Differences in Molecular Features of Triple Negative Breast Cancers Based on Age at Diagnosis

INTRODUCTION

The population of the United States is aging with about 20% (approximately 70 million) of the population expected to be over 65 years of age by 2030.¹ The incidence of breast cancer increases with age, resulting in a high percentage of women with breast cancer being over the age of 65 years.² Forty-five percent of all women diagnosed with breast cancers in the U.S. in 2014 were aged 65 years or older, and is projected to be half by the year 2030.^{3, 4} However, much of the understanding of breast cancer treatment is based on trials in which older women are underrepresented or excluded by design.^4–7 Breast cancers among older women typically have a less aggressive biology based on a higher percentage of hormone receptor positive tumors or a higher proportion of luminal subtype, lower proliferation rate, and lower grade tumors compared to younger women with breast cancer.^{2, 4, 8, 9} Patterns of molecular gene expression and genetic mutations also differ between breast cancers diagnosed in younger versus older women.^{8, 10} These differences in breast tumor biology according to age may have implications for treatment and survival.^{2, 4, 11} Triple negative breast cancers (TNBC), lacking expression of estrogen and progesterone receptor (ER and PR) and the human epidermal growth factor receptor HER2/neu (HER2), have poor prognosis and present a particular treatment challenge since there are no targeted therapies available. The triple negative (TN) phenotype is a frequent feature of breast cancers among women aged 35 and younger, with about one-third of breast cancers in this age group being TN. Even though a smaller proportion of breast cancers diagnosed among older women are TN¹², the number of TNBC in this age group is substantial because of the high incidence of breast cancer in older women.

TNBC based on clinically-determined estrogen, progesterone receptors, and HER2 protein expression and/or gene amplification is considered a reasonable surrogate for Basal-like gene expression subtype, but the correlation of TN phenotype and Basal-like subtype is not perfect, with an average positive predictive value of 70%.^{13, 14} Further, classification accuracy of clinical TN status for the Basal-like subtype of breast cancer may not be uniform across age groups.¹⁵

We examined molecular subtypes and gene expression in a population-based cohort of women with breast cancer using PAM50, an FDA-approved quantitative reverse transcriptase PCR (qRT-PCR) intrinsic subtyping tool that assays expression of 50 genes.^{13, 16} To our knowledge, no study has compared molecular subtypes and gene expression patterns of TNBC in older versus younger women. We described the classification accuracy of clinical TN status for the Basal-like molecular subtype of breast cancer for older ages at diagnosis compared to younger, and identified potential differences in the gene expression patterns of TNBC.

METHODS

RESULTS

The two cohorts (Pathways and LACE studies) included women diagnosed with breast cancer from age 25–91 years with the largest proportion of women between the ages of 50 and 59 years. The 1,319 cases with PAM50 assay results were similar to the parent cohorts with respect to clinical characteristics, patient age, and race/ethnicity.⁸

The proportion of patients with TNBC as defined by clinical tests who were white and non-Hispanic increased with age in a similar manner to the other phenotypes (Table 1). Histologic grade of poorly- or un-differentiated tumors predominated among women diagnosed with TNBC before age 50 years (85%) but was somewhat less prevalent (72%) among TNBC in the older age groups. The frequency of large tumors, in the 2–5 cm size category, was higher among TNBC than non-TNBC in each age group. Among TNBC, there was no indication that older women had a higher frequency of large tumors than younger women.

The proportion of both TNBC and Basal-like breast cancers among women older than 65 years at diagnosis was lower than the proportion in the younger age group (Table 2). The sensitivity and positive predictive value of the clinically-determined TN phenotype to correctly classify a breast cancer as Basal-like intrinsic subtype as determined by PAM50 decreased with age (Table 2). The trends of lower prevalence of TN and lower positive predictive value for Basal-like subtype with older age appeared to be present within subgroups by race/ethnicity, although due to small numbers within some age-race categories, statistical power was limited. The proportion Basal-like cancers for each age group was examined within subgroups by race and ethnicity. The proportion Basal-like declined with age among white non-Hispanics, among Hispanics, and among Asian and Pacific Islanders. For black or African-American women, the proportion of Basal-like cancers remained relatively high, ≥25%, for all age groups (Table 2).

