound in the first two observations; all IG groups (IN3, IS3, IN5and IS5) showed significantly more defensive behavior than both GGgroups (N3N3 and S3S3) (p's < 0.05). In addition, there were maineffects of days for aggressive (p < 0.05) and defensive (p <0.001) behavior. Aggressive and defensive behavior decreased fromobservation day 0 to day 1, and both behaviors remained low fromobservation days 1-7 (Fig. 21).Group x dominance interactions were found for aggressive (p= 0.004) and defensive behavior (p = 0.005). Generally, foraggressive behavior, significant differences between dominant andsubordinate males were present only within the IG groups (IN3, IN5and IS5, p's < 0.05). Differences in defensive behavior weresignificant in the IN3 (p < 0.001) and IS3 (p = 0.051) groups.The size x sibling x dominance x days ANOVA revealed a sizex dominance x days interaction for aggressive behavior (p < 0.05,Fig. 22). Dominant mice in groups of 3 consistently demonstratedhigh aggressive behavior, on all days except day 1 (DOM-3 > SUB-3,SUB-5, p's < 0.05), while dominant mice in groups of 5 showed highaggressive behavior only on observation days 1, 5 and 7 (DOM-5 >SUB-3, SUB-5, p's < 0.05).Finally, a sibling x dominance x days effect was found for82Figure 21. Frequency of aggressive and defensive behaviors overthe eight post-injection observations at 1900 hr for mice grouphoused post-injection. N = 81 subjects.Main effect of days for aggressive behavior, F (7,483) = 2.08, p< 0.05.Main effect of days for defensive behavior, F (7,483) = 6.51, p <0.001.5r:v4'2x 344tal20f:4^144(310Z 1^2^3^4^5^6^7OBSERVATION DAYS84Figure 22. Frequency of aggressive behavior over the eight post-injection observations at 1900 hr for mice in the IG treatment.DOM-3 = dominant mice from groups of 3, SUB-3 = subordinate micefrom groups of 3, DOM-5 = dominant mice from groups of 5, SUB-5 =subordinate mice from groups of 5, ( ) = number of subjects pergroup.Size x dominance x days interaction, F (7,385) = 2.34, p < 0.05;days 0, 2, 4, 5, 6, 7: DOM-3 > SUB-3, SUB-5, p's < 0.05;days 1 & 4: DOM-5 > SUB-5, p < 0.05; DOM-5 > SUB-3, p = 0.09;day 3: DOM-3 > SUB-3, SUB-5, DOM-5, p's^0.003;days 5 & 7: DON-5 > SUB-3, SUB-5, p's^0.013.6r:414PCI 30 244 10 0^1^2^3^4^5^6^7OBSERVATION DAYS86aggressive behavior (p < 0.01). On observation days 4, 5 and 6,dominant mice in nonsiblOing groups showed high aggressivebehavior (DOM-NS > SUB-NS, SUB-SIB, p's < 0.05); however, on day7, dominant mice in sibling groups were highest (DOM-SIB > SUB-NS,SUB-SIB, p's < 0.001).Social Behavior. The group x dominance x days ANOVA revealed maineffects of dominance for passive (p < 0.05) and active socialbehavior (p = 0.01). Dominant mice showed more active socialbehavior and less passive social behavior than subordinate mice(Fig. 23).There were main effects of group for active (p < 0.05),passive (p < 0.001), and total social behavior (p = 0.002). The185 group was highest on all 3 measures (p's s 0.058).A main effect of days occurred for all 3 measures of socialbehavior (p's < 0.001, Fig. 24). Generally, social behavior washigh on observation day 0, and markedly reduced on observation day1. Following observation day 1, active social behavior remainedstable, and passive and total social behavior gradually declinedto day 3. All three types of social behavior showed slightelevations on day 4, at approximately the first day of the secondfive-day stress period.Fights/Attacks. The group x days ANOVA revealed a main effect ofgroup (p < 0.001). The IG groups (IN3, IN5, and IS5) generallyexhibited significantly more fighting than the GG groups (N3N3 andS3S3, p's < 0.05), and were also higher than the IS3 group (p's <0.05). A main effect of days (p < 0.001) indicated that a markeddecrease in fighting occurred from observation day 0 to day 1,87Figure 23. Mean frequency of active (Graph A) and passive (GraphB) social behaviors averaged across the eight post-injectionobservations at 1900 hr for mice group housed post-injection. DOM= dominant, SUB = subordinate, ( ) = number of subjects per group.Graph A: Main effect of dominance for active social behavior, F(1,69) = 6.97, p = 0.01; DOM > SUB.Graph B: Main effect of dominance for passive social behavior, F(1,69) = 4.10, p < 0.05; SUB > DOM.89Figure 24.^Frequency of active, passive and total socialbehaviors over the eight post-injection observations at 1900 hrfor mice group housed post-injection. N = 81 subjects.Main effect of days for active social behavior, Fp < 0.001. (7,483) = 5.95,Main effect of days for passive social behavior, F (7,483) = 12.62p < 0.001.Main effect of days for total social behavior, F (7,483) = 10.54;p < 0.001.3000^1^2^3^4^5^6^7OBSERVATION DAYS91followed by a stabilization of the fighting level.The size x sibling x days ANOVA revealed a size x daysinteraction (p < 0.001, Fig. 25). On day 0, groups of 5 exhibiteda large amount of fighting, significantly more than groups of 3 (p= 0.001). On day 1, groups of 3 and 5 both showed reducedfighting; groups of 5 showed a greater reduction than groups of 3,but remained significantly higher (p < 0.012). On days 2-6, thetwo group sizes did not differ, but on day 7, more fightingoccurred in groups of 5 than groups of 3 (p = 0.001).Post-injection behavior: Post-novelty stress observations, days1-12Nonsocial behavior. The group x days ANOVA revealed a main effectof group (p < 0.001). The N3I group displayed the most nonsocialbehavior, significantly more than all groups except S3S3 (Fig.26). This result differs from the results of the firstobservation alone.The change x size x sibling x dominance x days ANOVA for micewhich experienced a change in housing revealed a change x daysinteraction (p < 0.001, Fig. 27). In general, GI mice initiallyshowed high nonsocial behavior but decreased over days, and ICmice initially showed low nonsocial behavior but increased overdays. On observation days 8 and 9, IC mice were higher than GImice, but both treatments were equal by day 10.Sleep/rest. A main effect of group was found for sleep (p <0.001, Fig. 28). The IN3, N3I, and S3S3 groups slept the least,although for the IN3 group only the comparison with IN5 reached92Figure 25. Frequency of fights and attacks over the eight post-injection observations at 1900 hr for mice in the IG treatment.(SZ=5) = 5 mice per cage, (SZ=3) = 3 mice per cage, ( ) = numberof subjects per group.Main effect of size, F (1,13) = 20.411, p = 0.001, SZ=5 > SZ=3.Size x days interaction, F (7,91) = 5.459, p <0.001; days 0 and 7,SZ=5 > SZ=3, p's = 0.001; day 1, SZ=5 > SZ=3, p = 0.012.94Figure 26. Mean frequency of nonsocial behaviors averaged acrossthe 12 post-novelty stress observations for mice group housedpost-injection (Graph A) and mice individually housed post-injection (Graph B). ( ) = number of subjects per group.Main effect of group, F (10,137) = 4.763, p < 0.001.N3I > IN3, IS3, IN5, IS5, N3N3, S3I, N5I, S5I, II, p's < 0.05;S3S3 > IN5, p = 0.096.INDIVIDUALGROUP1 5 151057 IN3 (24) I^I IS5 (10)N3N3 (9)[IIIE IS3 (9)■X•Vi IN5 (20) 1^ 1 S3S3 (9) +NA N5I (14)N3I (18) I^I S5I (10)= S3I (12)^H (14)'96Figure 27. Frequency of nonsocial behaviors across the 12 post-novelty stress observations for mice experiencing a change inhousing conditions. ( ) = number of subjects per group.Change x days interaction, F (11,1100) = 9.47, p < 0.001; days 1-4, GI > IG, p's < 0.05, day 5, GI > IG, p = 0.089; days 8 & 9, IG> GI, p's < 0.05.15r:v4)x 1 0a)440r:4^5tx101 2 3 4 5 6 7 8 9 10 11 12OBSERVATION DAYS98Figure 28. Mean frequency of sleep averaged across the 12 post-novelty stress observations for mice group housed post-injection(Graph A) and mice individually housed post-injection (Graph B).