Breast cancer in Ethiopia: evidence for geographic difference in the distribution of molecular subtypes in Africa

Breast cancer is the most common malignant neoplasm among women in Addis Ababa (Ethiopia). The Addis Ababa Cancer registry reports that breast cancer accounts for 34% of all female cancer cases, followed by cervical cancer at 16% [1]. Although breast cancer has a markedly higher incidence in developed countries, half of new breast cancer diagnosis and about 60% of breast cancer deaths occur in the developing world [2].

A significant increase in breast cancer incidence is reported in most Sub-Saharan Africa countries suggesting an increasing public health problem in a continent with existing infrastructures having been developed mainly for maternal, child health and infectious diseases [3]. Widespread urbanization, changing patterns of reproductive and environmental risks factors, obesity, decreased physical activity, and increasing life expectancy [3, 4] are among the salient factors implicated in the steady rise in breast cancer incidence across low income countries.

Breast cancer is a heterogeneous disease with different morphological and molecular subtypes [5, 6]. The morphological classification is still the foundation of histopathological diagnosis, but in the era of modern personalized medicine, a number of molecular classification systems have been introduced. In this context, estrogen receptor (ER) and progesterone receptor (PR) analysis in breast carcinomas were the first established biomarker assays with both prognostic and predictive power and they have been in use since the 1970s [7]. Human Epidermal Growth Factor 2 (HER2) was discovered in the 1990s and HER2-targeted therapy was subsequently introduced; the introduction of targeted antibodies that interfere with HER2 signaling followed and their use has led to improved survival among breast cancer patients whose tumors overexpress HER2 [8].

Further studies have revealed distinct molecular subtypes of breast cancer based upon the gene expression profiling of tumors [9]. Since classifications based on gene expression are still considered unaffordable in most global settings outside high income countries of the Western hemisphere, immunohistochemistry (IHC) studies have gained widespread acceptance as potential robust surrogates for more complex bar-code type assays. A combination of ER, PR, and HER2 defines the molecular classes initially identified by gene expression profile [10]. The classification systems used by Nielsen [11], Sotiriou [12] and the St. Gallen international panel of experts [13] in 2004, 2006, and 2011 respectively are widely used among research communities. The St. Gallen international panel of experts incorporated the well-established proliferation-associated antigen Ki67 that may also provide additional prognostic and predictive value to the previous classification systems [13, 14]. Several studies show that molecular subtypes of breast cancer vary substantially in their behavior and response to treatment [15]; however, at present none of these schema are widely used outside of Western countries.

With increasing opportunities for implementing modern health care in Sub-Saharan African countries, more in-depth knowledge of African breast cancer molecular characteristics compared to the more studied western countries is essential for improving prognosis and management of breast cancer.

Few studies have been conducted in Africa on the clinicopathological and biological characteristics of breast cancer in Sub-Saharan African countries. The published studies conducted so far exhibit some degree of divergence in their results. The majority of these studies, reviewed by Brinton et al., were mainly carried out on west and central Africa and reported that breast cancer patients exhibit specific features compared to the western countries [16]. According to these authors, breast cancer in Sub-Saharan Africa is reported to occur in younger age and show more aggressive features such as high-grade tumors and triple-negative phenotypes (negative for expression of ER and PR and for over-expression of HER2) [16]. Similar patterns to these African studies have been reported among African American women with disproportionately high incidence of triple negative phenotype as compared to White Americans [17–19]. Preliminary evidence, however, shows geographical diversity in the distribution of breast cancer molecular subtypes in Sub-Saharan African region with the rate of triple negative breast cancer reported to be relatively lower in East Africa [20–22]. In the present study, we aimed to study a large collection of cases from a major Ethiopian institution where women are treated for breast cancer and investigate the molecular subtypes using the most recent receptor-based classification [13, 14].

Progress in molecular research have led to the classification of breast cancers into distinct subgroups (luminal, normal breast-like, Her2/neu-positive and basal-like subtypes) based on mRNA expression profiles [25]. However, gene expression profiling using cDNA microarray or RNA seq technology is not currently feasible in clinical settings due to its high cost and technical complexity. Therefore, IHC markers have been validated and used as surrogates for cDNA microarray in molecular subtyping of breast cancer [26].

In the present study, we found that luminal A subtype was the most prevalent followed by luminal B subtype, Triple Negative, and HER2-enriched derived exclusively. Our findings, from TMA IHC, are in contrast to most other studies performed in standard paraffin sections, in tissues from Sub-Saharan Africa particularly in west and central African countries where triple negative phenotype was reported to be the most common molecular subtype (43–82%) of breast cancer [16]. In Europe, North America, Asia and middle eastern countries, 30–70% of breast cancers are luminal A tumors and our result (40%) is comparable to the distribution seen in the western societies [27, 28]. However, Luminal B which is the second most common molecular subtype in our study (26%) is present at higher proportion among our study participants when compared to the rate seen in the Western high resource regions, Asia and middle east (10–20%) [27, 28].

