RESEARCH ARTICLE Open AccessLong-term mortality among women withepithelial ovarian cancer: a population-based study in British Columbia, CanadaNimisha Arora1, Aline Talhouk1, Jessica N. McAlpine1, Michael R. Law2 and Gillian E. Hanley1,3*AbstractObjectives: Among women with epithelial ovarian cancer (EOC), histotype is one of the major prognostic factors.However, few data are available on histotype- specific survival and mortality estimates among these patients.We therefore examined survival and causes of death among women with EOC by histotype.Methods: A population- based cohort including all ovarian cancer patients diagnosed in British Columbia (BC)between 1990 and 2014 was built using population-based administrative datasets. We compared causes of deathwithin histotypes, by age at diagnosis, BRCA status, and time since diagnosis.Results: A total of 6975 women were identified as having been diagnosed with EOC between 1990 and 2014 in BC.The most common cause of death among these women was ovarian cancer until 10 years post diagnosis when othercauses surpassed ovarian cancer as the leading cause of death. Among women with serous EOCs, ovarian cancer wasthe leading cause of death 12 years after diagnosis, whereas ovarian cancer was the leading cause of death for 8 yearsamong women with non- serous EOCs. Among women with serous EOCs, ovarian cancer was the leading cause ofdeath for 12 years among younger women (< 60 years of age) compared to 8 years among women > = 60 years ofage, and those with BRCA mutations were more likely to die from ovarian cancer than those without a BRCA mutation.Conclusions: Within 10 years from diagnosis, ovarian cancer is the leading cause of death among women diagnosedwith EOC.Keywords: Ovarian Cancer, Histotype, Mortality, SurvivalIntroductionOvarian cancer is the leading cause of death amongwomen with gynecologic cancers. Although lifetime riskof ovarian cancer in the general population is relativelylow (1.4%) [1], it is the fifth leading cause of cancerdeaths among women in Canada, with a 5-year survivalrate of 44% [2] compared with nearly 90% [3]for breastcancer, more than 80% [2] for endometrial cancer, andnearly 73% [2] for cervical cancer. While survival ismuch improved when ovarian cancer is detected in theearly stages, there are presently no effective screeningmethods demonstrated to reduce mortality [4]. Ovariancancer is also largely asymptomatic in early stages,thus approximately 70% of women are diagnosedwhen the disease is already at advanced stages (StageIII and IV) [5].Causes of death among cancer patients has been rela-tively well studied across many different forms of cancer.It is often reported that within 5 years of a cancer diag-nosis, cancer is the most common cause of death.However, the risk of dying from cancer decreases withtime from diagnosis and cancer patients become morelikely to die from other causes [5]. This has been re-ported for breast cancer [6], prostate cancer [7], headand neck cancer [8], and lung cancer [9]. With respectto ovarian cancer, a previous American study usingSurveillance Epidemiology, and End Results (SEER) datahas reported that the probability of dying from ovariancancer decreases with time, but that ovarian cancer* Correspondence: Gillian.hanley@vch.ca1Department of Obstetrics & Gynecology, University of British Columbia,Vancouver, BC, Canada3Diamond Health Care Center, Vancouver General Hospital, 6207A 2775Laurel St., Vancouver, BC V5Z 1M9, CanadaFull list of author information is available at the end of the article© The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.Arora et al. BMC Cancer (2018) 18:1039 https://doi.org/10.1186/s12885-018-4970-9remains the leading cause of death 15 years post diagno-sis among women diagnosed in advanced stages [5].Here, we have focused on EOC, which represents 90%of all ovarian cancers [10]. Over the past decade, it hasbecome apparent that EOC is a heterogenous disease,comprising of distinct histotypes that differ in presenta-tion, response to therapy, molecular features, hereditarypredisposition, site of origin and clinical outcomes.Although EOC histotypes share an anatomical location(the ovary), they are now considered distinct diseases.