PAM50 subtypes were significantly different among different age groups with TNBC (P < 0.001) (Figure 1). Eighty-seven percent of TNBC diagnosed in women under the age 50 years were of Basal-like subtype whereas this ratio was only 52% in women over 65 years of age. The HER2-enriched subtype was observed in 35% of women over 65 versus 10% among women younger than age 50 years. Molecular expression of ESR1 and PGR did not vary significantly by age at diagnosis among women with TNBC after adjusting for race and ethnicity (Figure 2). Expression of ERBB2 was positively associated with age (P = 0.02). When the expression of proliferation-associated genes among women with TN breast cancer was evaluated, age was inversely associated with proliferation score (average of the 10-gene signature) and with each proliferation gene individually (P < 0.01) after adjusting for race and ethnicity (Figure 3). We repeated the analysis of gene expression by age at diagnosis after restricting the sample to tumors confirmed as Basal-like by PAM50 (data not shown). Within the Basal-like sample, there was no association between age and expression of ESR1, PGR, or ERBB2. Among Basal-like tumors, trends for proliferation genes were in the same direction as those shown in Figure 3 among TNBC, i.e. higher expression of proliferation genes associated with younger age. However, the association between age and proliferation gene trends were attenuated and were statistically significant for only three of the genes (CEP55 P = 0.01, MYBL2 P = 0.005, and KIF2C P = 0.02).

When survival of women with TN tumor phenotype was compared to women with hormone-receptor positive breast cancer in the same cohort, TNBC was associated with increased risk of breast cancer recurrence and breast cancer death in the first five years after diagnosis in all age groups (Fig 4, Table 3). The association between TNBC and survival was independent of tumor stage and grade. When the period of follow-up was extended to the end of study follow-up, HRs for TNBC were attenuated but remained significant for recurrence in the 65+ age group (HR 1.91, 95% CI: 1.03–3.54). Among women diagnosed at age 65 years or older, the association between TNBC and adverse breast cancer outcomes was more apparent among those with tumors confirmed as Basal-like by PAM50.

DISCUSSION

It has long been postulated that breast cancer among older women has less aggressive characteristics, including a higher rate of hormone-receptor-positive tumors, lower histologic grade, and lower proliferation rate.^{2, 4} Although TNBC are less prevalent among older women, TN/Basal-like cancers are still diagnosed in significant numbers due to a higher incidence of breast cancers overall in this age group. In this study, poor prognostic factors such as poor and undifferentiated cancers, larger tumor size, and higher number of positive lymph nodes were more frequent among younger women both in TN and other phenotypes combined. The prevalence of both clinically determined TN and qRT-PCR (PAM50) determined Basal-like cancers were lower among women diagnosed with breast cancer after age 65 years. Sensitivity and positive predictive value of clinical TN status for classifying Basal-like molecular subtype also decreased with age. The prevalence of TN and Basal-like breast cancers vary by race/ethnicity and minority populations have different age distributions of breast cancer. Proportion of TN and Basal-like cancers was highest among African American women although TN phenotype was least sensitive in predicting Basal-like subtype in this racial group.