( ) = number of mice per group.Main effect of group, F (10,137) = 4.90, p < 0.001;IN3 < IN5, p = 0.004;N3I, S3S3 < IN5, S5I, N3N3, p's < 0.05;N3I < S3I, p < 0.05;N3I < IS3, p = 0.092;S3S3 < IS3, p < 0.05;S3S3 < S3I, p = 0.076.IS5 (10)N3N3 (9)S3S3 (9)GROUP INDIVIDUAL100significance (IN3 < IN5, p = 0.004; N3I, S3S3 < IN5, S5I, N3N3,p's < 0.05; N3I < S3I, p < 0.05; N3I < IS3, p = 0.092; S3S3 < IS3,p < 0.05; S3S3 < S3I, p = 0.076).Aggressive and defensive behavior. In the group x dominance xdays ANOVA for group housed mice, main effects of dominance wererevealed for aggressive (p < 0.001) and defensive (p < 0.001)behaviors. Dominant mice showed more aggressive and lessdefensive behavior than subordinate mice. An interaction ofdominance x days occurred for aggressive behavior (p < 0.005).Subordinate mice showed consistently low aggressive behavior overdays; in contrast, dominant mice showed low aggressive behavior onobservation days 1-4 and high aggressive behavior on days 5-12.A group x dominance effect occurred for defensive behavior (p =0.01). The IN3 subordinate subgroup was highest in defensivebehavior, showing more than all subgroups except I55 dominant andIS5 subordinate mice (p's < 0.05, Fig. 29).The size x sibling x dominance x days ANOVA revealed a sizex days interaction for defensive behavior (p = 0.053), but therewas no main effect of size. From observation days 1-4, groups of5 showed marked variability. High defensive behavior levels ondays 1 and 3 alternated with low levels on days 2 and 4. Incontrast, groups of 3 showed a steady increase in defensivebehavior from a low level on day 1 to a high level on day 4.There were no consistent effects after day 4.A day x size x dominance interaction occurred for aggressivebehavior (p = 0.002). Dominant mice in groups of 3 were usuallyhighest, and showed more aggressive behavior than subordinate mice101Figure 29. Mean frequency of defensive behaviors averaged acrossthe 12 post-novelty stress observations for mice group housedpost-injection. ( ) = number of subjects per group.Group x dominance interaction, F (5,69) = 3.29, p = 0.01; 1N3-S >all others except 1S5-D, 1S5-S, p's < 0.05; 1N5-S > 1N3-D, p =0.097, 1S5-S > 1N3-D, p < 0.05; 1S5-S > S3S3-D, p = 0.101.N3N3 (3,6)IS3 (3,6)IN3 (8,1 6) I^I 155 (2)INS (4.16) I^ I S3S3 (9)1023.000 2.502.0044^1.500P4 1.000.500.00DOM^SUBDEFENSIVE103in groups of 3 or 5 on observation days 1, 6, 7, 9 and 12 (p's s0.059). When dominant mice in groups of 5 were compared tosubordinate mice in groups of 3 or 5, only the comparisons fordays 8 and 11 reached significance (p's < 0.05).Social Behavior. A group x dominance x days interaction was foundfor active social behavior (p < 0.01). IS5 dominant mice showedvery high active social behavior on observation days 5 and 8,significantly more than the subordinate mice in all othergroups(p's < 0.05). There were main effects of days for active (p= 0.004), passive (p < 0.001), and total (p < 0.001) socialbehavior. All three social behaviors followed a pattern of sharpelevations and reductions from observation days 1 to 4, remainedstable at a moderate level for seven observation days, and finallyfollowed another series of small elevations and reductions fromdays 10-12 (Fig. 30).The size x sibling x dominance x days ANOVAs for mice in theIG treatment revealed main effects of group size for active (p =0.067), passive (p < 0.002) and total (p < 0.01) social behavior.On average, mice in groups of 5 showed more active, passive andtotal social behaviors than mice in groups of 3. Size x daysinteractions (p's < 0.001) and sibling x days interactions (p's s0.001) were found for all three kinds of social behavior (p's s0.001). Active, passive and total social behaviors showed similarvariations by size and sibling over days. Regardless of groupsize or sibling relationship, all mice followed the overallincreases and decreases in social behavior over the first fourobservation days (see Figure 30). The major reason for the size104Figure 30.^Frequency of active, passive and total socialbehaviors over the 12 post-novelty stress observations for micegroup housed post-injection. N = 81 subjects.Main effect of days for active social behavior, F (11,759) = 2.50,p = 0.004.Main effect of days for passive social behavior, F (11,759) =5.11, p < 0.001.Main effect of days for total social behavior, F (11,759) = 4.34,p < 0.001.10PI 92 87Z14^6Fil44^5o 4g14^3C42Z 101 2 3 4 5 6 7 8 9 10 11 12OBSERVATION DAYS106x days interaction was higher variability over days in groups of5 than groups of 3. Generally, over the first three observationdays, groups of 5 were higher than groups of 3 (p's < 0.05), andover the first four observation days (except day 3), siblings werehigher than nonsiblings (p's < 0.05), but thereafter there was noconsistent pattern over days.Fights/Attacks. A main effect of group in the group x days ANOVAjust reached significance (p < 0.05), but there were nosignificant differences among groups in post-hoc tests. However,there was a tendency toward a higher number of fights in the IS5group than in the N3N3 group (p = 0.083).107DISCUSSIONThe aim of this study was to determine how psychosocialfactors are related to tumor growth rate and behavior. Groupversus individual housing condition, change in housing anddirection of change, group size, sibling relationship, anddominance status were investigated. These factors are relevant tothose involved in human breast cancer because they are based onsocial interactions and/or alterations in the social environmentwhich can produce or alleviate stress. Similar psychosocialfactors shown to modulate breast cancer in humans include severelife stressors such as loss of a loved one, as well as the amountof social support available to an individual.Tumor Growth RateThe major finding of this study was that animals thatexperienced a change from group to individual housing condition(GI treatment) showed the fastest tumor growth rates, whereasanimals that experienced a change from individual to group housingcondition (IG treatment) showed the slowest tumor growth rates.Groups in which animals did not experience a change in housingcondition (II and GG treatments) displayed intermediate tumorgrowth rates, indicating that the change in housing rather thangroup or individual housing per se was primarily responsible forthe differential tumor growth rates.In an early study in our laboratory (Emerman & Weinberg,1989; Weinberg & Emerman, 1989), we found that animals in the II108condition showed a more rapid tumor growth rate than animals inthe GG condition. However, in subsequent studies on immume andendocrine function, we found that animals in the II condition werenot significantly different from animals in the GG condition inlevels of natural killer cell activity or in testosterone levels.Consistent with this data, tumor growth rate in the II group inthe present study was not significantly different from that of theGG groups (N3N3 and S3S3). It would appear that the finding ofrapid tumor growth rate in II animals in our initial study is notrobust. Since the initial study, our subsequent replicationsusing this animal-tumor model have shown that animals in the IIcondition typically show tumor growth rates that are intermediateto those of animals in the GI and IG conditions and may not besignificantly different from animals in the GG condition.In animals group housed post-injection and group formation,social dominance influenced tumor growth rate differently indifferent treatments and was dependent upon whether animalsexperienced a change in housing. For animals that experienced achange from individual to group housing (IG), subordinate animalsshowed reduced tumor growth rates compared to dominant animals.However, for animals that remained in their rearing groups (GG),subordinate animals showed relatively increased tumor growthrates. Previous studies suggest that dominant animals showgreater resistance to some forms of cancer than subordinateanimals. For example, in a study by Temoshok and colleagues(1987), melanoma appearance was more rapid in subordinate than indominant hamsters. Ebbesen and colleagues (1991) demonstrated109that, for mice housed in small groups, a virally produced leukemiacould be induced in subordinate mice, but could not be induced indominant mice. In the present study, the dominance effects for GGmice but not IG mice are consistent with