Stage at Diagnosis and Comorbidity Influence Breast Cancer Survival in First Nations Women in Ontario, Canada

This article is featured in Highlights of This Issue, p. 1991

Canada's indigenous populations comprise the Inuit, Métis, and First Nations (FN), who collectively represent 3.8% of the national population (1). Ontario has the largest indigenous population of approximately 243,000, although this represents a small proportion of the total provincial population (2%) (1). FN peoples are further categorized according to their relationship with the federal government under the Indian Act, and the indigenous Ontario population of “Registered Indians” was approximately 70% (2). This study is restricted to FN women who are “Registered Indians,” hereafter referred to simply as “FN.” FN people experience more health disparities than other Canadians (1). Factors embedded in the social determinants of health, including historical and economic contexts, community infrastructure and resources, and health behaviors, along with challenging physical and social environments, all contribute to the experienced inequalities (3).

Breast cancer has the highest incidence of cancers among FN (4) and non-FN (5) women in Ontario. Only limited data on cancer survival of FN populations in Canada are available. When examined, poorer survival for FN than non-FN populations is found for all cancers combined and for breast cancer specifically (6, 7). Outside of Canada, literature on breast cancer prognosis comparing indigenous to nonindigenous populations within the same geographic area is emerging. These studies have found poorer all-cause (8–11), cancer-specific (8, 12, 13), and/or breast cancer–specific survival (10, 14–18) among their indigenous populations with breast cancer. Where further investigations have been conducted for breast cancer, poorer stage- (10–12, 16, 18) and treatment- (10, 11, 18) adjusted survival have been observed. These analyses suggest that other factors may be contributing to the observed poorer breast cancer survival among indigenous populations.

This is the first study to examine a wide range of demographic, personal, tumor, and treatment factors potentially associated with survival after a breast cancer diagnosis among an indigenous population. Also, it is likely that the same or similar factors are largely responsible for the survival disadvantage observed for other types of cancer in indigenous populations in Ontario and elsewhere. The purposes of this study were to compare stage-specific survival for Ontario FN and non-FN women diagnosed with breast cancer between 1995 and 2004, controlling for demographic, personal, tumor, and treatment factors, and to examine potential determinants of survival for FN women according to stage at diagnosis.

The Ontario Cancer Registry (OCR) is a population-based cancer registry that includes all cancers diagnosed in Ontario residents (19). As ethnicity is not recorded in the OCR, it is not possible to routinely estimate cancer incidence, mortality, or survival rates for any specific ethnic group directly from OCR data. Therefore, a cohort of Ontario FN people included in the files of Indian and Northern Affairs Canada as “Registered Indians” at any time between 1968 and 1991 was created (4). The cohort was then linked to both the OCR (through 2004) and the Ontario mortality database (through 2007) to include information on all causes of death.

Subjects for this study were identified from the OCR as those seen at one of Ontario's specialized cancer centers. The Regional Cancer Centers (RCC) deliver all cancer radiotherapy in the province and are often involved in the diagnostic work-up and treatment planning. Chemotherapy is not restricted to the RCCs.

This study used a case–case design, comparing FN women diagnosed with invasive breast cancer (International Classification of Diseases, 9th Revision, code 174) that were pathologically confirmed in 1995–2004 at ages 15+ and seen at an Ontario RCC, with a frequency-matched random sample of non-FN women with breast cancer (20). All FN women satisfying study eligibility criteria were included. The general population sample was frequency matched (2:1) on age at diagnosis (<50 vs. ≥50), period of diagnosis (1995–1999, 2000–2004), and RCC location first attended. Because more than 90% of eligible women of both groups of interest were seen at an RCC, we did not expect to introduce any bias by restricting our study to those women.

Two trained chart abstractors reviewed the medical charts at the RCCs. All data elements were collected manually with the use of an abstraction manual. A pilot test revealed 94% accuracy between the 2 abstractors prestudy.

