We investigated clinicopathological features of breast cancers in a patient population unusual for its racial/ethnic and socioeconomic diversity. We focused on triple-negative tumours. Our results confirmed previously described associations between patient and tumour characteristics, and uncovered new associations. In particular, we found a three-fold increased prevalence of triple-negative tumours in black women, who comprised 43% of our patients compared with non-black women. Triple-negative tumours comprised equal fractions – approximately 30% – of breast cancers in younger and older black women, regardless of their likely menopausal status. Triple-negative tumours comprised equal fractions – about 30% – of breast cancers in obese and in non-obese black women. Considering all women together, regardless of race/ethnicity and age, we did note a trend toward an inverse association between triple-negative tumours and elevated BMI. Overall, these results suggest that black women of diverse backgrounds are much more likely to be diagnosed with triple-negative tumours, and therefore a poorer prognosis, regardless of older age or higher BMI, factors that in other populations may be associated with hormone receptor-positive tumours with a better prognosis.
Our observation that both younger and older black women have increased, equivalent proportions of triple-negative tumours contrasts with results from the Carolina Breast Cancer Study. That study found a higher prevalence of basal-like tumours only in premenopausal African-American patients [6
]. The contrast may be due to the unusual heterogeneity of our population, with consequent diversity of socioeconomic, lifestyle and genetic factors. Our study highlights the complexity surrounding the issue of race/ethnicity in medical research, and the potential differences in how each can be defined, measured and interpreted [41
]. The contrast could also reflect differences between the tumours that were studied. We defined tumours by a triple-negative phenotype and did not routinely determine if they were basal-like. The triple-negative tumours we could examine further showed morphological and immunohistochemical characteristics (i.e., medullary features and increased CK5/6 and EGFR staining) in proportions similar to what has been reported elsewhere, and are consistent with estimates that 80 to 90% of triple-negative tumours are basal-like [43
Perhaps our most intriguing result is the relation between BMI and tumour subtype. Existing studies reveal a complicated relation between BMI and breast cancer. Some studies find associations between increased BMI and increased risk of developing breast cancer, higher stage at diagnosis, greater likelihood of expressing markers of high cell proliferation, poorer response to neodjuvant chemotherapy and increased disease-specific mortality [46
]. On the other hand, there is a clear association between elevated BMI and postmenopausal, ER-positive and PR-positive breast cancers [52
], perhaps due to oestrogen production from adipose tissue. If obesity is important in determining hormone receptor status, then obese black and non-black women should have similar proportions of hormone receptor-negative tumours. However, we find that obese black women have four-fold more triple-negative tumours than obese non-black women. Therefore, factors other than whole body obesity must be crucial in determining hormone receptor expression. A specific type of obesity, an elevated waist:hip ratio, was associated with the basal subtype in the Carolina dataset [40
], but overall, the factors that determine subtype – genetics, microenvironment, environmental or developmental exposures – remain unclear.
The lack of association between obesity and hormone receptor expression in black women may contribute to our observation that, when we considered all 415 cases, we found a trend, rather than a significant association, between increased BMI category and decreased proportions of triple-negative tumours. The inclusion of a large subset of black women (43%), for whom increased BMI is not clearly associated with hormone receptor positivity, in the dataset may prevent us from seeing the association reported in other populations between increased BMI and hormone receptor positivity.
Our observations are consistent with the lack of association seen between BMI and overall risk of postmenopausal breast cancer in the Black Women's Health Study [15
]. Those authors speculated that if obesity confers mainly a risk of hormone receptor-positive tumours, then it would be difficult to detect an association between elevated BMI and postmenopausal breast cancer if a large proportion of the black women's cancers were hormone receptor negative. This is precisely our finding. Our observations are also consistent with observations that high BMI is associated with hormone receptor-negative tumours in cohorts of different ethnic composition [50
Potential limitations of our study include its relatively small size and lack of data on clinical outcome or on potential confounders, such as parity. These limitations, however, are balanced by strengths: the unusual, highly heterogeneous population, data extraction from computerised records augmented with manual abstraction for quality control, and a consistent, single-institution approach to pathological diagnosis and patient care. Although we do not have the socioeconomic status of each patient, the population in the database is highly likely to reflect the institution's overall socioeconomic status data. These strengths contributed to our database confirming many previously noted associations between clinical and pathological features.
The observations we report from this database have potential clinical implications. Several studies have documented the poor outcome of patients with triple-negative or basal-like tumours [35
], and the greater mortality of black women with breast cancer compared with other women, regardless of age [6
]. Our findings that a diverse group of both younger and older, normal-weight and obese, black women have approximately three-fold more triple-negative tumours than other groups, may be one factor contributing to the unfavourable prognosis of black women with breast cancer.