Receptor-Defined Breast Cancer in Five East African Countries and Its Implications for Treatment: Systematic Review and Meta-Analysis.

PURPOSE: Estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor-2 (HER2) are determinants of treatment and mortality for patients with breast cancer (BC). In East Africa, the estimated 5-year survival (37.7%) is far lower than the US average (90%). This meta-analysis investigates BC receptor subtypes within five East African countries to ascertain cross-country patterns and prioritize treatment needs.
METHODS: From a PubMed search, January 1, 1998-June 30, 2019, for all English-only BC articles for Ethiopia, Kenya, Rwanda, Tanzania, and Uganda, eligible studies had receptor distributions for female BC samples ≥ 30 patients. Outcomes were proportions of ER+, PR+, and HER2-positive (HER2+), and/or molecular subtypes. Data included study characteristics and mean or median patient age. Using metaprop, Stata 16, we estimated pooled proportions (ES) with 95% CIs and assessed heterogeneity.
RESULTS: Among 36 BC studies with receptor data, 21 met criteria. Weighted mean age was 47.5 years and median, 48. Overall ES were as follows: 55% for ER-positive (ER+) (95% CI, 47 to 62), 23% for HER2+ (95% CI, 20 to 26), and 27% for triple-negative BC (TNBC) (95% CI, 23 to 32).
CONCLUSION: We found differences between countries, for example, lower distribution of TNBC in Ethiopia (21%) compared with Uganda (35%). ER+, the dominant BC subtype overall at 55%, emphasizes the need to prioritize endocrine therapy. Overall proportions of HER2+ BC (with or without ER+ or PR+), 23%, approached proportions of TNBC, 27%, yet HER2 testing and treatment were infrequent. Testing and reporting of receptor subtypes would promote delivery of more effective treatment reducing the mortality disparity.

INTRODUCTION

Among women worldwide, breast cancer (BC) is the most frequently diagnosed cancer and the leading cause of cancer-associated deaths, accounting for almost one quarter of incident cancer cases and 15% of cancer deaths.[1] Although age-adjusted incidence in Eastern Africa is about one-third that of regions composed of high-income countries (HIC, ie, Northern America, Northern and Western Europe, Australia, and New Zealand),[1] this rate may underestimate the true incidence as registry data, even when available, are often restricted to a specific city or county. Also, multiyear assessments show increasing rates likely linked to population aging and lifestyle changes.[2,3] Despite substantially lower incidence, age-standardized BC mortality rates in Eastern Africa (15.4 per 100,000) are similar to rates in Western Europe and higher than other affluent regions.[1] The estimated 5-year BC survival in East Africa is 37.7%, compared with 35.2% and 48.1% in West Africa and South Africa, respectively.[4] For US women, the average 5-year BC survival is 90%.[5]

CONTEXT

Key Objectives

To understand the overall distribution and differences in breast cancer (BC) subtypes among five East African countries. Estrogen receptor (ER), progesterone receptor, and human epidermal growth factor receptor-2 (HER2) are determinants of BC treatment and mortality, yet there has been limited review and meta-analysis of receptor studies within this region.

Knowledge Generated

Overall, ER-positive (ER+) was the dominant subtype, whereas the proportion of HER2 positive approached that of triple-negative BC (TNBC). Country differences, for example, lower distribution of TNBC in Ethiopia (21%) compared with Uganda (35%), were only partly explained by publication year.

Relevance

Improved testing and reporting of receptor subtypes are essential for the delivery of effective treatment to reduce the mortality disparity. ER+, as the dominant BC receptor, emphasizes the need to prioritize accessible multiyear endocrine therapy. More robust interventions for TNBC should be incorporated as available. Anti-HER2 treatment will require global cost reduction and access strategies.