Demographic, personal, tumor, and treatment factors were abstracted from the medical charts at the RCCs either as potential confounders of the association between FN status and survival or as independent predictors of survival. Age at diagnosis, period of diagnosis, and RCC attendance were derived from the OCR data. The demographic factor that captured distance to an RCC was computed by calculating the straight line distance (to the nearest kilometer) between an RCC and a woman's residence (postal code) at diagnosis (21) and then categorizing the variable for “close,” “moderate,” and “far” with the cutoff points defined by the distance tertiles of FN women. Taking road routes, which are longer than a straight line route, into account did not change which distance category the women were situated in. The 2 northern RCCs are located in central Ontario, whereas the remaining southern locations are far south in the province. Therefore, the variable was defined as [close (northern women: 0 < 118 km, southern women: 0 < 30 km); moderate (northern women: 118 < 267 km, southern women: 30 < 81 km); and far (northern women: 267+ km; southern women: 81+ km)]. The personal factors included the following: body mass index (BMI), smoking status, comorbidity, and family history. These data were derived from a structured questionnaire administered by an oncology nurse at the women's first visit at the RCC. BMI was calculated as weight (in kg) divided by height (in m²) and was defined as (normal weight = BMI <25 kg/m²; overweight = BMI ≥25 to <30 kg/m²; obese = BMI ≥30 kg/m²) (22). Because of the risk potential of current and/or former active cigarette smoking on breast cancer survival (23), we grouped all the “ever smokers” together for the smoking status variable. Comorbidity was defined as having a concurrent health condition(s) at the time of diagnosis that was (were) likely to influence the probability of mortality, decrease the adherence to the treatment regime, or result in procedural or drug contraindication. The conditions included in this study were those specified in the Charlson index: acute myocardial infarction, cerebrovascular disease, chronic obstructive pulmonary disease or other respiratory diseases, congestive heart failure, dementia, diabetes, hemiplegia or paraplegia, HIV/AIDS, liver disease, peripheral vascular disease, renal disease, rheumatic-like diseases, and ulcers of the digestive system (24). This variable was then categorized as having no comorbidity, having diabetes, or having the other comorbid conditions. Having a “family history” included women with a first-degree relative with breast cancer and/or ovarian cancer.

The tumor-related factors included method of detection, stage at diagnosis, and estrogen and progesterone receptor statuses. Method of detection was grouped as “screen detected” for women whose breast cancer was detected through participation in routine mammography. “Nonscreen detected” included women whose breast cancer was detected by themselves, by a physician, or by other health care professional. We utilized the American Joint Committee on Cancer TNM classification scheme for staging breast cancers (25). The TNM system is an expression of the anatomic extent of disease and is based on 3 components: T, the extent of the primary tumor; N, the absence or presence and extent of regional lymph node involvement; and M, the absence or presence of distant metastasis. Stage at diagnosis was categorized as stage I, stage II, and stages III–IV. Hormonal receptor status of estrogen and progesterone receptors was defined as having “either positive” or “neither positive.” A representative variable for stage- and age-appropriate treatment using data that captured receipt of radiotherapy, chemotherapy, and surgeries (including whether a woman had a mastectomy or a lumpectomy plus radiotherapy) was created on the basis of guidelines developed by the Physician's Data Query (14). The data for receptor status were excluded from the calculation of this variable because hormonal therapy use was not consistent over the study time period.

The distributions of the demographic, personal, tumor, and treatment factors were compared between groups using χ² tests. Survival time was calculated for each woman, beginning at her date of diagnosis until death from any cause or censoring. Deaths not only from breast cancer but also from all causes were used in our analyses because data on fact of death were available up to the end of 2007 whereas the data on specific causes of death were only available up to December 31, 2005. Women were censored at the end of the study (December 31, 2007) if death had not occurred. Kaplan–Meier curves for FN versus non-FN women were compared by stage at diagnosis (stages I, II, and III–IV).

To examine the effect of FN status on risk of death, a separate Cox proportional hazards regression model including matching variables was created for each stage (base model). Potential confounders were added to the base model, and those significant at P < 0.05 were included in that model.

To examine the determinants of survival after a breast cancer diagnosis among FN women, Cox proportional hazards models were developed for each factor adjusted by age at diagnosis and period of diagnosis with stratification by stage. Because the non-FN women were matched to the characteristics of the FN women, factors associated with breast cancer survival were not assessed for those women because the results would not be generalizable to other non-FN women. All analyses were conducted using SAS (26), and all reported P values were for 2-sided alternatives.