There was a significant difference in molecular subtypes among older and younger women. Only 52% of women over 65 years of age with TNBC at the time of diagnosis were of Basal-like molecular subtype compared to 87% of women younger than age 50 years at diagnosis. Thirty-five percent of TNBC seen in women over 65 were of HER2-enriched subtype. Among women with TNBC about 20–30% are of non-Basal like molecular type.²⁰ In this study the proportion of non-Basal like tumors, especially HER2-enriched cancers, was found to be significantly higher among older women who may benefit from targeted therapies. In post-menopausal women with hormone positive metastatic breast cancer having clinical HER2 negative/molecular HER2-enriched subtype have been shown to benefit from HER2-targeted (Lapatinib) therapy.²¹ Androgen receptor (AR) mRNA is highly expressed in both luminal and HER2-enriched molecular subtypes of TNBC. Luminal / Androgen Receptor (LAR) subtype, although being ER negative, has hormonally regulated pathways including androgen/estrogen metabolism. LAR TNBC have lower proliferation rates compared to other TNBC subtypes resulting in relative chemo-resistance in a neoadjuvant setting.²² The percentage of tumor cells positive for AR by immunohistochemistry is significantly higher in LAR subtype compared to other subtypes within TNBC. The AR is the focus of therapeutic targets in adjuvant and neoadjuvant setting in recent and ongoing clinical trials.^{23, 24} AR is not among the genes included in PAM50 and we did not have access to tissue blocks to perform immunohistochemistry for AR to test if the increase in HER2-encriched subtype in older patients correlates with LAR subtype.

ESR1 and PGR gene expression levels were similar in younger and older women with TNBC and unlikely to have contributed to the difference in the sensitivity of predicting Basal-like molecular type in these age groups. ERBB2 expression was higher among older women. TNBC are diverse group of cancers with non-uniform response to therapy.²⁵ Some significant differences between older and younger women with TNBC have been described. In studies using IHC staining, older women with TNBC were reported to have higher frequency of cytokeratin 5/6 expression, lower expression of EGFR,^{9, 12} lower rate of Ki67 labeling index and cytokeratin 7/8 positivity, and a higher rate of Bcl2 and cytokeratin 18 positivity.⁹ In the present study we used different methods, i.e. qRT-PCR, and evaluated a set of 10 proliferation-related genes: ANLN, CDC20, CENPF, CEP55, KIF2C, RRM2, UBE2C, MKI67, CCNB1, MYBL2. These proliferation genes correlate to Ki67 expression, and the last four genes are also included in the set of genes of commonly used commercially available OncotypeDX ® testing. We observed that all 10 proliferation genes were significantly lower among older women with TNBC. This result is consistent with the IHC-based studies in indicating that lower proliferation is characteristic of TN tumors from older women.

Several characteristics of TNBC in the population of women diagnosed at age 65 years or greater compared to TNBC in younger women documented here, including lower grade distribution, lower predictive value of clinical TNBC for molecular Basal-like subtype, and lower expression of proliferation genes, are suggestive of prognostic differences for TNBC by age group. However, when tumor size, nodes, and grade were accounted for, TNBC remained associated with breast cancer recurrence and death in elderly women. HRs in the age group diagnosed at age 65 or greater indicated that the worse survival outcomes were more apparent in those TNBC confirmed as Basal-like molecular subtype, but numbers of outcomes after subgrouping by age and PAM50 results were small, limiting ability to distinguish different survival patterns among these groups. The higher recurrence rates and breast cancer deaths among the youngest woman with non-Basal like TNBC compared to Basal-like TNBC is likely due to a lower response to chemotherapy due to lower proliferative rates and / or lower proliferation association associated gene expression levels. Although we do not have treatment information, it is likely that a significant proportion, if not all, of the patients younger than 50 received systemic therapy following the diagnosis of TNBC.

TNBC is a heterogeneous breast cancer which has been further sub-classified into four molecular subtypes with significantly different response to similar neoadjuvant therapy.^{25, 26} There is interest in developing targeted treatment to these subclassifications. Although the genes assayed in our study did not include all genes needed to define the sub-classes of Basal-like cancers, our findings indicate that age is a determinant of some of the heterogeneity among TNBC. Our findings of higher ERBB2 expression and lower proliferation-related gene expression raise the possible need for intrinsic subtyping in older women with TNBC when benefits of systemic chemotherapy are questionable or when potential targetable therapies become available for specific molecular subclassifications of TNBC.