BC mortality in East African countries has been exacerbated by late diagnosis and treatment that fails to meet common international standards. Primary care physicians, nurses, and other healthcare professionals are in short supply, and credentialed specialists in oncology are rare.[6,7] There is limited or no access to radiation,[8] which constrains the possibility for breast-conserving surgery even among women with early-stage BC. Many cancer medications routinely used in HIC for treatment, palliative care, or to control adverse effects of treatment are unavailable, and supplies of available less expensive medications fall short of need.[9-11] Cancer treatment, such as chemotherapy, is confined to a few locations in an entire country in East Africa and care at these sites may not conform to current standards for diagnosis and treatment.[7,12] There is no system for routine preventive care or early detection of BC. Women seek care in response to breast symptoms.[13,14] In addition, because clinical pathways for BC diagnosis are often poorly defined, women may have to pursue several levels of care before reaching a facility with the capacity to diagnose BC.[11,13]

BC is a heterogeneous disease with different morphologic and molecular subtypes.[15,16] Based on immunohistochemistry (IHC), it is characterized by the following receptor subtypes in the clinical setting: estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor-2 (HER2). These subtypes are prognostic markers for mortality and major determinants of evidence-based treatment guidelines.[12] ER-positive (ER+) tumors typically have a better prognosis since hormone-targeting therapy (such as tamoxifen and aromatase inhibitors) substantially decreases mortality.[17,18] HER2-positive (HER2+) patients acquire a significant survival advantage when treated with humanized monoclonal antibodies against HER2/neu. By contrast, triple-negative BC (TNBC), defined by the absence of ER, PR, and HER2 expression, is more aggressive and negatively associated with survival.[19]

Relative distribution of BC subtypes would provide prognostic and therapeutic information to guide treatment and health system planning. The first systematic review and meta-analysis of BC subtypes identified seven studies from countries in East Africa (two each from Kenya, Tanzania, and Uganda, and one from Madagascar) among 26 from sub-Saharan Africa.[20] Although results from this review suggested that ER+ BC was identified in 41% of patients in East Africa,[20] recent research has reported substantially higher proportions.[21-23] The clinical relevance of receptor subtypes for treatment and the variable findings for the relative proportions of ER+ BC and TNBC in East Africa highlight the need to update the prior review. Because individual countries frame policies and priorities for cancer care,[11,24] it is advantageous to examine receptor-defined patterns for BC nationally. We performed a systematic review and meta-analysis to characterize receptor-defined subtypes of BC in five East African countries, namely Ethiopia, Kenya, Rwanda, Tanzania, and Uganda. Although the region consists of 21 countries, we focused on the four most populous countries in the region with established cancer centers, plus Rwanda, a country undergoing recent expansion of its cancer services.[6] The countries we designated make up approximately 62% of the total population for East Africa.[25]

METHODS

We conducted the systematic review and meta-analysis following PRISMA guidelines.[26] Searches of PubMed for English-only articles used the term breast cancer combined with names of each of the five countries of interest for the period January 1, 1998 through June 30, 2019. For inclusion, studies had to report single receptor or combination subtypes (eg, ER+ or PR+/HER2−). Studies were required to have female BC samples of at least 30 patients. Study authors were not contacted.

Data abstraction was independently performed by two reviewers (E.M.G. or P.P.) using a prespecified Excel spreadsheet. Prior to coding, authors designated and defined variables in a data dictionary. Coding disagreements were resolved by consensus between the two reviewers or by a third coauthor (C.O.). Categories of information included (1) study characteristics (Table 1); (2) patient age, assessed by median or mean age; (3) BC pathology, consisting of reported cancer stage, grade, and histologic subtype (ductal proportion); (4) receptor testing, incorporating timing (prospectively collected v retrospectively archived tissue), and testing or scoring used for HER2+; and (5) outcomes, composed of three receptor statuses (ER+, PR+, and HER2+) and/or four receptor combinations (ER+ or PR+/HER2− [luminal A]; ER+ or PR+/HER2+ [luminal B]; ER−/PR−/HER2+ [HER2 enriched]; or ER−/PR−/HER2− [TNBC]). These receptor combinations, used by the US SEER registries, are the most widely used.[27] When reported, we described treatment with oral endocrine therapy (tamoxifen or aromatase inhibitors), including treatment numbers and linkage of therapy to hormone receptor-positive (HR+, that is either ER+ and/or PR+) BC.