[11–15]. This histotype-specific approach has completelychanged the approach to clinical care and research.Histotype and stage remain the strongest prognosticfactors in EOC. Some international work examiningsurvival among ovarian cancer patients by histologic-subtypes has reported that survival rates were lowestamong women with high-grade serous cancers [13].However, little is known about causes of death amongwomen with EOC, and we do not currently have evi-dence on whether causes of death differ among theseaffected women by histotype.Histotype-specific mortality estimates are of clinicalimportance as information may impact advice given orinterventions undertaken for patients and physicians in-volved in their care. For women with OC histotypes lesslikely to recur, information is currently lacking on theirpotential health concerns following EOC diagnosis Wetherefore examined survival and causes of death amongwomen with EOC in British Columbia between 1990and 2014 by histotype and years since EOC diagnosis.MethodsData sourcesIn this descriptive study, we built a population-based co-hort using data from the BC cancer registry [16], Vitalstatistics [17], BC hereditary cancer program (HCP) [16],and BC’s insurance registry file (the Consolidation file)[18]. The BC Cancer Registry is a population-basedregistry of all cancers diagnosed in BC residents. It re-ceives notifications of cancer from many sources includ-ing pathology, cytology and other labs, hospital charts,death certificates, and admissions to cancer centers op-erated by the BC Cancer Agency. The Registry containspersonal and demographic information and informationabout the specific cancer diagnosis. The vital statisticsdeath file is an extract of the deaths registration file pro-vided by BC vital statistics agency. It contains informa-tion on all deaths in BC, including underlying cause ofdeath (UCOD) and exact date of death. We accesseddata from the HCP, the source of all BRCA1 and BRCA2testing in the province of BC. We classified patients ashaving a BRCA mutation if they had either a BRCA1 orBRCA2 mutation or both. The consolidation file is a com-prehensive data set, containing information on individualsreceiving health services and/or individuals eligible to re-ceive health services in BC, Canada (˜4.6 million people).We used the consolidation data to access demographicdata and information on registration for health insurancein BC, in order to assess whether a woman had movedout of BC [19]. As the data linkage maintained patientanonymity (all identifiers were removed before being pro-vided to the researchers) and the population-based admin-istrative datasets in BC operate based on passive consent(i.e. patients may withdraw their consent and their datawill be removed from the administrative datasets), directpatient consent was not required. With the permission ofall relevant data stewards, and ethics approval from theUniversity of British Columbia’s behavioural research eth-ics board, data were retrieved from PopData BC [20]. Allinferences, opinions, and conclusions drawn are those ofthe authors and do not reflect the opinions or policies ofthe Data Stewards.Study cohortOur study population consists of all patients diagnosedwith ovarian cancer in BC between 1990 and 2014. TheInternational Statistical Classification of Diseases codes:tenth revision (ICD-10) was used to identify thesewomen with ovarian cancer with the codes of C56.0(ovary), and C57. 0 (fallopian tube) in the BC cancerregistry. Following WHO criteria (histotype classifica-tion), the study cohort was restricted to patients diag-nosed with serous, mucinous, endometrioid, and clearcell tumours. The codes are unable to distinguish be-tween high grade and low-grade serous cancers, andthus we have classified them as serous cancers. Thereare also ovarian cancers with morphology codes that arenot detailed enough to classify into histologic subtypesand were included as “Not classified” histologic subtype.Women with these unclassified tumours are analyzedseparately in all histotype-specific analyses. To ensurewe had complete follow-up on women included in ourcohort and to prevent misclassifying women as alive ifthey had left the province, we first assessed whetherwomen were registered for health care in BC in the finalyear of follow-up (2014). If women were not regis-tered in 2014 and were not