those reported in theliterature. This finding is understandable, in that animals inthe studies cited above were housed continuously in groups; i.e.,GG condition. Whether social dominance status is alreadyestablished (GG animals) or newly acquired (IG animals) may thusmodulate the effects of dominance on tumor growth rate.In contrast, dominance in the pre-injection period alone(i.e. for GI mice) did not influence tumor growth rate in thepost-injection period, suggesting that any physiological changesrelated to pre-injection dominance either were gone by the post-injection period or were not relevant to tumor growth rate.Consistent with this finding is primate research showing that thephysiological characteristics accompanying dominance depended uponcontinued dominance status, and these characteristics disappearedwhen the animal's rank changed to subordinate (Keverne, Meller, &Eberhart, 1982; Sapolsky, 1983; Shively & Kaplan, 1984).To our knowledge, this is the first study to investigate theeffects of both group size and sibling relationship on tumorgrowth rate. In previous studies in this laboratory (Emerman &Weinberg, 1989; Weinberg & Emerman, 1989), animals in the GI andIG conditions differed on these two variables. The present datademonstrate that all groups within the IG treatment (with thesingle exception of IN3 animals) showed significantly lower tumorgrowth rates than did mice in the GI treatment. Thus, the factors110of group size and sibling relationship are clearly of minorimportance in tumor growth rate. In addition, since there were nosignificant differences in tumor growth rate within the GG or GItreatments, it can be concluded that the characteristics of thepre-injection group, i.e. group size and sibling relationship, arealso unimportant factors in tumor growth rate.Pre-injection BehaviorIn terms of pre-injection behavior, mice housed individuallypre-injection (individually reared) showed more nonsocial behaviorand less sleep than mice housed in groups pre-injection (groupreared). One possible explanation is that group housed mice,unlike individually housed mice, also engaged in aggressive,defensive and social behaviors, resulting in less nonsocialbehavior. The low sleep levels exhibited by individually housedmice, however, suggest that they were actually more active overallthan group housed mice. Increased motor activity has beendemonstrated in a variety of novel environments and in the homecage in individually reared rats (Dalrymple-Alford & Benton, 1984;Einon & Morgan, 1978, Holson et al., 1988; Morgan, 1973) and mice(Essman, 1966, 1968). Our nonsocial category represents a broadspectrum of behaviors; nevertheless, all behaviors in thiscategory require motor activity, and some behaviors (climb, jump,rear, run, walk) require a great deal of motor activity. Thus,the present findings agree with those of previous studies. Theincreased behavioral activity observed here could be due to a highlevel of arousal in the individually reared animals.Interestingly, observations of group housed animals pre-111injection suggest that dominance was the most consistent influenceon behavior. In addition to displaying more aggressive behavior,dominant mice exhibited more nonsocial behavior and more activesocial behavior than subordinate mice. In contrast, subordinatemice showed more defensive behavior, sleep and passive socialbehavior than dominant mice. Thus, dominance was associated witha more active role, and subordinance with a more passive role, forevery aspect of behavior measured. High aggressive behavior indominant animals and high defensive behavior in subordinateanimals has been well-documented in intermale confrontations(Blanchard & Blanchard, 1988; Fauman, 1987; Poole & Morgan, 1973).Other behavioral differences between dominant and subordinateanimals have also been observed. For example, mice exposed torepeated social defeat by an aggressive male showed a more timidand inhibited behavioral profile than did mice exposed to anonaggressive male (Puglisi-Allegra, Cabib, & Mele, 1989).Compared to controls, defeated mice showed much less active socialand nonsocial behavior, and much more defensive behavior andimmobility. These findings are consistent with the behavior ofsubordinate compared to dominant animals in the present study.Other work has shown that the behavior of male mice infectedcongenitally with the parasite Toxoplasma, which causes cysts inthe brain, was similar to that of our dominant mice: highaggressive and active social behavior and low defensive behaviorcompared to that in uninfected mice (Arnott et al., 1990). Unlikeour dominant mice, these infected/dominant mice did not show morenonsocial behavior than uninfected/subordinate mice.112Group size also played an important role in pre-injectionbehavior. Mice housed in groups of five exhibited morefights/attacks and less active, passive and total social behaviorsthan mice housed in groups of three. In addition, animals housedin groups of five slept more than those in groups of three orindividually housed mice. A previous study compared smallergroups (three or four/cage) and larger groups (nine or 12/cage) ofmice, measuring home cage attacks and social rank for 21 daysfollowing group formation (Poole & Morgan, 1973). In the smallergroups, only the dominant mouse exhibited attacks, and this mouseremained dominant throughout the study. In the larger groups,subordinate mice were attacked by other subordinates as well, andchanges in dominance were frequent. Groups of five showed anintermediate social structure; subordinate mice exhibited attackbehavior, but there were no changes in dominance. In our study,it is possible that attacks by subordinate mice housed in groupsof five but not groups of three could account for the increasedtotal attacks observed in groups of five. Because attacks weremeasured by cage rather than by mouse, we cannot determine at thistime how many attacks were initiated by the dominant andsubordinate mice in each group.Compared to mice in groups of three, mice in groups of fivealso showed reduced active and passive social behaviors, whichnormally make up a large amount of the total behavior in our grouphoused mice, and increased sleep/rest. Overall, the resultssuggest that the inhibited social behavior and increasedsleep/rest of mice in groups of five may be a kind of adaptive113behavior that enabled animals to avoid provoking attacks.Post-injection BehaviorOverall effects.^In the post-injection period, overallbehavioral patterns were different at the 1900 hr and post-noveltystress observation times, indicating that the factors governingbehavior are modulated both by time of day and by whether animalsare exposed to stressors prior to observation. Behavioralactivity was lower in the post-novelty stress observations than inthe 1900 hr observations. In our colony, mice typically sleepduring the lights-on portion of the cycle and are active duringthe lights-off portion of the cycle. During the 1900 hrobservations, which took place following lights-off, behavioralactivity was high throughout the observation session. Incontrast, in the post-novelty stress observations, mice initiallyshowed high overall behavioral activity and low sleep. Thus,exposing mice to daily novelty stress increased behavioralactivity beyond the normal amount for the lights-on portion of theday, but levels dropped to normal over the course of theobservation session, so that most mice were asleep or resting bythe end of the observation.The amount of overall behavior varied over days. At 1900 hrbehavior was markedly higher on the injection day (day 0) than onall other days, for all behaviors except sleep, which showed theopposite pattern. The drop in behavior from day 0 to day 1 wasseen most dramatically in the IG animals. Therefore, the changein housing from individual to group on day 0 (perhaps compounded114by injection stress) stimulated behavioral activity, but theanimals appeared to adapt within a day or two. In contrast, atthe post-novelty stress observation