TABLE 1

Overview of Included Studies

We examined three factors that may influence receptor status results: timing of receptor testing, study date, and proportion of grade 3 BC. In the prior meta-analysis composed of studies from both North and sub-Saharan Africa, ER+ was decreased by 10% among studies using archival tissue compared with prospectively collected tissue. In our study, we operationalized timing of receptor testing similarly.[20] We compared studies by year of publication (≥ 2014 [last 5 years] v < 2014) to reflect expansion of cancer care in the region, as well as potential for increased recognition of receptor status for BC treatment.[6,23,28] In addition, ER+ was 9% lower among studies reporting higher grade 3 disease compared to those with lower levels (≥ 40% grade 3 v < 40%).[20] To ensure sufficient numbers in our comparison groups, we dichotomized studies based on the median value for proportion of grade 3.

We calculated the proportions with each relevant receptor outcome using the number with a positive status as the numerator and the total number with a known status as the denominator. Median or mean age in years was abstracted from each study as reported. Alternatively, we calculated a weighted estimate when means or medians were available only for subgroups.

Statistical Analysis

Using Stata 16 and the metaprop procedure, version 10.1 (2016), from Boston College Statistical Software Components, we pooled proportions from multiple studies applying a random-effects model that displayed results as forest plots.[29] We employed the exact or Clopper-Pearson method to generate CIs for proportions from the selected individual studies and a Freeman-Tukey transformation to normalize outcomes before calculating pooled proportions (ES). Metaprop tests for intragroup heterogeneity of pooled proportions through the I[2] statistic and its P value, although only if the pooled proportion includes at least four studies. The null hypothesis reflects variation consistent with chance alone (presumed homogeneity), whereas rejection of the null hypothesis based on a significant P value reflects intragroup heterogeneity linked to clinical and methodologic diversity.[30] Given two or more groups, metaprop tests for subgroup or intergroup heterogeneity based on the heterogeneity statistic, degrees of freedom, and P value. If the P value for subgroup heterogeneity lacked significance, groups were considered homogeneous.[29]

We generated country-specific and overall pooled proportions, along with associated tests of heterogeneity, using available study data for all single and combination receptor statuses. In a second set of results, we explored the effect of potential bias and explanatory factors on receptor outcomes, by categorizing available studies separately on timing of receptor testing, publication year, and proportion of grade 3 disease, and then generating pooled proportions for receptor outcomes by these groupings. We applied a separate Bonferroni correction to our significance level of 0.05 for the nine and 12 tests of heterogeneity by explanatory factors for the single (0.05/9 = 0.005) and combination (0.05/12 = 0.004) receptor outcomes, respectively. Only levels below this threshold were considered significant.

RESULTS

Our PubMed search identified 319 publications after eliminating duplicates (Fig 1). After all exclusions, we had 21 eligible articles with the following distribution: Ethiopia = 4,[21,22,31,32] Kenya = 8,[23,33-39] Rwanda = 1,[9] Tanzania = 4,[40-43] and Uganda = 4.[44-47] Although 36 articles had receptor data, 13 articles were excluded because of identical or overlapping samples[48-60] and two did not meet study criteria.[61,62] When multiple publications addressed similar samples, we chose the earliest or most comprehensive profiling of receptors for inclusion and designated other publications as supplementary. For three eligible publications,[34,41,44] we amplified available data by using results from the supplementary publications (Table 1).

FIG 1

Flow diagram of search results.

The overall weighted mean age was 47.5 years (SD 3.2) among 3,307 patients. By country, mean ages were 42.3 (SD 0.4) Ethiopia, 46.0 (0.4) Uganda, 49.3 (SD 2.5) Kenya, and 49.5 (SD 0.4) Tanzania (no mean age for Rwanda). The overall weighted median age was 48 years among 2,051 patients with country rankings similar to those for mean age. The weighted median age for Rwanda was 48 years, identical to Kenya.