captured in the death file,we required that they be registered for > 5 years fol-lowing their EOC diagnosis in order to be included inour study.Assessment of causes of deathWe classified the UCODs using ICD-10 categories. TheUCODs were classified into specific categories such asovarian cancer, breast cancer, colorectal cancer, ‘other’cancers (lung cancer, gastrointestinal tract cancer, bloodlymph cancer, other malignancy, non malignant and un-specified), cardiovascular diseases (rheumatic, hypertension,Arora et al. BMC Cancer (2018) 18:1039 Page 2 of 9ischemic, heart failure, congenital, pulmonary, cardiomyop-athy etc), other chronic conditions (diabetes, COPD, AIDS/HIV, pneumonia, other infectious and parasitic disease,asthma, cerebro and other vascular disease, liver disease,pulmonary fibrosis etc), external causes (Motor vehicle ac-cidents, poisoning, falls, suicide, other unintentional injuriesetc), and unclassified causes (causes of death that did notmeet the criterion of the above categories) (see Table 1).Patients were considered to have died of ovarian cancer ifthe cause of death was reported as ovarian cancer orcancer-related likely due to ovarian cancer (which includeddeaths from neoplasm of uterus, cervix, placenta, ovary andadnexa, vagina and external genitalia following an initialdiagnosis of ovarian cancer).Statistical analysisWomen were monitored as of the date of their ovariancancer diagnosis (as recorded in the BC cancer registrydatabase) until their death or until December 31st, 2014(the end of the follow-up period). Causes of death werestratified based on histotype categories (Serous, Mucinous,Endometrioid, Clear cell, Not classified). We further strati-fied based on age at diagnosis (< 60 or > =60 years) andBRCA mutation status. Causes of death were calculated aspercentages with 95% confidence intervals. 95% confi-dence intervals were calculated using Mid- P exact test.All analyses were performed with R version 3.3.2 [21].ResultsA total of 6975 women were identified as having beendiagnosed with ovarian cancer between 1990 and 2014in BC. After excluding women who did not have EOC(n = 407), and after excluding women who were not cap-tured in the death registry and were not registered forhealth care in BC in 2014 or for at least 5 years postdiagnosis (n = 141), 6427 were included in our study.The study cohort included 2996 (46.6%) serous, 366(5.7%) mucinous, 719 (11.2%) endometrioid, 431 (6.7%)clear cell and 1915 (29.8%) not classified EOCs.Table 1 Underlying causes of death and their ICD-10 codesCauses of Death ICD-10 codesOvarian cancer or ovarian cancer related C510-C58Breast cancer C500-C509Colorectal cancer C180-C218Other cancer Lung: C33, C340-C349, C384, C450;GI: C150-C179, C220-C269;Blood lymph: C810-C969, C463;Other malignancy: C000-C148, C300-C449, C451-C462, C467-C499, -C609, C620-C768, C5091, C80;Non malignant and unspecified: D000-D489Cardiovascular Rheumatic/Valvular: I050-I099, I340-I38;Hypertension: I10-I159; Ischemic: I200-I259;Conductive & Dysrhythmic: I440-I499; heart failure: I500-I509;Congenital: Q200-Q249; Pulmonary: I260-I289;Cardiomyopathy: I420-I429; Unspecified: I312-I318, I510-I513, I515-I519Other chronic Diabetes: E100-E149;COPD: J440-J449AIDS/HIV: B200-B24;Pneumonia: J120-J181, J188-J189Other infectious and parasitic disease: A000-B199, B250-B999, U049;Asthma: J450-J459, J46;Cerebro and other vascular disease: I600-I698, I700-I879, I950-I959, I880-I899;Liver disease:K700-K7699;Pulmonary fibrosis: J841;ALS/MS: G122, G1221, G35;Lung disease due to external agents: J60-J709External causes of death Motor vehicle accidents (MVA): V020-V049, V090-V099, V120-V149,V190-V196, V200-V799, V803-V805, V820-V821, V823-V839, V840-V878,V880-V888, V8900-V8909,V8920-V8929, V8990-V8999, Y850Poisoning: X40-X49Falls: W00-W19Suicide: X60-X84, Y870Other external:Y10-Y369, Y890-Y899Other unintentional injuries:, V010, V019, V050-V069, V091, V099, V100-V119, V150-V189, V198-V199,V250-V259, V350-V359, V450-V459, V550-V559, V650-V659, V750-V759, V800-V802, V806-V819, V822,V879, V889, V910-V919, V930-V949, V950-V978, V98-V99, W20-W64, W75-W99, X20-X39, X50-X59,Y40-Y849, Y859, Y86, Y880-Y883Unclassified cause of death Codes that did not meet criterion aboveArora et al. BMC Cancer (2018) 18:1039 Page 3 of 9The