time, behavioral activity wasrelatively stable over all observation days.Nonsocial behavior and sleep/rest. Group housed animalsspent most of their time in social behaviors, but it was foundthat even group housed animals spent a significant portion oftheir time engaged in nonsocial behavior. Direction of change inhousing was the most consistent predictor of nonsocial behaviorand sleep in the post-injection period. At both observationtimes, GI mice showed more nonsocial behavior and less sleep thanIG mice, but for the post-novelty stress observations, theincreased level of nonsocial behavior was no longer evident afterthe first five observation days. It is possible that IG miceshowed less nonsocial behavior than GI mice because they alsoengaged in aggressive, defensive and social behaviors. However,the finding of a high level of sleep for IG mice suggests thatthese animals were genuinely less active than GI mice. Previousstudies suggest that adult-isolated mice previously housed ingroups (GI) show characteristic physiological signs of stress suchas increased adrenal and gonadal weights and increased plasmacorticosterone and testosterone (Brain, 1975, Sayegh, 1990). GImice also show hyperreactivity to stimuli (Brain, 1975), anddecreased locomotion and exploration in novel environments(Valzelli, 1969, 1973). In the present study, the high activitylevels exhibited by GI mice appear to contradict previous resultsof decreased locomotion and exploration. However, it is possible115that activity may be decreased in novel environments whileincreased in the home cage. Increased home cage behavior appearsto be consistent with increased anxiety and stress, which couldcontribute to the high tumor growth rate observed in GI mice. Itwould be interesting to observe nonsocial behavior and sleep/restin differentially housed mice during exposure to novelenvironments, and to compare those data with behavioral activityin the home cage.Dominance. As expected, in the post-injection analyses,dominant mice were consistently high in aggressive and low indefensive behavior, while subordinate mice were high in defensiveand low in aggressive behavior. Over all eight 1900 hrobservations, dominant mice showed higher active social behaviorand lower passive social behavior than subordinate mice, andwithin the IG treatment, dominant mice also showed more nonsocialbehavior and less sleep than subordinate mice. Dominance resultedin a generalized increase in behavioral activity, which concurswith the pre-injection observations and with other behavioralstudies on dominance, as described in the discussion of pre-injection data.Aggressive behavior. Aggressive behavior was influenced bydominance as well as by other factors. On the day of injectionand rehousing, dominant mice in IG nonsibling groups showed moreaggressive behavior than those in IG sibling groups. Within 24-48hr, however, aggression in IG nonsibling groups decreased and wasno longer different trom that in IG sibling groups. Mice insibling groups may have been familiar with one another from the116preweaning period, and this may have reduced the aggressivebehavior observed initially. In another study, dominant mice ingroups of three to five showed an exponential decline inaggressive behavior over a 21-day period following group formation(Poole & Morgan, 1973), but in the present study, no significantchange in aggressive behavior was found at 1900 hr following theinitial drop from day 0 to day 1. This suggests that our strainis more aggressive than other strains. In the post-novelty stressobservations, aggressive behavior was not affected by socialhousing conditions. However, dominant mice were more aggressivethan subordinate mice, particularly on days 5-12. Together, thesedata suggest that neither group size nor sibling relationship wasa significant factor in aggressive behavior.Fights/attacks. In terms of fighting at 1900 hr, IG animalsfought most, and within the IG treatment, groups of five showedmore fighting than groups of three, consistent with the pre-injection results. Increased fighting in groups of five mayreflect high instability of the social group compared to groups ofthree. For the post-novelty stress observations, there were nosignificant differences among groups.For animals housed in groups post-injection, those thatexperienced a change from individual to group housing conditions(IG) showed more fights/attacks than those that did not experiencechange (GG) at the 1900 hr observation time. There is extensiveevidence that for many strains of mice, being individually housedrather than group housed increases intermale aggression, aphenomenon described as isolation-induced aggression (Brain, 1975;117Lister & Hilakivi, 1988; Valzelli, 1969, 1973). Prior to thisstudy, isolation-induced aggression was typically documentedbetween pairs of unfamiliar mice in a neutral testing arena. Thepresent study, however, found that a similar phenomenon may occurin the home cage. Thus, increased fighting following socialisolation does not appear to be dependent upon the novelty of thetesting environment. Adult isolation has also been shown toincrease social behavior (Brain, 1975; Lister & Hilakivi, 1988),but we failed to find increased social behavior in our IG mice.Thus, although increased fighting may occur in both a novelenvironment and the home cage, in our study, familiarity withcagemates might be a factor that decreased the tendency to engagein social behavior.The present data suggest that for IG mice, frequent home-cagefighting may play a critical role in their decreased tumor growthrates compared to those in GG mice. Previous research supportsthis suggestion that for group housed animals, fighting maysuppress cancer development. Home cage fighting has been shown toreduce tumor growth in both male and female mice (Amkraut &Solomon, 1972; Sklar & Anisman, 1980). In addition, delayedcancer appearance was associated with home-cage fighting in femalehamsters exposed to daily cage-shaking stress (Temoshok et al.,1987).Defensive behavior. Subordinate mice showed more defensivebehavior than dominant mice at 1900 hr. Defensive behavior wasalso influenced by change in housing conditions; IG mice showedhigh defensive behavior at 1900 hr. Whether or not defensive118behavior contributes to the decreased tumor growth rate in the IGtreatment, especially in subordinate mice, is not clear at thistime. Submissive behavior in the home cage was correlated withearlier tumor appearance in female hamsters (Temoshok et al.,1987), indicating that data on tumor growth and defensive behavioris not consistent. The discrepancy in effects observed may dependon experimental variables such as species and sex of animals, ortype of tumor (Justice, 1985). In addition, at 1900 hr, IG miceshowed more defensive behavior than GG mice.On day 1 post-novelty stress, mice in the IS5 group showedmore defensive behaviors than mice in the other groups, but therewas no effect of dominance. Over the remaining post-noveltystress observations, mice in the IN3 group showed high defensivebehavior across days. Interestingly, mice in the IN3 group alsoshowed a higher tumor growth rate than mice in the other IGgroups. It appears that whereas high defensive behavior at 1900hr is negatively associated with tumor growth rate, high defensivebehavior post-novelty stress is positively associated with tumorgrowth rate in the IN3 group. More research is needed in order tounderstand the relationships among subordinance, defensivebehavior, and tumor growth rate.In mice housed continuously in groups (GG), the literatureindicates that subordinance is stressful (McKinney & Pasley,1973). If social stress stimulates tumor growth, then one wouldexpect subordinate mice to show a high tumor growth rateregardless of housing conditions. Although subordinance wasaccompanied by a high tumor growth rate in the GG groups, it was119accompanied by a low tumor growth rate in the IG groups,indicating that the stressfulness of subordinance is modulated bysocial housing