Clinical characteristics included BC grade, stage, and histologic subtype, respectively addressed by 16, 14, and 15 publications with aggregated samples ranging from 2,167 (grade) to 1,951 patients (stage). The overall ES were 55% for grade 3 BC (95% CI, 50 to 60), I2 of 82.7%, P ≤ .001, range 75% in Rwanda (95% CI, 66 to 82) to 50% in Tanzania (95% CI, 33 to 67); 25% for stage 1 or 2 BC (95% CI, 18 to 33), I2 of 93.5%, P ≤ .001, range 37% in Kenya (95% CI, 33 to 42) to 17% in both Tanzania (95% CI, 7 to 29) and Uganda (95% CI, 13 to 20); and 88% for ductal subtype (95% CI, 84 to 91), I2 of 76.8%, P ≤ .001, range 89% in Kenya (95% CI, 84 to 93) to 86% in Ethiopia (95% CI, 70 to 97). Tests of heterogeneity by subgroup (country) showed significant differences by proportions of grade 3 (P ≤ .001) and stage 1 or 2 (P ≤ .001), although not ductal subtype (P = .97) (results not shown).

Eighteen publications (n = 2,875) addressed ER+ BC (Fig 2). The overall ES was 55% (95% CI, 47 to 62), range 62% in Ethiopia (95% CI, 52 to 71) to 42% in Uganda (95% CI, 36 to 49). Although the highest ES, 68% (95% CI, 59 to 75), was calculated for Rwanda, we substituted the combined number of ER+/PR+ as ER+ alone was not reported. Sixteen publications (n = 2,661) addressed PR+. The overall ES was 46% (95% CI, 38 to 54), I2 of 93.2%, P ≤ .001, range 52% in Kenya (95% CI, 40 to 63) to 28% in Uganda (95% CI, 22 to 36) (not shown). Eighteen publications (n = 2,689) addressed HER2+ BC (Fig 3). The overall ES was 23% (95% CI, 20 to 26), range 27% in Ethiopia (95% CI, 18 to 37) to 21% in both Tanzania (95% CI, 14 to 30) and Uganda (95% CI, 15 to 29). Tests for heterogeneity by country showed significant differences in proportions of ER+ (P ≤ .001) and PR+ (P ≤ .001), not HER2+ (P = .85).

FIG 2

Proportion of ER+ breast cancer by study, designated country, and overall. ER+, ER-positive.

FIG 3

Proportion of HER2+ breast cancer* by study, designated country, and overall. *Among 18 studies, determinations of HER2+ were based on IHC + fluorescent in situ hybridization n = 6 (33.3%), IHC alone, score of 2 n = 2 (11.1%), IHC alone, score of 3 n = 6 (33.3%), not described n = 4 (22.2%). HER-2+, human epidermal growth factor receptor-2-positive; IHC, immunohistochemistry.

For molecular subtypes, fourteen publications (n = 2,445) addressed luminal A and luminal B, whereas fifteen publications (n = 2,489) and sixteen publications (n = 2,575) addressed HER2-enriched and TNBC, respectively. No subtype data were available for Rwanda. For luminal A, 48% was the overall ES (95% CI, 43 to 54), I2 of 84.9%, P ≤ .001, range 53% in Kenya (95% CI, 45 to 60) to 41% in Uganda (95% CI, 37 to 45). For luminal B, 11% was the overall ES (95% CI, 8 to 14), I2 of 82.0%, P ≤ .001, range 19% in Ethiopia (95% CI, 14 to 25) to 7% in Uganda (95% CI, 4 to 11). For HER2-enriched, 11% was the overall ES (95% CI, 8 to 14), I2 of 80.3%, P ≤ .001, range 16% in Uganda (95% CI, 12 to 20) to 3% in Tanzania (95% CI, 0.3 to 7). For TNBC, 27% was the overall ES (95% CI, 23 to 32), range 35% in Uganda (95% CI, 29 to 41) to 21% in Ethiopia (95% CI, 16 to 27) (Fig 4). There was heterogeneity by country for proportions of luminal A (P = .04), luminal B (P ≤ .001), HER2-enriched (P ≤ .001), and TNBC (P = .005).