clinical characteristics of BC women diagnosedwith EOC between 1990 and 2014 are outlined inTable 2. Among all histotypes, serous carcinomas werecommonly observed (n = 2996), accounting for approxi-mately 46.6% of total EOCs (66.4% if we remove EOCsthat were not classified by histotype). The majority(68.2%, n = 4382) of affected women were diagnosedwith EOC between 50 and 79 years of age and were notBRCA mutation carriers (96.9%, n = 6228).Median follow- up was 3.5 years (IQR: 3.4–3.5 years)for women with serous EOCs, 9 years (IQR: 8–10 years)for women with mucinous EOCs, 7.5 years (IQR: 7–8.5 years) for women with endometrioid EOCs,7 years (IQR: 6.5–8 years) for women with clear cellEOCs, and 4 years (IQR: 3.5–4 years) for womenwith not-classified EOCs.By the end of this study, 55.9% of the study cohort (allhistologies) died from their ovarian cancer (n = 3592),33.9% were alive (n = 2181), 10.2% of women died fromcauses other than ovarian cancer (n = 654); 0.8% frombreast cancer (n = 49), 0.5% from colorectal cancer (n =34), 3.5% from ‘other’ cancers (n = 227), 1.8% from car-diovascular disease (n = 116), 1.8% from other chronicconditions (n = 114), 0.3% from external causes (n = 20)and 1.5% from unclassified causes (n = 94). Amongwomen with serous EOCs, 29.3% (n = 877) were alive atthe end of follow-up while 62.2% had died from theirdisease (n = 1864). 54.4% women with mucinous EOCswere alive (n = 199) while 32% died from ovarian cancer(n = 117). The greatest number of women alive at theend of follow-up was observed among women withendometrioid EOC (62.0%, n = 446). Among thesewomen who had died, 26.7% died from ovarian cancer(n = 192). There was a considerable number of womenwith clear cell EOCs who were alive at the end offollow-up (55.9%, n = 241). However, 35.3% of womenwith clear cell EOCs died from ovarian cancer (n = 152).Among women with EOCs that could not be classifiedinto histotypes, 21.8% of cases were alive at the end offollow-up (n = 418) while 66.2% died from ovarian can-cer (n = 1267). Within each histotype, the most commoncause of death after ovarian cancer was death from othercancers (Additional file 1: Table S3).Figure 1a displays the frequency distribution of deathsfor all histotypes. It reveals that ovarian cancer is theleading cause of death among women diagnosed withovarian cancer for 10 years post diagnosis. It is first sur-passed by other causes of deaths 11 years post diagnosis.Figure 1b displays the frequency distribution of deathsfor serous EOCs. Ovarian cancer remains the leadingcause of death among women diagnosed with serousEOCs for 12 years following diagnosis. Figure 1c depictsthe frequency distribution of deaths for non- serous(endometrioid, clear cell, mucinous) EOCs. Other causesof death surpass ovarian cancer as the leading causeamong women diagnosed with non-serous EOCs at8 years post diagnosis. Figure 2a and b display the fre-quency distribution of deaths for serous EOCs by agegroup. Figure 2c and d displays the frequency distribu-tion of deaths for non-serous EOCs by age group. Ovar-ian cancer remains the leading cause of death for longeramong younger women (< 60 years) than among olderwomen (60 years or more). Other causes of death firstsurpass ovarian cancer as the leading cause of deathamong older women with serous EOCs at 8 years postdiagnosis. Whereas ovarian cancer is the leading causeof death among younger women with serous EOCs for12 years post diagnosis. Among women with non- ser-ous EOCs, ovarian cancer is the leading cause of deathamong older women for 5 years after diagnosis in com-parison to younger women where it is 8 years after diag-nosis. There were too few deaths from causes other thanovarian cancer in the women with a BRCA mutation toreliably examine causes of death over time in this group.Differences between age groups in the causes of death,stratified by histotype, are reported in Additional file 2:Table S4. In each histotype group, a greater number ofwomen who were diagnosed under 60 years of age sur-vived than older women. Women diagnosed with serouscancer at 60 or older were most likely to die fromTable 2 Clinical characteristics of the study cohortYear of diagnosis N %1990–1994 1106 17.21995–1999 1265 19.72000–2004 1252 19.52005–2009 1395 21.72010–2014 1409 21.9Histology