conditions. In terms of behavior, subordinate micein the IG groups showed the most defensive/receiving aggressivebehavior, suggesting that subordinates in the IG groupsexperienced the most social stress. However, there was also agreater amount of fighting in IG animals, which in itself mightmodulate tumor growth rate, and could contribute to the differencein tumor growth rates of subordinate IG and GG animals. Althoughthe same animals remained dominant throughout the study, the datasuggest that dominant IG animals fought more to maintain theirstatus than did dominant GG animals. It is likely that socialdominance relationships in GG groups, which are alreadyestablished, are more stable than those in IG groups, which arenewly formed. In our study, the dominant mouse in each cagetended to participate in most of the fights occurring in thatcage. For dominant mice in GG groups, fighting was at a lowlevel, but for dominant mice in IG groups, fighting was frequentand may thus have been a source of stress. Support for thissuggestion comes from a study of psychosocial factors in malemacaque monkeys, in which stable and unstable housing groups werecompared for degree of coronary artery atherosclerosis in animalsfed a high fat diet (Kaplan et al., 1982). Unstable groups werecreated by frequent reorganization (rehousing) of group members.Highly aggressive dominant males that remained dominant throughoutthe study had high atherosclerosis, but only if housed underconditions of social instability. Subordinate monkeys and120dominant monkeys in the stable groups had relatively lowatherosclerosis. Similar results were found for animals on a lowfat diet (Kaplan et al., 1983). Thus, our data suggest that tumorgrowth rate in influenced by dominance, but that the effect ofdominance is modulated by the stressfulness of social housingconditions. One may suggest that dominant males in IG groups,which fight more than dominant males in GG groups, are under morestress than dominant males in GG groups.The immune system may mediate the interactive effects ofchange in housing conditions and dominance on tumor growth rate.Dominance has been shown to produce decreased antibodyresponsiveness in groups of mice rehoused from individual to agroup of two or eight (Fauman, 1987). In addition, for mice in IGgroups of two, the magnitude of antibody responsiveness waspositively associated with submissive behavior and negativelyassociated with aggressive behavior. Immune function has alsobeen studied in animals exposed to daily fighting bouts with anaggressive male conspecific in a neutral cage. In this type ofsituation, repeated social defeat results in immunosuppression inrats and mice (Hardy et al., 1990; Ito et al., 1983; Fleshner etal., 1989).Social behaviors. While aggressive and defensive behaviorswere strongly modulated by dominance at both 1900 hr and post-novelty stress, social behaviors were modulated by dominance onlyat 1900 hr. At 1900 hr, as in the pre-injection observations,dominant mice showed more active social and less passive socialbehavior than subordinate mice. These effects did not emerge121until after the first two observations, suggesting that in mostgroups, immediately following group formation, dominant andsubordinate mice were engaged in aggressive and defensivebehaviors instead of active and passive social behaviors. Also at1900 hr, active, passive and total social behaviors were highestin the IS5 group. Mice rehoused in sibling groups of five mightrecognize their siblings from the preweaning period, which couldexplain the increased social behavior. During the preweaningperiod, siblings engage in frequent social interactions(unpublished observations). It is not clear why sibling groups offive but not three show high social behavior. It is possible thatthe higher concentration of animals in groups of five may increasethe probability of social interactions.Consistent with 1900 hr results, on day 1 of post-noveltystress observations mice in the IS5 group showed the highestsocial behavior. After day 1, there was a main effect of groupsize and a main effect of sibling relationship, not due solely tothe IS5 group. Siblings were higher than nonsiblings and groupsof five were higher than groups of three in active, passive andtotal social behaviors. The group size effect is opposite to thatfound during pre-injection observations, when groups of threeshowed the most social behaviors. A number of factors, such aschange in housing, duration of group housing, exposure to noveltystress, or the presence of tumors, could contribute to thediscrepancy of the pre- and post-injection results. For example,prior to the start of behavioral observations in this study,animals were housed together for an average of eight weeks in the122pre-injection period. In contrast, animals were housed in groupsfor only about three weeks in the post-injection period. Socialbehaviors may be differentially affected by the duration of grouphousing in groups of three versus groups of five.Active, passive and total social behaviors were higher insibling than nonsibling groups during the post-novelty stressobservations, primarily during days 1-5. In contrast, there wasno effect of sibling relationship on social behaviors during thepre-injection observations. Thus, sibling relationship had onlya limited influence on social behavior, apart from its interactionwith group size.Conclusion. In the present study, tumor growth rate wasdetermined primarily by direction of change in housing; a changefrom group to individual increased tumor growth rate whereas achange from individual to group decreased tumor growth rate, asfound in our previous studies. Previous findings of a high tumorgrowth rate in mice in the II treatment were not replicated.Within the IG groups, dominant mice had a faster tumor growth ratethan subordinate mice, but within the GG groups, the oppositeoccurred. Neither social housing conditions nor behavior duringthe pre-injection period appeared to have any infuence on tumorgrowth rate. Thus, change in social housing conditions and socialstatus within a group appear to have an interactive effect ontumor growth rate.Pre-injection behavioral analysis revealed that individuallyhoused mice showed the most nonsocial behavior whereas grouphoused animals showed the most sleep. Groups of five slept most123and fought most, and groups of three showed the most socialbehavior. Dominant animals were more active overall, showing themost nonsocial, aggressive, and active social behavior, whereassubordinate animals showed the most sleep, defensive, and passivesocial behavior. Pre-injection behavior and housing did notappear to influence tumor growth rate.In the post-injection observations, mice consistently showedmore behavioral activity at 1900 hr than post-novelty stress. At1900 hr, behavioral activity was generally much higher on day 0than on any day thereafter. In contrast, for the post-noveltystress observations, behavior was low overall and showed onlyminor variations over days.Of all the housing conditions, GI mice showed the highestlevel of behavioral activity, consistent with literature showingother behavioral and physiological indicators of stress occurringin GI mice. The apparent stress experienced by mice in the GIhousing condition may contribute to a high tumor growth rate.Of mice housed in groups post-injection, IG mice fought most,especially in groups of five, and showed the most defensivebehavior as well. Fighting and defensive behavior may help reducetumor growth rate in these mice. Social behaviors were highest ingroups of five and in sibling groups. Dominant mice played a moreactive role than did subordinate mice, exhibiting highernonsocial, aggressive, and active social behavior and lower sleep,defensive and passive social behavior.In summary, a change in housing from group to individualincreases tumor growth rate, whereas a change from individual to124group decreases tumor growth rate. A change from individual togroup also results in increased fighting and