FIG 4

Proportion of triple-negative breast cancer (TNBC) by study, designated country, and overall.

For factors potentially linked to receptor outcomes, we showed the following (Table 2): in comparisons by timing of testing (retrospective v prospective), we found no significant effects for timing in results for ER+, PR+, or HER2+ BC. By publication year (2014-2019 v 2000-2013), the heterogeneity test showed significant effects for publication year for ER+ (P ≤ .001) and PR+ (P = .003), although not for HER2. For ER+ BC, the ES among recent versus earlier studies was 64% (95% CI, 58 to 69) compared with 40% (95% CI, 33 to 49). Similarly, the proportions of PR+ BC among recent versus earlier studies were 54% (95% CI, 46 to 61) and 34% (95% CI, 24 to 45%), respectively. By proportion of grade 3 BC (≥ 54% v < 54%), we found a significant subgroup effect (P ≤ .001) only for HER2+ proportions such that the ES among studies having higher versus lower distributions of grade 3 BC were 26% (95% CI, 24 to 29) and 17% (95% CI, 14 to 20), respectively.

TABLE 2

Breast Cancer Receptor Outcomes by Timing of Testing, Publication Year, and Proportions of Grade 3 Breast Cancer: Estimated Proportions (ES) and 95% CI With Number of Published Studies (Pubs) and Weight Percentages (WT)a

We also examined the effect of explanatory factors on the four receptor combinations (table not shown). Tests between subgroups for timing of testing were only significant for proportions of luminal A (P = .004). Among studies conducted with prospective versus retrospective testing, luminal A was 54% (95% CI, 46 to 61) compared with 42% (95% CI, 39 to 45), respectively. For publication year, although tests for heterogeneity did not reach the corrected significance level for proportions of luminal A (P = .06), luminal B (P = .21), HER2-enriched (P = .04), and TNBC (P = .04), there was a consistent pattern of higher ES for luminal subtypes and lower ES for HER2-enriched and TNBC in more recent publications. For example, for TNBC, the ES in recent compared with prior years was 25% (95% CI, 20 to 29) versus 34% (95% CI, 26 to 43), respectively. We found no significant differences by proportions of grade 3 BC for any receptor combination.

Six of 21 studies addressed treatment with endocrine therapy. Three limited endocrine therapy to patients who were ER+ and/or PR+ with the following proportions treated: 42.0% (29/69),[32] 84.3% (86/102),[39] and 89.8% (88/98).[9] In two studies, endocrine therapy was not limited to patients with HR+ BC[37,42] and one early study reported no tamoxifen treatment.[43]

DISCUSSION

In our meta-analysis, ER+ was the predominant BC receptor, 55%, overall in the five East African countries, compared with 72% in the United States.[27] The review revealed large between-country variability in the estimated pooled proportions of ER+ BC, ranging from 62% in Ethiopia to 42% in Uganda. In studies published from 2014 to 2018, the ER+ proportion was almost 25% higher compared with those from earlier years, 64% versus 40%. This difference in ER+ may explain why the prior meta-analysis had an overall proportion of 41% for East Africa, as six of seven studies were published prior to 2014.[20] Improved tissue handling with quality histopathology and IHC may have enhanced detection of ER+, sometimes through support and telepathology from international cancer facilities.[9,63] We found higher proportions of luminal A BC with prospective testing (54%) versus archival testing (42%). However, similar comparisons addressing other receptors and combinations subtypes were not significant.

Study recency and improved IHC evaluation may explain the higher overall proportions of ER+ BC in Ethiopia, Kenya, and Rwanda compared with Tanzania and Uganda. Biologic factors may also influence the lower ER+ proportions in Tanzania and Uganda. Prior BC studies have shown that younger women have lower rates of ER+ BC and higher rates of TNBC.[27] Yet, one study found a higher proportion of ER+ BC among Ethiopian women compared with African American and Ghanaian women despite their younger mean age (42.7 years compared with 60.2 and 49.3 for African Americans and Ghanaians, respectively).[31]