N %Serous 2996 46.6Mucinous 366 5.7Endometrioid 719 11.2Clear cell 431 6.7Not classified 1915 29.8Age at diagnosis N %< 40 410 6.440–49 815 12.750–59 1477 23.060–69 1457 22.770–79 1448 22.580+ 820 12.8BRCA status N %BRCA null 199 3.1BRCA wild type 6228 96.9Arora et al. BMC Cancer (2018) 18:1039 Page 4 of 9ovarian cancer (66.8%, n = 1231), whereas women di-agnosed with endometrioid cancer under the age of60 were least likely to die from ovarian cancer(19.5%, n = 87) followed closely by women diagnosedwith mucinous cancer under the age of 60 (21.3%, n= 46). For each age cohort, and for each histotypeovarian cancer was the leading cause of death. In all,death from ovarian cancer was most common, followedby ‘other’ cancers.The outcomes of women with ovarian cancer and aBRCA mutation, stratified by serous or non- serous his-totype is reported in Additional file 3: Table S5. We wereunable to stratify by other histotypes due to small sam-ple sizes. Among women diagnosed with serous cancer,Fig. 1 a: Frequency distribution of deaths among patients diagnosed with all the histotypes. b: Frequency distribution of deaths among patientsdiagnosed with serous epithelial ovarian cancers. c: Frequency distribution of deaths among patients diagnosed with non- serous epithelial ovarian cancersArora et al. BMC Cancer (2018) 18:1039 Page 5 of 9Fig. 2 (See legend on next page.)Arora et al. BMC Cancer (2018) 18:1039 Page 6 of 9those with a BRCA mutation (BRCA positive) facegreater risk of death from breast cancer (4%, n = 6) incomparison to women without a mutation (0.7%, n =20). In all, regardless of a woman’s BRCA mutation sta-tus, and for all histotypes ovarian cancer was the leadingcause of death.DiscussionHerein we report that among EOC patients, ovarian can-cer is the leading cause of death for 10 years post diag-nosis before it is surpassed by other causes of death.However, this differed significantly based on histotypeand age at diagnosis. For example, among women withserous EOC, ovarian cancer was the leading cause ofdeath for 12 years compared to 8 years among womenwith non- serous EOCs. When stratified by age, ovariancancer was the leading cause of death for 8 years postdiagnosis among women with serous EOCs diagnosedafter age 60 compared to 12 years among youngerwomen (< 60 years of age) with serous EOCs. Our re-sults suggest that women with non- serous EOCs weremore likely to die from causes other than ovarian cancerin comparison to serous patients, as the majority ofwomen with serous EOCs died from ovarian cancer. Forinstance, women with endometrioid and mucinousEOCs were more vulnerable to die from cardiovasculardiseases, chronic conditions and unclassified causes thanwomen with serous EOC.Although the percentages of death from breast cancerwere not significantly different across the histotypes, wefound that these percentages varied when stratified byBRCA status. Women with a BRCA mutation and withserous carcinomas were less likely to die from othercauses, and more likely to die from ovarian cancer, thanwomen without a BRCA mutation. Whereas womenwith a BRCA mutation and a non- serous cancerwere at relatively comparable risk of death from othercauses as women without a BRCA mutation and anon- serous cancer.Our results reporting a significant risk of death fromovarian cancer for many years post diagnosis is consist-ent with those previously reported using SEER data [5].The SEER study reported that the probability of deathfrom ovarian cancer decreases with increased survivalyears post diagnosis and the probability of death from allother causes increases. We observed that ovarian can-cer is the leading cause of death among BC womendiagnosed with EOC for 10 years post diagnosis.The SEER study reported that ovarian cancer wassurpassed by other causes of death 7 years afterdiagnosis.Our results are also consistent with a small body of lit-erature reporting minimal risk of breast cancer amongwomen with EOC and a BRCA mutation [22]. Our find-ings indicate a low incidence of death from breast canceramong women with ovarian cancer and a BRCA muta-tion. When stratified by histotype, BRCA mutation car-riers diagnosed with serous EOCs were at the greaterrisk of death from breast cancer, breast cancer was onlyresponsible for 4% of deaths among these