defensive behavior.Dominance is related to increased tumor growth rate in IG groupsbut to decreased tumor growth rate in GG groups. Regardless ofhousing, dominant animals are more active, while subordinateanimals are more passive. Neither group size nor siblingrelationship per se appear to have much effect on tumor growthrate. It is unlikely that aggressive, social, or nonsocialbehavior or sleep have much influence on tumor growth rate, butthey do reflect an animal's social status and housing condition.A possible limitation of this study is that in the behavioralobservations, the experimenter was aware of the housing conditionsof the animals. This situation was unavoidable because there wasonly one experimenter to conduct all aspects of the study; e.g.rehousing, stressor application, and behavioral observations; thusraising the possibility of observer bias. To reduce the potentialfor bias, the observation procedure was designed to be assystematic as possible. All behaviors were strictly defined priorto the beginning of the study (see Table 2, p. 27). During thebehavioral observations, a routine procedure was followed in whicheach mouse was observed as a focal animal for four 30-secondintervals, spaced evenly over the course of the session. Eachbehavior occurring during the observation interval was recorded ona standardized sheet. Animals were observed in the samepredetermined order for each observation session. Due to thesystematic nature of the protocol, there was little room forbiased collection of data based upon knowledge of the animals'125housing conditions. Similarly, the data were analyzed in anobjective, systematic manner, further reducing the probability ofexperimenter bias.This animal-tumor model is currently being utilized toinvestigate the physiological pathways that may mediate theobserved differences in tumor growth rate, such as pituitary-adrenal, pituitary-gonadal and immune functions. Future studiesusing this model should investigate the mediating role ofdominance and agonistic behavior on tumor growth rate.126REFERENCESAarstad, H. J.; Thiele, D.; Seljelid, R. 1991. The effect ofvarious contexts of stress on the mouse spleen lymphocytesand macrophage co-stimulatory activity. Scandinavian Journal of Immunology 33:461-72.Amkraut, A.; Solomon, G. F. 1972. Stress and murine sarcomavirus (Moloney)-induced tumors. Cancer Research32:1428-1433.Armario, A.; Garcia-Marquez, C.; John, T.^1987.^Crowding-induced changes in basal and stress levels of thyrotropin andsomatotropin in male rats. Behavioral and Neural Biology48:334-43.Arnott, M. A.; Cassella, J. P.; Aitken, P. P.; Hay, J. 1990.Social interactions of mice with congenital Toxoplasmainfection. Annals of Tropical Medicine and Parasitology84:149-156.Baider, L; Peretz, T.; Kaplan De-Nour, A. 1992. Effect of theHolocaust on coping with cancer. Social Science & Medicine34:11-15.Baltrusch, H. J.; Stangel, W.; Titze, I. 1991. Stress, cancerand immunity. New developments in biopsychosocial andpsychoneuroimmunologic research. Acta Neurologica 13:315-27.Barinaga, M.^1989.^Can psychotherapy delay cancer deaths?Science 246:448-449.Bartrop, R. W.; Lazarus, L.; Luckhurst, E.; Kiloh, L. G.; Penny,R. 1977. Depressed lymphocyte function after bereavement.Lancet 1:834-836.Berkman, L. F.; Syme, S. L.^1979.^Social networks, hostresistance and mortality: A nine-year follow-up study ofAlameda County residents. American Journal of Epidemiology109:186-204.Blanchard, D. C.; Blanchard, R. J.^1988.^Ethoexperimentalapproaches to the biology of emotion. Annual Review ofPsychology 39:43-68.Blanchard, R. J.; Flannelly, K. J.; Blanchard, D. C.^1988.Life-span studies of dominance and aggression in establishedcolonies of laboratory rats. Physiology and Behavior 43:1-7.Brain, P. F.^1975. What does individual housing mean to amouse? Life Science 16:187-200.Bruchovsky, N.; Rennie, P. S.^1978.^Classification of127dependent and autonomous variants of Shionogi mammarycarcinoma based on heterogeneous patterns of androgenbinding. Cell 13:273-280.Burchfield, S. R.; Woods, S. C.; Elich, M. S. 1978. Effects ofcold stress on tumor growth. Physiology and Behavior21:537-400.Burright, R. G.; Freeman, M. J.; Donovick, P. J.^1988.Repeated tests of intermale aggression in mice (Musmusculus)are influenced by housing and test conditons. Journal ofComparative Psychology 102:303-305.Chiueh, C. C.; McCarty, R.^1981.^Sympatho-adrenalhyperreactivity to footshock stress but not to cold exposurein spontaneously hypertensive rats. Physiology and Behavior26:85-89.Cohen, F. 1984. Coping. In: Matarazzo, J. D.; Weiss, S. M.;Herd, J. A.; Miller, N. E., eds. Behavioral Health. NewYork: John Wiley & Sons; 261-274.Cooper, C. L.; Cooper, R.; Faragher, E. B. 1989. Incidence andperception of psychosocial stress: The relationship withbreast cancer. Psychological Medicine 19:415-422.Cutler, M. G.; Piper, D. C. 1970. Chronic administration ofthe 5-HT3 receptor antagonist BRL 43694: Effects on reflexepilepsy and social behaviours of the Mongolian gerbil.Psychopharmacology. 101:244-249.Dalrymple-Alford, J. C.; Benton, D.^1984.^BehaviouralInhibition and the age at social isolation in rats.Ouarterly Journal of Experimental Psychology 36B:27-38.Dechambre, R. P.; Gosse, C. 1973. Individual vs. group cagingof mice with grafted tumours. Cancer Research 33:140-144.Dewsbury, D. A.^1988. Kinship, familiarity, aggression anddominance in deer mice (Peromyscus maniculatus) inseminatural enclosures. Journal of Comparative Psychology102:124-128.Dorian, B.; Garfinkel, P. E.^1987.^Stress, immunity andillness -- a review. Psychological Medicine 17:393-407.Ebbesen, P.; Villadsen, J. A.; Villadsen, H. D.; Heller, K. E.1991. Effect of subordinance, lack of social hierarchy, andrestricted feeding on murine survivial and virus leukemia.Experimental Gerontology 26:479-486.Einon, D. F.; Morgan, M. J.^1978. Early isolation producesenduring hyperactivity in the rat, but no effect uponspontaneous alternation.Psychology 30:151-156.Ekman, P.; Snochowski, M.;Gustafsson, J.A. 1979.prostatic carcinoma andCancer 44:1173-1181.128Ouarterly Journal of Experimental Zetterberg, A.; Hogberg, B.;Steroid receptor content in humanresponse to endocrine therapy.Emerman, J. T.; Weinberg, J. 1989. Influence of psychosocialstress on mouse mammary tumor growth. In: Zinder, O.;Bresnitz, S., eds. Molecular Biology of Stress, UCLASymposia, vol. 97. New York: Alan R. Liss, Inc.; 295-305.Ewertz, M.^1986.^Bereavement and breast cancer.^BritishJournal of Cancer 53:701-703.Fauman, M. A. 1987. The relation of dominant and submissivebehavior to the humoral immune response in BALB/c mice.Biological Psychiatry 22:776-779.Fleshner, M.; Laudenslager, M. L.; Simons, L.; Maier, S. F.1989. Reduced serum antibodies associated with social defeatin rats. Physiology and Behavior 45:1183-1187.Fox, B. H. 1981. Psychosocial factors and the immune system inhuman cancer. In: Ader, R., ed. Psychoneuroimmunology. NewYork: Academic Press; 103-157.Gamallo, A. Villanua, A.; Trancho, G.; Fraile, A. 1986. Stressadaptation and adrenal activity in isolated and crowded rats.Physiology and Behavior 36:217-221.Gehde, E.; Baltrusch, H. J.^1990.^Early experience anddevelopment of cancer in later life: Implications forpsychoneuroimmunologic research. International Journal ofNeuroscience 51:257-260.Geller, J.; Albert, J.; Nachtsheim, D.; Loza, D.; Lippman, S.1979. Steroid levels in cancer of the prostate: Markers oftumor differentiation and adequacy of anti-androgen therapy.Progress in Clinical Biological Research 33:103-111.Gentsch, C.; Lichtsteiner, M.; Driscoll, P.; Feer, H.^1981.Differential hormonal and physiological responses to stressin Roman high- and low-avoidance rats. Physiology & Behavior28:259-263.Giralt, M.; Armario, A.^1989.