Endocrine therapy is an essential component of treatment for women with ER+ and/or PR+ BC, regardless of HER2 status,[12,18] although only six of the 21 studies included data on treatment with endocrine therapy. Furthermore, of the five that reported administration of endocrine therapy, only three limited such treatment to patients with HR+ BC, suggesting that other studies started patients on endocrine therapy when IHC for ER was unknown or the results were delayed. Despite its ease of use and the potential for clinic personnel to monitor administration outside specialized cancer centers, suboptimal use of endocrine therapy appears widespread. Because generic tamoxifen is relatively inexpensive, it could be readily available, although access is determined by multiple factors.[64]

The overall estimated proportion of HER2+ BC in our study was 23%, compared with under 15% in the United States.[27] Our review revealed minimal variability in pooled proportions of HER2+ across countries (21% to 27%), despite variability within each country. Because incongruency in defining HER2+ status may have contributed to within-country variability, standardization in scoring for future studies is recommended.

Although HER2+ BC is subdivided by ER and/or PR expression, both groups benefit from treatment with HER2-targeting therapy, for example, trastuzumab. Trastuzumab is now included in the WHO’s model list of essential medicines (2015),[65] as its use in conjunction with chemotherapy has modified the formerly poor outcomes for this subtype.[66,67] Although the standard of care for HER2+ early BC has been one year of adjuvant trastuzumab with chemotherapy,[68] accumulating evidence indicates equivalence between 6 and 12 months of trastuzumab for selected patient profiles.[69]

In many countries of sub-Saharan Africa, there is currently minimal assessment of HER2, primarily because anti-HER2 agents are prohibitively expensive and often unavailable.[6,70] In a recent study, trastuzumab was not shown to be cost-effective in 11 African countries at current prices.[70] Shortened trastuzumab regimens for early BC, as well as availability of multiple biosimilars for all BC,[71] will reduce drug costs. Yet, affordability in sub-Saharan Africa may require global cost-reduction strategies similar to prior collaborations to bring effective antiretroviral medications to low-income countries.[72] The introduction of subcutaneous trastuzumab may increase treatment feasibility in East African countries,[73-75] where there are very few hospitals to administer infusion treatments.[65]

The overall estimated proportion of TNBC in our study was 27% with across-country variability from 21% to 35% in Ethiopia and Uganda, respectively. In the United States, the proportion of TNBC in SEER data was 12.2%,[27] although the proportions among White (10.7%) and African American women (22.5%) showed marked disparity.[27] Higher proportions of TNBC among people of African ancestry may be associated with inherited genetic susceptibility to TNBC.[76] In our meta-analysis, the proportion of TNBC was 9% lower among studies conducted in the last 5 years compared with studies conducted in prior years. The lower proportion of TNBC in Kenya and Ethiopia needs further study as it could be linked to testing and/or biologic factors.

Numerous studies have also identified multiparity and early age for the start of childbearing as strongly associated with the elevated incidence of TNBC.[77-79] Despite high average parity and early childbearing in Kenya and Ethiopia, these countries have lower estimated proportions of TNBC compared with Tanzania and Uganda. Median age at first birth ranged from 18.9 years (Uganda) to 23 years (Rwanda) and children per women from 3.4 (Kenya) to 5.5 (Uganda).[80]

Although TNBC currently has the worst prognosis, achievement of a pathologic complete response (pCR) after neoadjuvant chemotherapy confers a strong prognostic advantage.[81] Although studies do not include an African population, the proportion of patients achieving a pCR after optimal neoadjuvant chemotherapy is 33% to 45%.[82,83] Per NCCN guidelines, BRCA1/2 testing is part of treatment planning in HIC countries for patients that meet designated criteria. Women with TNBC have a higher prevalence of germline BRCA mutations.[84] In two population-based studies in HIC, among patients with TNBC, 19.5%-22.7% had BRCA mutations.[85,86] Women with BRCA mutations have more than a 50% chance of developing TNBC, and founder BRCA1 mutations have been identified in BC patients of African ancestry.[87] Most data regarding BRCA mutations in Africa are based upon northern and western African populations[88,89]; therefore, BRCA research needs to be conducted in East Africa.