women. Thissupports the assertion that there is no need to rush toperform mastectomy to prevent breast cancer amongwomen with a BRCA mutation and EOC. Other researchhas suggested that there are greater improvements insurvival with mastectomy among women who hadalready survived 10 years from diagnosis with EOC andthose with stage I or II ovarian cancer. Mastectomy isalso often recommended for ovarian cancer patientswho are under age 55 and those with serum CA125levels within normal limits [23, 24].The population-based nature of the study and its in-clusion of all women diagnosed with EOC in BC be-tween 1990 and 2014 is an important strength of ourresearch; however, some limitations are noted. Our reli-ance on the ICD morphologic codes to classify tumoursinto histologic subtypes likely introduced some mis-classification. Although there have been considerable ad-vances in categorization of epithelial ovarian cancersubtypes with high interobserver agreement in histotypeassignment for this disease [25], our study includes can-cers subtyped prior to these publications, and most didnot have the benefit of additional immunohistochemicaltests to help characterize challenging cases. Thus, we ex-pect that there has been some misclassification of histo-types in this study. In addition, the rarity of ovariancancer combined with the relatively small BC populationhas resulted in small numbers of women with histotypesother than serous ovarian cancer, and thus we had togroup these histotypes in some analyses. Also, we cannotcomment on the lynch syndrome status of our studypopulation as we lack these data. Based on the past lit-erature, women with endometroid and clear cell EOCswere observed to be at risk of Lynch syndrome [26].Associated with a high risk of colorectal cancer, lynchsyndrome status may have factored into the risk of deathfrom colorectal cancer.(See figure on previous page.)Fig. 2 a: Frequency distribution of deaths among older patients (60 years or more) diagnosed with serous epithelial ovarian cancers. b: Frequencydistribution of deaths among younger patients (under 60 years) diagnosed with serous epithelial ovarian cancers. c: Frequency distribution of deathsamong older patients (60 years or more) diagnosed with non- serous epithelial ovarian cancers. d: Frequency distribution of deaths among youngerpatients (under 60 years) diagnosed with non- serous epithelial ovarian cancersArora et al. BMC Cancer (2018) 18:1039 Page 7 of 9Importantly, we were unable to examine long-termmortality by stage because these data were missing inour cancer registry. Previous research has shown thatwomen diagnosed with early stage EOC are less likely todie from ovarian cancer [5], but this has never been ex-amined by histotype. Future research should examinewhether stage influences long-term mortality differentlyby histotype. We are also missing data on importanttreatment variables, including the receipt of neoadjuvantor adjuvant chemotherapy, both intravenous and intra-peritoneal, and the residual disease status of the studypopulation following surgery. A better understanding ofthe effect of treatment by histotype would be useful forproviding a more nuanced picture of what influencessurvival. Finally, our percentage of ovarian cancer patientswith BRCA mutations is much lower than expected, giventhat recent research indicates approximately 20% ofHGSC patients have a BRCA mutation [27]. This reflectsthe fact that histotype-based referral did not comeinto effect in British Columbia until 2010, and thateven following this recommendation, only 46% ofHGSC patients were being tested for BRCA mutations[28]. The high number of deaths from unclassifiedcauses is unexpected and may reflect inaccuracies ormissing information in the death certificates for someof those women. To limit this possible bias, our mainanalysis compared ovarian cancer specific mortality toall other causes of death.The results of this study have implications for clini-cians and ovarian cancer patients. The findings can helpclinicians better understand the differences in outcomesamong women with ovarian cancer based on histotype,age at diagnosis and BRCA status.ConclusionsOur data show that ovarian cancer remains the lead-ing cause of death among women with EOC for10 years post diagnosis. However, the probability ofovarian- cancer related death varies significantly basedon histotype, age at diagnosis and BRCA mutationstatus. The differences in proportions of deaths be-tween the histological groups support the evidence ofovarian cancer heterogeneity and support the asser-tion that histotype-specific research is valuable whenresearching ovarian cancer.Additional filesAdditional file 1: Table S3. Cause of death stratified by histotype.