^Individual housing does notinfluence the adaptaion of the pituitary-adrenal axis andother physiologic variables to chronic stress in adult malerats. Physiology and Behavior 45:477-481.Glaser, R.; Rice, J.; Speigcher, C. E.; Stout, J. C.;129Kiecolt-Glaser, J. K. 1986. Stress depresses interferonproduction by leukocytes concomitant with a decrease innatural killer cell activity. Behavioral Neuroscience100:675-678.Greenberg, A. H.; Dyck, D. G.; Sandler, L. S. 1984. Opponentprocesses, neurohormones and natural resistance. In: Fox,B. H.; Newberry, B. H., eds. Impact of PsychoneuroendocrineSystems on Cancer and Immunity. Toronto: C. J. Hogrefe,Inc.; 225-257.Greenman, D. L.; Kodell, R. L.; Sheldon, W. G. 1984.Association between cage shelf level and spontaneous andinduced neoplasms in mice. JNCI. Journal of the National Cancer Institute 73:107-113.Greer, S.; Morris, T. 1975. Psychological attributes of womenwho develop breast cancer: A controlled study. Journal ofPsychosomatic Research 19:147-153.Greer, S.; Morris, T.; Pettingale, K. W. 1979. Psychologicalresponse to breast cancer: Effect on outcome. Lancet2:785-787.Grossarth-Maticek, R.; Eysenck, H. J. 1989. Length of survivaland lymphocyte percentage in women with mammary cancer and afunction of psychotherapy. Psychological Reports 65:15-321.Hardy, C.; Quay, J.; Livnat, S.; Ader, R. 1990. Altered T-lymphocyte response following aggressive encounters in mice.Physiology and Behavior 47:1245-1251.Hislop, T. G.; Waxier, N. E.; Goldman, A. J.; Elwood, J. M.;Kan, L. 1987. The prognostic significance of psychosocialfactors in women with breast cancer. Journal of ChronicDisease 40:729-735.Hoffman-Goetz, L.; Simpson, J. R.; Arumugam, Y. 1991. Impactof housing condition on mouse natural killer cell activity.Physiology and Behavior 49:657-660.Holson, R. R.; Scalett, A. C.; Ali, S. F.; Sullivan, P.; Gough,B. 1988. Adrenocortical, beta-endorphin and behavioralresponses to graded stressors in differentially reared rats.Physiology and Behavior 42:125-130.Irwin, J.; Ahluwalia, P.; Zacharko, R. M.; Anisman, H. 1986.Central norepinephrine and plasma corticosterone followingacute and chronic stressors: Influence in social isolationand handling.^Pharmacology Biochemistry and Behavior24:1151-1154.Irwin, M.; Daniels, M.; Bloom, E. T.; Smith, T. L.; Weiner, H.1301987. Life events, depressive symptoms, and immune function.American Journal of Psychiatry 144:437-441.Ito, Y.; Mine, K.; Ago, Y.; Nakagawa, T.; Fujiwama, M.; Veki, S.1983. Attack stress and IgE antibody production in rats.Pharmacology, Biochemistry and Behavior 19:883-886.Jessop, J. J.; Bayer, B. M. 1989. Time-dependent effects ofisolation on lymphocyte and adrenocortical acivity. Journal of Neuroimmunology 23:143-147.Jessop, J. J.; Gale, K.; Bayer, B. M. 1987. Enhancement of ratlymphocyte proliferation after prolonged exposure to stress.Journal of Neuroimmunology 16:261-271.Jessop, J. J.; Gale, K.; Bayer, B. M. 1988. Time-dependentenhancement of lymphocyte activation by mitogens afterexposure to isolation or water scheduling. Life Science43:1133-1140.Justice, A. 1985. Review of the effects of stress on cancer inlaboratory animals: Importance of time of stress applicationand type of tumor. Psychological Bulletin 98:108-138.Jurcovicova, J.; Vigas, M.; Klir, P.; Jezova, D. 1984.Response of prolactin, growth hormone and corticosteronesecretion to morphine administration or stress exposure inWistar-AVN and Long Evans rats. EndocrinologiaExperimentalis 18:209-214.Kandil, 0.; Borysenko, M. 1987. Decline of natural killer celltarget binding and lytic activity in mice exposed to rotationstress. Health Psychology 6:89-99.Kandil, 0; Borysenko, M. 1988. Stress-induced decline inimmune responsiveness in C3H/HeJ mice: Relation to endocrinealterations and tumor growth. Brain, Behavior and Immunity2:32-49.Kant, G. J.; Lenox, R. H.; Bunnell, B. N.; Mougey, E. H.;Pennington, L. L.; Meyerhoff, J. L. 1983a. Comparison ofacute stress mseponses in male and female rats: Pituitarycyclic AMP and plasma prolactin, growth hormone, andcorticosterone. Psychoneuroendocrinoloay 8:421-428.Kant, G. J.; Mougey, E. H.; Pennington, L. L.; Meyerhoff, J. L.1983b. Graded footshock stress elevates pituitary cyclic AMPand plasma beta-endorphin, beta-LPH, corticosterone andprolactin. Life Sciences 33:2657-2663.Kaplan, J. R.; Manuck S. B.; Clarkson, T. B.; Lusso, F. M.;Taub, D. M. 1982. Social status, environment andatherosclerosis in cynomolgus monkeys.^Arteriosclerosis1312:359-368.Kaplan, J. R.; Manuck S. B.; Clarkson, T. B.; Lusso, F. M.;Taub, D. M.; Miller, E. W. 1983. Social stress andatherosclerosis in normocholesterolemic monkeys. Science220: 773-735.Keverne, E.; Meller, R.; Eberhart, J.^1982.^Dominance andsubordination:^concepts of physiological states?^In:Chiarelli, V.; Corruccini, J., eds.^Advanced Views inPrimate Biology.  New York: Springer-Verlag; 223-247.Kiecolt-Glaser, J. K.; Glaser, R.; Williger, D.; Stout, J.;Messick, G.; Sheppard, S. et al. 1985a. Psychosocialenhancement of immunocompetence in a geriatric population.Health Psychology 4:25-41.King, R. B. J.; Yates, J. 1980. The use of cultured mammarytumor cells to study effects of steroid hormones.^In:Richards, R. J.; Rajan, K. T., eds.^Tissue Culture inMedical Research II. New York: Pergamon Press; 221-227.Laudenslager, M.; Ryan, S. M.; Drugan, R. C.; Hyson, B. L.;Maier, S. F. 1973. Coping and immunosuppression:Inescapable but not escapable shock suppresses lymphocyteproliferation. Science 221:568-570.Laudenslager, M. L.; Fleshner, M.; Hofstadter, P.; Held, P. E.;Simons, L.; Maier, S. F. 1988. Suppression of specificantibody production by inescapable shock: Stability undervarying conditions. Brain, Behavior and Immunity 2:92-101.Levy, S. M. 1985. Behavioral factors contributing directly toprogression of cancer. In: Levy, S. M., ed. Behavior andCancer. San Fransisco: Jossey-Boss Publishers; 148-180.Levy, S. M.; Herberman, R. B.; Maluish, A. M.; Schlein, B.;Lippman, M. 1985. Prognostic risk assessment in primarybreast cancer by behavioral and immunological parameters.Health Psychology 4:99-113.Levy, S. M.; Herberman, R. B.; Whiteside, T.; Sanzo, K.; Lee,J.; Kirkwood, J. 1990. Perceived social support and tumorestrogen/progesterone receptor status as predictors of NKcell activity in breast cancer patients. Psychosomatic Medicine 52:73-85.Lewis, J. W.; Shavit, Y.; Terman, G. W.; Gale, R. P.;Liebeskind, J. C. 1983/84. Stress and morphine affectsurvival of rats challenged with a mammary ascites tumor (MAT13762B). Natural Immunity and Cell Growth Regulation3:43-50.132Linkins, R. W.; Comstock, G. W.^1990.^Depressed mood anddevelopment of cancer. American Journal of Epidemiology132:962-72.Lister, R. G; Hilakivi, L. A. 1988. The effects of novelty,isolation, light and ethanol on the social behavior of mice.Psychopharmacology 96:181-187.Maes, S.; Vingerhoets, A.; Van Heck, G.^1987. The study ofstress and disease: Some developments and requirements.Social Science and Medicine 25:567-578.Maier, S. F.; Seligman, M. E. P. 1976. Learned helplessness:Theory and evidence. Journal of Experimental Psychology[Generall 105: 3-46.Marsh, J. T.; Miller, B. E.; Lamson, B. G.^1959.^Effect ofrepeated brief stress on growth of Ehrlich carcinoma in mice.JNCI. Journal of the National Cancer Institute 22:961-977.McKinney, T. D.; Pasley, J. N. 1973. Effects of social rankand social disruption in adult male house mice. General andComparative Endocrinology 20: 579-583.Miller, R. G.^1987.^Immunology related to cancer.^In:Tannock, I. F.; Hill, R. P., eds.^The Basic Science ofOncology. Toronto: Pergamon Press; 223-233.Monjan, A. A.^1981.^Stress and immunologic competence:Studies^in^animals.^In:^Ader,^R.,^ed.Psychoneuroimmunology. New York: Academic Press; 185-228.Monjan, A. A.; Collector, M. I.^1977.