A previous study among Ethiopian women carried out by Kantelhardt [20] in 2014 compared ER negativity in 352 patients out of 1208 consecutive patients treated at Addis Ababa-University Hospital, Ethiopia, from June 2005 through December 2010. They reported ER negative tumors to be around 35% in agreement with our present results [20]. A comparable result was also reported in a recent study done by groups at the University of Michigan in 2016 from one of the referral hospitals in Addis Ababa where they found ER negative tumors to be 26.5% [22].

Similarly the frequencies of triple negative tumors in these East African studies were lower than reported from west and central Africa. TNBC in the Kenyan study was reported in 20.2% of all cases [21] and TNBC in the Michigan study among Ethiopian women was reported to be around 15% which is lower than our finding (23%), but all of these studies indicate TNBC tumors to be less common than reported elsewhere in Africa. Hence, taken together, the data from East Africa suggest that it would not be accurate to associate African ancestry with increased probability of diagnosis with ER negative or TNBC tumors, as the Sub-Saharan African populations are known to be themselves highly heterogeneous in lifestyles, exposures, and genetic admixtures.

Present study has shown HER2 positivity at a slightly lower proportion than the rate seen among Ethiopians in the Michigan study [22] study (33%), the only other study (to our knowledge) which has incorporated HER2 assessment for molecular phenotyping of breast cancer from Ethiopia. However, HER2 positivity in our study (23%) is comparable to white Americans, African Americans, and West Africans which is about 17%, 19% and 20% respectively [22].

Our result shows an earlier onset of luminal B compared to the other molecular subtypes with statistically significant difference in median age at diagnosis (P < 0.05). The median age at diagnosis in our study also reveals a decline in luminal B subtypes as patients’ age increase with the incidence of luminal A exceeding that of luminal B at age ranges > = 50. This finding is comparable to a studies in the west where a bimodal age distribution at diagnosis is seen where incidence of the more aggressive phenotype luminal B peaks at earlier ages whereas luminal A type peaks at older ages [30].

The median age at diagnosis in our study was comparable to most previous studies in Africa and the few studies done among breast cancer patients in Ethiopia [20, 22]. However, it is much lower than the median age at diagnosis in most Western countries which is (55–60 years) [31]. The difference could be explained by the fact that African nations have younger population pyramids and the proportions reported in this and earlier studies are not age adjusted so the distributions are expected to always be shifted to younger ages in African cases compared to the cases in western countries; only 5% of African population is older than 60 years as compared to 24% of the population of Europe [32] being > 60 years old. More extensive population based registries in Africa, with active registration of cases, are urgently needed to understand the burden of cancer and to aid in cancer control program. Gaining a better understanding of environmental and lifestyle risk factors is crucial, as the incidence of breast cancer appears to be increasing at all ages in all African regions.

There was no statistically significant difference between median age at diagnosis and the ER status of breast cancer in this study. In this study, invasive ductal carcinoma was the predominant histological type (60%), which is comparable to the study by Kantelhardt in 2014 [20]. No statistically significant correlation was found between the molecular subtypes of breast cancer and histological type of breast cancer in this study, although the numbers of non ductal histologies were very low in the study. High tumor grade (Grade III) was reported in 34% of our study participants which is comparable with the study done by Kantelhardt [20]. The increased rate of high grade tumors observed in this study is in part possibly due to late diagnosis of breast cancer. No statistically significant correlation was found between the molecular subtypes of breast cancer and tumor grade in this study. The numbers of cases in each category were likely too small to appreciate differences, if they exist.

Limitations of the present study include selection bias which is due to enrollment of participants with available FFPE in the pathology laboratory, missing clinicopathological information and small sample size. Larger studies from population based samples are necessary to help guide cancer control programs. In spite of its stated limitations, our study adds strength to the notion that investments in utilization of hormonal and anti-HER2 therapies have potential to impact survival in Ethiopia, so efforts to spread education about implementing these strategies are recommended in light of the relatively high proportion of tumors amenable to these treatments.

In conclusion, our study confirms the findings of studies from Ethiopia and other East African countries that hormone receptor negative tumors are not the most common molecular subtypes of breast cancer in this particular part of Africa. Hence, majority of breast cancer cases in this population may benefit from hormone therapy and/or anti-HER2 or other targeted therapy. Additionally, our findings and other East African Studies confirm the geographic variability in the distribution of the molecular subtypes of breast cancer in Africa and hence have important clinical and policy implications for breast cancer control. Furthermore, Ethiopian breast cancer patients exhibit highly proliferative Luminal B tumors at young ages. Future research should examine currently recognized as well as novel genetic and environmental factors that may contribute to the tumor characteristic differences between different populations in Africa.