(DOCX 18 kb)Additional file 2: Table S4. Cause of death stratified by histotype andage at diagnosis. (DOCX 19 kb)Additional file 3: Table S5. Cause of death stratified by histotype andBRCA status. (DOCX 16 kb)AbbreviationsBC: British Columbia; EOC: Epithelial ovarian cancer; HCP: BC hereditarycancer program; ICD-10: International Statistical Classification of Diseasescodes: tenth revision (ICD-10); SEER: Surveillance Epidemiology, and EndResults; UCOD: Underlying cause of deathAcknowledgmentsNot applicable.FundingCanadian Cancer Society Research Institute, Canadian Institutes for HealthResearch, Vancouver General Hospital, University of British Columbia HospitalFoundation.Availability of data and materialsThe data used herein are available from Population Data BC, the BC CancerAgency, and the BC Hereditary Cancer Program but restrictions apply to theavailability of these data, which were used under a data sharing agreementfor the current study, and so are not publicly available. Data are howeveravailable from Population Data BC, the BC Cancer Agency, and the BCHereditary Cancer Program following completion of a data access request andwith permission of the data stewards.Authors’ contributionsThe paper is authored by Nimisha Arora (NA), Aline Talhouk (AT), Jessica N.McAlpine (JM), Michael R. Law (ML), Gillian E. Hanley (GH). NA and GH cameup with the study proposal and wrote the manuscript. NA performed thedata analysis, thereby prepared required figures and Tables. GH assisted NAin the final analysis and interpretation of the data. AT, JM, ML contributed tomanuscript revision. The final draft of manuscript was read and approved byall the authors.Ethics approval and consent to participateThe University of British Columbia’s behavioural research ethics board hasapproved the study. As the data linkage maintained patient anonymity(all identifiers were removed before being provided to the researchers)and the population-based administrative datasets in BC operate basedon passive consent (i.e. patients may withdraw their consent and theirdata will be removed from the administrative datasets), direct patientconsent was not required.Consent for publicationNot applicable.Competing interestsThe authors declare that they have no competing interests.Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.Author details1Department of Obstetrics & Gynecology, University of British Columbia,Vancouver, BC, Canada. 2Centre for Health Services and Policy Research,School of Population and Public Health, University of British Columbia,Vancouver, BC, Canada. 3Diamond Health Care Center, Vancouver GeneralHospital, 6207A 2775 Laurel St., Vancouver, BC V5Z 1M9, Canada.Received: 30 May 2018 Accepted: 17 October 2018References1. Canadian Cancer Society's Advisory Committee on Cancer Statistics.Canadian Cancer Statistics 2014 - Special Topic: Skin Cancers. CanadianCancer Society; 2014. http://www.cancer.ca/. Accessed 20 May 2017.2. Canadian Cancer Society's Advisory Committee on Cancer Statistics.Canadian Cancer Statistics 2016 - Special topic: HPV-associated cancers; 2016.http://www.cancer.ca/. Accessed 20 May 2017.3. Nuttall R. Canadian cancer statistics 2017. Canadian Cancer Society; 2017.http://www.cancer.ca/. Accessed 20 July 2017.Arora et al. BMC Cancer (2018) 18:1039 Page 8 of 94. Gérvas J. Ovarian cancer screening: could you recommend it? No. EBM2016; doi: https://doi.org/10.1136/ebmed-2016-110385.5. Dinkelspiel HE, Champer M, Hou J, et al. Long-term mortality amongwomen with epithelial ovarian cancer. Gynecol Oncol. 2015. https://doi.org/10.1016/j.ygyno.2015.06.005.6. Colzani E, Liljegren A, Johansson ALV, et al. Prognosis of patients withbreast cancer: causes of death and effects of time since diagnosis, age, andtumor characteristics. JOC 2011; doi: https://doi.org/10.1200/JCO.2010.32.6462.7. Van Hemelrijck M, Folkvaljon Y, Adolfsson J, et al. Causes of death in menwith localized prostate cancer: a nationwide, population-based study. BJUInt. 