Introduction
In many areas of Africa affected by the HIV epidemic, HIV-associated malignancies are among the most common cancers in the overall population [
1]. For example, despite growing availability of antiretroviral therapy (ART), Kaposi’s sarcoma (KS), as of 2018, remains the most incident cancer in many parts of eastern and southeastern Africa, including Malawi, Mozambique, Uganda, and Zambia [
2]. Not only is KS common, but survival in these settings is also very poor. Our work across five African countries found that 45% of KS patients were lost to care by 2 years [
3], and, once those patients were tracked, corrected mortality was 54% at 3 years [
4]. One year survival after diagnosis of HIV-related KS in sub-Saharan Africa is estimated to be between 60 and 76%, [
4‐
8] though slightly higher in a tightly monitored randomized controlled trial setting [
9]. In contrast, in resource rich settings such as the U.S. and Europe, one year survival is 80–95% [
10,
11].
Much of the explanation given for the high mortality of HIV-related KS in Africa has been attributed to advanced stage at presentation, but the role of therapy must be considered as well. There are two established therapies: ART and chemotherapy. ART is indicated for all HIV-infected patients with KS regardless of cancer severity [
12], and can be effective alone as treatment in early stage disease. Indications for when to use chemotherapy are less clear. International guidelines vary slightly on when to begin chemotherapy [
12,
13], but do agree that chemotherapy is indicated for all but early stage KS. In resource rich settings, liposomal anthracyclines (e.g. Doxil) or paclitaxel are recommended first line agents, based on efficacy [
14]. Much less is known is about the use of chemotherapy in resource-limited settings such as sub-Saharan Africa, where cancer treatment facilities may have constrained resources and limited formularies. Most of the limited data on chemotherapy use for KS in sub-Saharan Africa either comes from clinical trials [
15,
16] or from specialized, intensive treatment centers operated in collaboration with major donor agencies [
17,
18]. Lacking comprehensive population-level data on real-world chemotherapy use in Africa, it is difficult to assess how much poor survival of KS is impacted by advanced stage presentation versus lack of access and effectiveness of available chemotherapy.
To address our limitations in knowledge about the epidemiology of chemotherapy use for HIV-related KS in Africa, we investigated all newly diagnosed cases of KS in a community-based HIV care program in East Africa. We determined the prevalence of indications for chemotherapy at the time of diagnosis, frequency of initiation of chemotherapy, drug regimens used, and factors associated with chemotherapy initiation.
Methods
Overall design
Via medical record review, we identified all patients newly diagnosed with HIV-related KS from 2009 to 2012 in the Academic Model Providing Access to Healthcare (AMPATH) consortium in Kenya. We determined indications for chemotherapy at the time of KS diagnosis (as detailed in Measurements below). In a cohort analysis, we then observed subsequent clinical course, including any therapy received, for 12 months following KS diagnosis.
Study population
To identify all KS cases over the specified time period, we reviewed the clinic network’s electronic database as well as pathology records, and then followed up with detailed paper chart review. We included all patients living with HIV > 18 years old receiving their care at AMPATH if they received a clinical or biopsy-confirmed diagnosis of KS during this time. At the time of this study, AMPATH provided care to approximately 160,000 patients with HIV through 26 major clinical care sites. There is an established KS biopsy service, which is free of charge. However, biopsy diagnosis was not universally standard of care at the time of the study, and patients were allowed to be treated based on clinical grounds alone when a biopsy was not done. ART is also free of charge. For patients needing chemotherapy, this treatment is provided by the Oncology Department at a tertiary care hospital in Eldoret, Kenya, as well as smaller outreach clinics. At the time of the study, chemotherapy and associated fees were paid for by the patient. All patients provided written consent for data derived from their care at the participating clinic sites to be used for purposes of research through International Epidemiology Databases to Evaluate AIDS (IeDEA).
Measurements
We ascertained KS disease severity at time of diagnosis and subsequent KS-specific treatment and survival through chart review. HIV/primary care charts, oncology charts, chemotherapy infusion records, and inpatient charts were retrieved for each participant. Paper charts were sought on three separate occasions before being deemed unobtainable. Record abstraction was performed by trained clinical officers using a structured chart review form. Double data entry was performed, and differences were reconciled by the principal investigator (PI).