In the present study, compared with BC among US women, the disease in East African women was characterized by presentation at a younger age, a more advanced stage, and a higher grade. Similar to other regions of the world, most BC among East African women had invasive ductal histology. Preventive health care and BC screening are limited or nonexistent in East Africa,[7,13,90] leading to BC diagnoses at advanced stages and higher grades. Although mammography screening has been linked to reduced BC mortality in US women,[91,92] timely access to BC diagnosis and treatment among symptomatic women in East Africa would likely improve BC mortality.[14,93-99] In our study, the overall proportion of grade 3 BC was 55%, and did not vary by proportion of receptor combination subtypes. In the United States, the proportions of grade 3 tumors were markedly different by subtype: 17% HR+/HER2−; 46% HR+/HER2+; 67% HER2-enriched; and 75% TNBC.[27] Our finding, which suggests ER+ tumors were often high grade, reinforces the need for extended endocrine therapy in East African countries.

The higher grade of BC on presentation could be because of the younger age of women diagnosed with BC in East Africa, about 15 years younger than their US counterparts, where it is 62 years.[5] The younger age in East Africa is likely related to the age profile rather than younger manifestations of BC.[34] In the United States, 30.8% of women are 55 years of age or older; in the countries we reviewed, 5%-7% are in this age category.[80] Relative to HIC, the disproportionate numbers of young and middle-aged women in East African countries has skewed the incidence of BC to younger ages. Yet, as average longevity increases, rates of BC are predicted to rise, a pattern already documented.[3]

Our study is the most comprehensive systematic review and meta-analysis for East Africa to date addressing five countries within this region and a total of 21 studies. Yet, the data available for the study had a number of limitations. Although the advantage of aggregating studies in a meta-analysis is to minimize biases in individual studies, it cannot compensate for insufficient study numbers or indefinite generalizability within a country. For ER+ BC, the range in number of studies per country was one to seven, whereas the median number of patients by country was 247 (range 148 in Rwanda to 1,594 in Kenya). We included the single Rwandan study[9] in our meta-analysis of ER+ by using their assessment of HR+ as a proxy for ER+. Rwanda was not included in the meta-analyses of receptor combinations because the single study[9] did not report this outcome. Future reviews should include available receptor studies from all countries designated as East Africa as those data become available.[25]

There was substantial within-country variability on receptor status and subtypes, which could be linked to sampling or methodologic issues. The significant variability among the seven studies of ER+ conducted in Kenya shows the secular pattern of higher and lower proportions of ER+ BC for recent and prior publication periods. The pooled estimate (56%) for Kenya may have been lowered by the inclusion of earlier studies. By contrast, the four studies of ER+ BC conducted in Tanzania appeared homogeneous despite variability in publication period.

Samples for included studies could be insufficiently representative or generalizable. Among the 18 studies, which drew samples from consecutive cases, more than a third (7 of 18) had receptor data for < 70% of the total.[21,22,37,41,42,44,47] The only population-based study sourced from the cancer registry in Kyadondo County, encompassing Kampala, Uganda, included only two-thirds of registry cases.[47]

The majority of studies were conducted at a single medical institution, typically in the country’s capital or largest city, had patient samples < 100 (n = 9) or between 100 and 250 (n = 9), and may not reflect BC patterns in other regions of the country.

In summary, this study provides updated estimates for the proportion of BC receptor subtypes defined by ER, PR, and HER2 status in five countries within East Africa. The proportion of ER+ BC is higher than reported,[20] underscoring a substantial role for extended endocrine therapy. Since endocrine therapy is available orally, broader delivery to patients appropriate for this treatment should be feasible. The sizable proportion of HER2+ BC, higher compared with HIC, suggests a clear need for strategies to build capacity in assessment and affordable treatment of HER2+ BC. Finally, TNBC remains a considerable proportion of BC in these five East African countries, as compared with HIC, emphasizing the need to maximize robust treatment of this aggressive subtype. Testing and reporting of receptor subtypes is critical for delivery of more effective treatment to reduce the mortality disparity for BC in East Africa.