2016. https://doi.org/10.1111/bju.13059.8. Massa ST, Osazuwa-Peters N, Christopher KM, et al. Competing causes ofdeath in the head and neck cancer population. Oral Oncol. 2017. https://doi.org/10.1016/j.oraloncology.2016.12.006.9. Janssen-Heijnen MLG, van Erning FN, De Ruysscher DK, Coebergh JWW,Groen HJM. Variation in causes of death in patients with non-small cell lungcancer according to stage and time since diagnosis. Ann Oncol. 2015.https://doi.org/10.1093/annonc/mdv061.10. Webb PM, Jordan SJ. Epidemiology of epithelial ovarian cancer. Best Pract ResClin Obstet Gynaecol. 2017. https://doi.org/10.1016/j.bpobgyn.2016.08.006.11. Kommoss S, Gilks CB, du Bois A, Kommoss F. Ovarian carcinoma diagnosis:the clinical impact of 15 years of change. Br J Cancer. 2016. https://doi.org/10.1038/bjc.2016.273.12. Matz M, Coleman MP, Sant M, et al. The histology of ovarian cancer: worldwidedistribution and implications for international survival comparisons (CONCORD-2).Gynecol Oncol. 2017. https://doi.org/10.1016/j.ygyno.2016.10.019.13. Matz M, Coleman MP, Carreira H, Salmerón D, Chirlaque MD, Allemani C.Worldwide comparison of ovarian cancer survival: histological group andstage at diagnosis (CONCORD-2). Gynecol Oncol. 2017. https://doi.org/10.1016/j.ygyno.2016.11.019.14. Gilks CB, Ionescu DN, Kalloger SE, et al. Tumor cell type can be reproduciblydiagnosed and is of independent prognostic significance in patients withmaximally debulked ovarian carcinoma. Hum Pathol. 2008. https://doi.org/10.1016/j.humpath.2008.01.003.15. Köbel M, Kalloger SE, Boyd N, et al. Ovarian carcinoma subtypes aredifferent diseases: implications for biomarker studies. PLoS Med. 2008.https://doi.org/10.1371/journal.pmed.0050232.16. BC cancer agency registry data (2011). Population data BC. BC cancer agency.2011 http://www.popdata.bc.ca/data. Accessed 20 May 2017.17. BC Vital Statistics Agency (2011). Vital statistics deaths. Population data BCBC vital statistics agency (2011). http://www.popdata.bc.ca/data. Accessed20 May 2017.18. British Columbia Ministry of Heath (2011): Consolidation file (MSPregistration and premium billing). Population data BC MOH (2011).http://www.popdata.bc.ca/data. Accessed 20 May 2017.19. Hanley GE. Equity in the use and financing of medicines in British Columbiabefore and after the move to income-based pharma care. 2011. https://open.library.ubc.ca. Accessed 10 December 2017.20. About popdata. https://www.popdata.bc.ca/. Accessed 16 April 2017.21. RStudio team (2016). RStudio: Integrated development for R. RStudio, inc.,boston, MA URL http://www.rstudio.com/. Accessed 17 May 2017.22. Gangi A, Cass I, Paik D, et al. Breast cancer following ovarian cancerinBRCAMutation carriers. JAMA Surg. 2014. https://doi.org/10.1001/jamasurg.2014.1081.23. McGee J, Giannakeas V, Karlan B, et al. Risk of breast cancer after adiagnosis of ovarian cancer in BRCA mutation carriers: is preventivemastectomy warranted? Gynecol Oncol. 2017. https://doi.org/10.1016/j.ygyno.2017.02.032.24. Kotsopoulos J, Narod SA. Prophylactic mastectomy for BRCA mutation carriersafter ovarian cancer treatment: is it beneficial? Expert Rev Anticancer Ther.2018. https://doi.org/10.1080/14737140.2018.1424547.25. Köbel M, Bak J, Bertelsen BI, et al. Ovarian carcinoma histotype determination ishighly reproducible, and is improved through the use of immunohistochemistry.Histopathology. 2014. https://doi.org/10.1111/his.12349.26. Vierkoetter KR, Ayabe AR, VanDrunen M, Ahn HJ, Shimizu DM, TeradaKY. Lynch syndrome in patients with clear cell and endometrioidcancers of the ovary. Gynecol Oncol. 2014. https://doi.org/10.1016/j.ygyno.2014.07.100.27. Schrader KA, Hurlburt J, Kalloger SE, et al. Germline BRCA1 and BRCA2mutations in ovarian cancer: utility of a histology-based referral strategy.Obstet Gynecol. https://doi.org/10.1097/AOG.0b013e31825f3576.28. Hanley GE, McAlpine JN, Miller D, et al. A population-based analysis ofgermline BRCA1 and BRCA2 testing among ovarian cancer patients in anera of histotype-specific approaches to ovarian cancer prevention. BMCCancer. 2018. https://doi.org/10.1186/s12885-018-4153-8.Arora et al. BMC Cancer (2018) 18:1039 Page 9 of 9