KS severity at time of diagnosis was based on any symptom recorded within the first 30 days of KS diagnosis. Information on weight loss, diarrhea, night sweats, fever, Eastern Cooperative Oncology Group (ECOG) performance status, extent of cutaneous KS, lymphadenopathy, oral disease, GI KS, pulmonary KS, ulceration, and edema was extracted from the charts. We considered chemotherapy indications to be the presence of at least one AIDS Clinical Trials (ACTG) T1 qualifying criteria (tumor-associated edema or ulceration, extensive oral KS, or non-nodal visceral KS) [
19] or the presence of at least one World Health Organization (WHO) KS Treatment Guidelines “severe disease” qualifying symptom (symptomatic visceral disease, extensive oral KS that interferes with swallowing, painful/disabling tumor-associated edema or ulcerated tumors, or life-threatening/functionally disabling disease) [
12]. Demographic information, CD4 count, and use of antiretroviral therapy was extracted from AMPATH’s electronic database. KS-specific treatment with chemotherapy was ascertained through the chemotherapy infusion record, and charts were additionally reviewed for other forms of treatment such as injections, surgery, or radiation. Survival was initially determined through medical record review and, subsequently, for those who became lost to follow-up, updated through intensive community field tracking, which has been described elsewhere [
3].
Statistical analysis
Time zero was set as the earliest date the clinician had information about the KS diagnosis and therefore could potentially act on it. When a biopsy was performed, the date of the receipt of those results to the clinician (taken to be 10 days after the biopsy was performed accounting for local time needed to process and communicate results of the biopsy) was used for time zero. If a patient had both a clinical diagnosis date and a biopsy diagnosis date, the earlier of the two dates was used to establish time zero. Observation was continued from time zero to the initiation of chemotherapy or death for 12 months. Patients not known to be dead or to have started chemotherapy were administratively censored at last known date alive.
Cumulative incidence of chemotherapy initiation with death as a competing event was calculated using the Aalen-Johansen estimator [
20]. Patients who were still alive at the end of observation time who had not received chemotherapy and had not died were administratively censored at last known date alive. We used binomial log-linear regression [
21,
22] to evaluate the independent association between exposures at time of KS diagnosis (age, sex, KS disease severity, CD4 count, and socio-economic indicators) and the outcome of rapid chemotherapy initiation, defined as chemotherapy initiation within 30 days of KS diagnosis.
Discussion
KS mortality in sub-Saharan Africa is high, which led us to investigate the use of chemotherapy for treatment of KS in this setting. Among patients in a primary care HIV cohort in Kenya, just over half of patients with KS who had an indication for chemotherapy received it by the end of 1 year. Patients with the most severe disease were the most likely to receive prompt treatment within 30 days of KS diagnosis (relative risk 2.30; 95% CI 1.46–3.60). However, a substantial group of patients (44%), who seemingly had an indication for chemotherapy based on the severity of their KS disease, did not receive it within 1 year. Given the high mortality of KS in sub-Saharan Africa [
2,
3], this considerable gap in treatment highlights a potential target for future interventions to improve survival.
Prior studies on KS treatment with chemotherapy in sub-Saharan Africa have focused on outcomes of particular chemotherapy regimens, [
17,
18,
25‐
27] but have neither evaluated real-world delays in chemotherapy initiation, nor, perhaps most importantly, chemotherapy non-initiation in patients with chemotherapy indications. We know that there is often a significant delay between symptom recognition and successful initiation of cancer care in Africa [
28]. Our results for KS echo data from other cancers in Africa, such as lymphoma and breast cancer, with similar delays [
29] and non-initiation of chemotherapy [
30]. A particular strength of our study comes from linking severity of disease at time of diagnosis with subsequent treatment course in a real-world setting.
Similar to other resource poor settings, most patients in this Kenyan setting were diagnosed with KS when their disease was already at advanced stage, and not all patients had histologically confirmed disease [
8,
25,
31,
32]. We found that patients who were on ART for > 60 days prior to KS diagnosis were diagnosed at less advanced KS disease stage, which could either be due to the effects of ART itself (though we do not know who was virally suppressed), or due to the more frequent contact these patients had with the health system. This finding is in line with what has been reported for other co-morbid cancers and HIV [
33,
34], although not universally [
35]. This finding also reinforces that being on ART does not completely eliminate the risk of developing KS [
36,
37].