Ethical approval of this research protocol was obtained from the Human Subjects Ethics Review Committee of the University of Toronto and from the RCCs where the women were treated for breast cancer.

Of the potential 1,052 cases, 958 women (287 FN and 671 non-FN) remained in the study (Fig. 1). The average age at diagnosis was similar in FN and non-FN women (Table 1). Only 30% of FN women lived close to an RCC compared with 51% among non-FN women. FN women had higher BMI (50% were obese compared with 29% among non-FN women), had ever smoked more often (68% vs. 49%), and were twice as likely to have one or more comorbid conditions (36% vs. 18%). In particular, the proportion of FN women with diabetes was more than 3 times that in the non-FN women (P ≤ 0.01 for each). FN women were diagnosed at later stages of breast cancer more often (P ≤ 0.05) than non-FN women. The treatment data revealed that FN women were significantly more likely to have received more intensive surgery (i.e., mastectomies compared with lumpectomies, or breast-conserving treatment plus radiotherapy) and significantly less likely to have had radiotherapy than did non-FN women (both P ≤ 0.01; data not shown). However, rates of stage- and age-appropriate treatment were similar between groups.

The only significant survival difference for FN women was for stage I breast cancer (P < 0.0001). Ninety-four percent of non-FN women survived 5 years after a stage I diagnosis compared with 86% of the FN women (Fig. 2). In the multivariate models adjusted for age at diagnosis, period of diagnosis, RCC, distance to RCC, BMI, smoking status, and comorbidity, the risk of death among FN women with a stage I breast cancer diagnosis was more than 3 times higher than that of non-FN women (HR = 3.10, 95% CI = 1.39–6.88; Table 2). FN women were not at an increased risk of death compared with non-FN women after a stage II or III–IV diagnosis.

Among FN women with a diagnosis of stage I breast cancer, only comorbidity significantly increased the risk of death. The risk of death in those with comorbidity other than diabetes and those with diabetes were about 5 times that of those without comorbidity (HR = 4.65, 95% CI = 1.39–15.53 and HR = 5.49, 95% CI = 1.69–17.90 respectively; Table 3). FN women diagnosed with stage III–IV breast cancer who also had diabetes had a significantly greater risk of death than those without any comorbidities (HR = 3.08, 95% CI = 1.26–7.50).

This study found that the survival disadvantage after a breast cancer diagnosis for FN women compared with non-FN women in Ontario was restricted to those diagnosed at stage I. Only a few studies examining breast cancer survival in an indigenous population have conducted separate analyses by stage at diagnosis (10, 16, 18). In these studies, contrary to our findings, the relative risk of death became increasingly worse with diagnosis at the later stage in the indigenous compared with nonindigenous populations across time periods in both the United States and Australia (10, 16, 18). These studies however did not adjust for distance to a treatment center, BMI, smoking status, or comorbidity, as was done in these analyses. The comparable survival for those diagnosed at the later stages of breast cancer for FN and non-FN women in our study is encouraging and suggests similar tumor biology and equivalent treatment of these stage II+ cancers.

Having a preexisting comorbidity was the most important prognostic factor among FN women. The variable used in these analyses was categorized as having a comorbidity (either diabetes or another comorbidity) from the Charlson index (24); however, 9% of FN women had 2 and 5% had 3 comorbid conditions compared with 4% and 1% among the non-FN women, respectively. Studies have shown that survival is poorer with accumulating comorbidities (24, 27) and thus may contribute to the survival difference. Another consequence of the presence of comorbidities is that physicians may be more reluctant to prescribe optimal stage or tumor-specific treatment because of potential complications (28).