^Stress-inducedmodulation of the immune response. Science 196:307-308.Morgan, M. J. 1973. Effects of post-weaning environment onlearning in the rat. Animal Behavior 21:429-442.Neiburgs, H. E.; Weiss, J.; Navarette, M.; Strax, P.; Teirstein,A.; Grillione, G.; Siedlecki, B. 1979. The role of stressin human and experimental oncogenesis. Cancer Detection andPrevention 2:307-336.Psychoneuroendocrine Systems on Cancer and Immunity.Toronto: C. J. Hogrefe, Inc.; 86-146.Variablesof currentImpact ofNewberry, B. H.; Liebelt, A. G.; Boyle, D. A. behavioral oncology: Overview and assessmentissues. In: Fox, B. H.; Newberry, B. H., eds.Noguchi, S.; Nichizawa, Y.; Nakamura, N.; Uchida, N.; Yamaguchi,K.; Sato, B.; Kitamura, Y.; Matsumoto, K. 1987. Growth-stimulating effect of pharmacological doses of estorogen onandrogen-dependent Shionogi carcinoma 115 in vivo but not in133cell culture. Cancer Research 47:263-268.Orr, T. E.; Meyerhoff, J. L.; Mougey, E. H.; Bunnell, B. N.1990. Hyperresponsiveness of the rat neuroendocrine systemdue to repeated exposure to stress. Psychoneuroendocrinoloay15:317-328.Plaut, S. M.; Friedman, S. B.; Grota, L. J. 1971. Plasmodiumberghei: Resistance to infection in group and individuallyhoused mice. Experimental Parasitoloav 29:47-52.Poole, T. B.; Morgan, H. D. 1973. Differences in aggressivebehaviour between male mice (Mus musculus L.) in coloniesof different sizes. Animal Behaviour 21:788-95.Priestman, T. J.; Priestman, S. G.; Bradshaw, C. 1985. Stressand breast cancer. British Journal of Cancer 51:493-498.Puglisi-Allegra, S.; Cabib, S.; Mele, A.^1989.^Stress andsocial behavior in the mouse: Experimental models forpreclinicalpsychopharmacology. In: Brain, P. F. ; Mainardi,D.; Parmigiani, S., eds. House Mouse Aggression.  New York:Harwood Academic Publishers; 253-272.Rabin, B. S.; Lyte, M.; Epstein, L. H.; Caggiula, A. R. 1987.Alteration of immune competency by number of mice housed percage. Annals of the New York Academy of Sciences 496:492-500.Rabin, B. S.; Lyte, M.; Hamill, E.^1987.^The influence ofmouse strain and housing on the immune response. Journal ofNeuroimmunoloay 17:11-16.Riley, V.; Fitzmaurice, M. A.; Spackman, D. H.^1981.Psychoneuroimmunologic factors in neoplasia: Studies inanimals. In: Ader, R., ed. Psychoneuroimmunoloay. NewYork: Academic Press; 31-102.Ramirez, A. J.; Craig, T. K. J.; Watson, J. P.; Fentiman, I. S.;North, W. R. S.; Rubens, R. D. 1989. Stress and relapse ofbreast cancer. British Medical Journal 298:291-293.Rodriguez-Enchandia, E. L.; Gonzalez, A. S.; Cabrera, R.;Fracchia, L. N. 1988. A further analysis of behavioral andendocrine effects of unpredictable chronic stress.Physiology and Behavior 43:789-795.Rowse, G. J.; Rowan, R. E.; Weinberg, J.; Emerman, J. T. 1990.Alterations in splenic natural killer cell activity inducedby the Shionogi mouse mammary tumor. Cancer Letters 54:81-87.Rowse, G. J.; Weinberg, J.; Bellward, G. D.; Emerman, J. T.1992. Endocrine mediation of psychosocial stressor effects on134mouse mammary tumor growth. Cancer Letters 65:85-93.Sayegh, J. F.; Kobor, G.; Lajtha, A.; Vadasz, C. 1990. Effectsof social isolation and the time of day on testosteronelevels in plasma of C57BL/6By and BALB/cBy mice. Steroids55:79-82.Scherg, H.^1987.^Psychosocial factors and disease bias inbreast cancer patients. Psychosomatic Medicine 49:302-312.Scherg H.; Cramer, I.; Blohmke, M. 1981. Psychosocial factorsand breast cancer: A critical reevaluation of establishedhypotheses. Cancer Detection and Prevention 4:165-171.Scherg, H.; Blohmke M.^1988. Associations between selectedlife events and cancer. Behavioral Medicine 14:119-124.Selye, H. 1936. A syndrome produced by diverse nocuous agents.Nature 138:1-32.Shanks, N.; Griffiths, J.; Zalcman, S.; Zacharo, R. M.; Anisman,J. 1990. Mouse strain differences in plasma corticosteronefollowing uncontrollable footshock. Pharmacology. Biochemistry and Behavior 36:515-519.Shively, C.; Kaplan, J.^1984. Effects of social factors onadrenal weight and related physiology of Macaca fascicularis. Physiology and Behavior 33:777-783.Simmel, E. C.; Wright, J. C.; Smith, M.^1974.^Changes inleukocyte levels associated with social-rearing condition inC57BL/10J mice. Bulletin of the Psychonomic Society3:269-270.Simpson-Herren, L.; Lloyd, H. K. 1970. Kinetic parameters andgrowth curves for experimental tumor systems. CancerChemotherapy Reports 54:143-174.Sklar, L. S.; Anisman, H.^1979.^Stress and coping factorsinfluence tumor growth. Science 205:513-515.Sklar, L. S; Anisman, H. 1980. Social stress influences tumorgrowth. Psychosomatic Medicine 42:347-365.Sklar, L. S.; Anisman, H.^1981.^Stress and Cancer.Psychological Bulletin 89:369-406.Sklar, L. S.; Bruto, V.; Anisman, A. 1981. Adaptation to thetumor-enhancing effects of stress. Psychosomatic Medicine43:331-342.Sorenson, G. 1987. Stereotyped behaviour, hyperaggressivenessand "tyrannic" hierarchy induced in bank voles (Clethrionomys135Glareolus) by a restricted cage milieu.^Progress inNeuropsychopharmacology and Biological Psychiatry 11:9-21.Spiegel, D.; Bloom, J. R.; Kraemer, H. C.; Gottheil, E. 1989.Effect of psychosocial treatment on survival of patients withmetastatic breast cancer. Lancet, 2:888-891.Stefanski, V.; Hendrichs, H.; Ruppel, H. G.^1989.^Socialstress and activation of the immune system in guinea pigs.Naturwissenschaften 76:225-226.Steplewski, Z.; Goldman, P. R.; Vogel, W. H. 1987. Effect ofhousing stress on the formation and development of tumours inrats. Cancer Letters 34:257-261.Steplewski, Z.; Vogel, W. H.; Ehya, H.; Poropatich, C.; McDonaldSmith, J. 1985. Effects of restraint stress on inoculatedtumor growth and immune response in rats. Cancer Research45:5128-5133.Tejwani, G. A.; Gudehithlu, K. P.; Hanissian, S. H.; Gienapp, I.E.; Whitacre, C. C.; Malarkey, W. B. 1991. Facilitation ofdimethylbenz[a]anthracene-induced rat mammary tumorigenesisby restraint stress: Role of beta-endorphin, prolactin andnaltrexone. Carcinogenesis 12:637-41.Temoshok, L.; Peeke, H. V. S.; Mehard, C. W.; Axelsson, K.;Sweet, D. M. 1987. Stress-behavior interactions in hamstertumor growth. Annals of the New York Academy of Sciences 496:501-509.Teshima, H.; Kubo, S. 1984. Changes in immune response andtumor growth in mice depend on the duration of stress. In:Fox, B. H.; Newberry, B. H., eds. Impact ofPsychoneuroendocrine Systems on Cancer and Immunity. Toronto:C. J. Hogrefe, Inc.; 208-244.Thomas, C. B.; Duszynski, K. R. 1974. Closeness to parents andthe family constellation in a prospective group study of 5disease states: Suicide, mental illness, malignant tumor,hypertension, and coronary artery disease. Johns Hopkins Medical Journal 134:251-270.Turney, T. H.; Harmsen, A. G.; Jarpe, M. A. 1986. Modificationof the antitumor action of Cornebacterium parvum by stress.Physiology and Behavior 37:555-558.Valzelli, L. 1969. The exploratory behaviour in normal andaggressive mice. Psychopharmacologia (Berlin) 15:232-235.Valzelli, L.^1973.^The "isolation syndrome" in mice.Psychopharmacologia (Berlin) 31:305-320.136Visintainer, M. A.; Volpicelli, J. R.; Seligman, M. E. P. 1982.Tumor rejection in rats after escapable or inescapable shock.Science 216:437-439.Wanatabe, S.; Nohno, T.; Omukai, Y.; Saito, T.; Senoo, T. 1982.Stimulatory effects of dexamethasone and indomethacin ongrowth of androgen-dependent Shionogi carcinoma 115 in themouse. Cancer Letters 16:261-266.Weinberg, J.; Emerman, J. T.^1989. Effects of psychosocialstressors on mouse mammary tumor growth. Brain, Behavior, and Immunity 3:234-246.Weiss, J. M.^1970.^Somatic effects of predictable andunpredictable shock. Psychosomatic Medicine 32:397-408.Weiss, J. M.; Pohoreckey, L. A.; Salman, S.; Gruenthal, M.1976. Attenuation of gastric lesions by psychological aspectsof aggression in rats. Journal of ComDarative Physiological Psychology 90:252-259.Weiss, J. M.; Goodman, P. A.; Losito, B. G.; Corrigan, S.Charry, J. M.; Bailey, W. H. 1981. Behavioral depressionproduced by an uncontrollable stressor: Relationship tonorepinephrine, dopamine and serotonin levels in variousregions of rat brain. Brain Research Reviews 3:167-205.Wirsching, M.; Stierlin, H.; Hoffmann, F.; Weber, G.; Wirsching,B. 1982. Psychological identification of breast cancerpatients before biopsy. Journal of Psychosomatic Research26:1-10.Workman, E. A.; LaVia, M. F.^1987.^T-lymphocyte polyclonalproliferation: Effects of stress and stress response styleon medical students taking national board examinations.Clinical Immunology and Immunopathology 43:308-313.


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items