The most common first line chemotherapy regimen in this setting was bleomycin plus vincristine. Liposomal anthracyclines (e.g., liposomal doxorubicin) and paclitaxel, which are first-line medications in resource-rich settings, were not used as initial therapy in any patient. Liposomal anthracyclines are recommended over BV when available in both the WHO and the National Comprehensive Cancer Network (NCCN) guidelines [
12,
13], although these international guidelines were not yet in place during the time period of the study. Additionally, in sub-Saharan Africa, BV has recently been shown to be less efficacious than paclitaxel (in trial ACTG 5263, which was stopped early) for treating AIDS-related KS [
38]. There has been increasing recognition for the need for highly efficacious chemotherapy in the African setting [
39,
40], although more advocacy is desperately needed to decrease drug costs and improve accessibility. Improvement in the effectiveness of chemotherapy available to patients in low resource settings could improve survival beyond the direct effects of the drug. Specifically, peer to peer evidence of good outcomes may reduce fatalism and encourage patients to start and adhere to therapy, as occurred with ART when it became more widespread [
41].
Patients with more chemotherapy indications (e.g. more advanced disease stage at time of diagnosis) are more likely to achieve rapid chemotherapy initiation within 30 days of diagnosis. That the most ill patients are being prioritized for chemotherapy is reassuring. The lack of impact of socioeconomic factors on chemotherapy initiation was initially surprising, since we hypothesized that patients who reside farther from clinic with less education and fewer resources would have a harder time accessing care. There are several possible reasons for this observation. This lack of effect may be due to the relative homogeneity of this population of care seekers, who are overwhelmingly economically disadvantaged and have at most a primary school education, where small variations in socioeconomic status may not impact access to chemotherapy. It is also possible that we need more sensitive socio-economic measures to delineate this effect. Alternatively, the community outreach programs of AMPATH, where chemotherapy is supplied to two other community clinics outside of the main hospital, may have helped to mitigate the effect of socioeconomic status on access to chemotherapy.
This study had several limitations. First, we identified people only after they had been diagnosed with KS. Therefore, we did not account for patients in the community who either did not seek care, or sought care but did not receive a formal KS diagnosis. Therefore, it may overestimate the true proportion of KS patients that received chemotherapy. Second, we were unable to locate paper charts for 13% of patients over the study period who were diagnosed with KS. These patients were excluded from the study since we were unable to ascertain whether or not they had received chemotherapy. Thirdly, severity of KS at time of diagnosis was limited to what clinicians recorded in the patients’ charts during routine clinical care. Record keeping was often incomplete. Absence of a symptom (e.g. a pertinent negative) was rarely recorded. Therefore, when a particular symptom was not documented, we made the assumption that the patient did not have that symptom. It is possible, therefore, that we underestimated the severity of disease at diagnosis. In particular, WHO severe disease was likely to be underestimated, since it was not common practice to record whether the disease was affecting the patient’s functional status.
Due to the retrospective nature of this chart review and incomplete paper records, we were not able to ascertain reasons that eligible patients did not start on chemotherapy. Literature suggests that for other cancers in sub-Saharan Africa affordability [
42,
43], transportation, and fear, all may contribute to non-initiation [
44], but prospective evaluation is needed to understand the differential contributions of these patient-level factors. Similarly, patients’ adherence to ART may influence the decision to start chemotherapy therefore should be investigated in future work as well. In addition, factors related to healthcare providers, rather than the patients themselves, such as provider non-recognition of disease severity, limited training, difficulty in obtaining pre-treatment tests, conflicting guidelines, or decisions made to prioritize end of life quality over treatment initiation [
45‐
47], could all factor in to provider’s decisions not to start a patient on chemotherapy. In order to develop interventions to improve chemotherapy initiation in this setting, additional prospective evaluation of both these patient and provider level factors is needed.
Regarding generalizability, this Kenyan cohort consists of the general population receiving HIV primary care in a resource-poor sub-Saharan Africa setting. However, the AMPATH program does have access to enhanced oncology services, including a public cancer center with routine outreach clinics and an IeDEA-sponsored KS biopsy service. Therefore, both diagnosis and treatment in the setting may be enhanced compared to other public sector settings elsewhere in sub-Saharan Africa [
17].
Conclusion
In conclusion, we found that 44% of patients with KS in this Kenyan setting who may have needed chemotherapy at the time of diagnosis did not initiate this treatment. The subset of patients who successfully initiated chemotherapy did, for the most part, receive it in a timely manner. Due to the retrospective nature of this study, we could not further elucidate the reasons for treatment non-initiation, or distinguish between patient and provider factors. Prospective evaluation is needed to assess barriers to chemotherapy from both patient and provider perspectives, so that appropriate interventions can target treatment initiation. The chemotherapy regimens used at the time of the study, and what is still routinely available in Kenya, are not equivalent to liposomal anthracyclines, which are first line in resource-rich settings. International advocacy for more efficacious, affordable, and accessible chemotherapy in sub-Saharan Africa is critical, and improved outcomes could encourage other patients to seek out and initiate chemotherapy.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.