Making payments to a health care facility or paying for treatment-related expenses would not be a barrier for Canadian FN women, as has been reported among Native American women (29), because these items are included in the provincial health care plan and/or by federally funded non-insured health benefits. However, other financial costs associated with treatment such as parking expenses at cancer centers may impose difficulties. Other barriers such as a lack of knowledge about breast cancer, its screening and treatment, or having to deal with multiple health care entities may certainly impact FN women in Ontario (29). Furthermore, traveling on average 160 km suggests that many FN women would have had to leave the comfort of their communities for diagnosis and subsequent treatment(s). The importance of community support and its relationship to thriving health, specific to Canadian indigenous women, has been shown in the literature (30). Our finding that FN women received significantly more mastectomies than lumpectomies and significantly less radiotherapy (which required a prolonged period of daily treatment) than non-FN women suggests that distance to a cancer center may impact treatment decisions; however, among these data, it may not necessarily impact stage- and age-appropriate treatment. It was unknown whether any of the study population had been offered brachytherapy, as only the treatment received data were collected from the medical charts. A course of brachytherapy would have facilitated a shorter treatment time.

This study had some limitations. We were unable to assess survival using breast cancer deaths separately from all causes of death. FN women were twice as likely to have 1 or more comorbid conditions compared with non-FN women; thus, it is possible that some of the early deaths among FN women were caused by consequences of those comorbidities. If rates of comorbidity had been comparable in the 2 groups of women, stage I breast cancer survival may not have been significantly different.

This study was designed to have the power to detect a significant difference between the stage I and II+ data. Because the stage II+ data were further stratified in the survival analyses, there may not have been an adequate sample size to detect significant effects separately, especially in the latter with only 58 FN women. Information about socioeconomic status would have been valuable but was not available at the RCCs. Furthermore, it has been reported that 44% of FN seniors (aged 55+) had used traditional medicine within the last year (31). Knowing more about traditional medicine use among the study sample is important in terms of the potential for nontraditional medicine compliance and/or for potential contraindications. Finally, although the therapeutic importance of Her2/neu receptor status is crucial, it was not included in the analyses because of the study time frame.

One of the strengths of this study is the accuracy of the indigenous status of the FN women because ethnicity was identified from a database linkage. However, it is possible that there may be other indigenous persons within the non-FN women group who did not meet the “Registered Indian” identity status. Upon review of the medical charts, we identified only one non-FN women as being indigenous; however, she was not an FN woman. Analyzing the data by stage at breast cancer diagnosis was also a strength of the study because it has been shown that detection of breast cancer occurs at a later stage (stage II+) more often among FN women than non-FN women in Ontario, Canada (32).

Future research is recommended on the stage I survival experience of FN women with and without diabetes, as well as other comorbidities that were observed in this study. The influence of diabetes, for instance, is important because some of the complications related to it such as renal impairment influence imaging modalities and may affect the dose of chemotherapy prescribed (8). Another research area that could explain some of the survival differential is wait times to treatment among FN women diagnosed at stage I, as women may have been presenting at a more advanced stage by the time they started their treatment regime (33).

In conclusion, this study determined that survival was more than 3 times poorer for FN women diagnosed at stage I breast cancer than non-FN women, even after adjustment for important cofactors such as comorbidity and BMI. The risk of death after a stage I breast cancer diagnosis was more than 5 times higher among FN women with a comorbidity other than diabetes, and those with diabetes was about 5 times compared with those without. Although FN populations make up a small proportion of the Canadian population, they are the fastest growing demographic (1), therefore, enhancing the pathway from screening to diagnosis to treatment could improve survival after a breast cancer diagnosis. We need to identify culturally competent approaches to improve the FN breast cancer survival disparity, approaches that consider their overall health status, and the complexity of factors embedded in the social determinants of health.

No potential conflicts of interests were disclosed.

The authors thank the study staff, Lindsay Stewart, Sarah Ahmad, Lucia Mirea, and Tanya Cecic, and endorsement by the Joint Ontario Aboriginal Cancer Committee composed of Aboriginal representatives from each of the 9 provincial/territorial Aboriginal organizations (PTO), including a representative from the Independent First Nation and representatives from Cancer Care Ontario. We also thank the participants of the Aboriginal Breast Cancer Care Workshop (Toronto, Canada, April 2009), who shared their relevant experiences and contributed their interpretations of the study findings (34).

The study was funded by a grant from the Canadian Breast Cancer Foundation, Ontario Region. A.J. Sheppard's doctoral training was supported by an indigenous Health Research Development Program scholarship, a strategic Training Program in the Transdiciplinary Approach to the Health of Marginalized Populations fellowship, and a doctoral award from the Canadian Breast Cancer Foundation, Ontario Region.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.