This study is one of the few looking at patient outcomes in HIV-associated KS, in the era of widespread antiretroviral therapy in the developing world. We found that the combination of HAART and vincristine therapy is associated with an impressive response rate, although there were no reports of complete response in our study. Liposomal doxorubicin is considered the standard of care in the developed world on the basis of two large randomized clinical trials [14
]; however, it is not only expensive, but can be associated with myelosuppression and gastrointestinal side effects which would be difficult to manage currently in our setting. Northfelt et al. reported 1 complete response and 60 partial responses out of 133 patients who received liposomal doxorubicin, for an overall response rate of 45.9% in their study [14
]. Median time to response was 39 days. The highly favourable response rate in our series is likely explained by a combination effect of vincristine, along with immune reconstitution associated with HAART.
The difficulties in distinguishing the antitumour effects of HAART from the effects of chemotherapy have been previously described [16
]. The FDA has previously suggested that patients should receive an extended period of HAART prior to being considered for chemotherapy. The Malawian guidelines, for the use of vincristine described above, are consistent with this suggestion. In our study, 74% patients were receiving HAART for a median of 6.8 months prior to commencing vincristine, which was consistent with these guidelines.
There are few studies looking at the use of chemotherapy for HIV-associated KS in Sub-Saharan Africa. Olweny et al. [17
] report on a randomized study of 495 patients in Zimbabwe comparing two chemotherapy regimens, radiotherapy and best supportive care. No complete responses were seen, however single agent etoposide was associated with improved quality of life, and a partial response rate of 31%. Combination chemotherapy (actinomycin-D, bleomycin, vincristine) was associated with a response rate of 49%, however with considerable toxicity (alopecia, mucositis, nausea/vomiting). No patient on this study received HAART. Bihl et al. report on a small South African study of 33 patients, randomized to HAART alone or HAART in combination with combination chemotherapy (doxorubicin, bleomycin, vincristine) [18
]. Both arms produced a profound decline in KSHV viraemia and improved CD4 counts. 16 patients received HAART and chemotherapy, with 7 (44%) achieving a complete response and 7 (44%) achieving partial response when measured at 11 months, however the authors acknowledge considerable toxicity of this regimen.
In our study, the combination of vincristine and HAART was generally well tolerated, although peripheral neuropathy was a dose-limiting factor in some patients. We found a baseline peripheral neuropathy rate of 50%. The prevalence of HIV-related peripheral neuropathy varies in different published series from 9–63%. References [19
], with higher rates seen in those with more advanced immunosuppression and those on HAART. A similarly high incidence of neuropathy was reported in a large cross-sectional Ethiopian study, with the majority of cases found in those receiving HAART [24
]. Likely contributing factors in our series were stavudine-containing HAART, prior tuberculosis treatment, and HIV neuropathy. Peripheral neuropathy is a particular concern in this patient population, causing significant morbidity with a limited choice of agents available for symptom management.
16 patients died during the six-month study period, including 6 early deaths that occurred within 6 weeks of recruitment. The median PFS in the ITT analysis was 30 weeks in our study, which is comparable with survival data reported by Olweny et al. [17
]. Additional analysis presuming all defaulted patients who had died revealed a median progression-free survival of 24.5 weeks. This emphasizes the need for taking a holistic approach to patient management, aiming to improve quality of life for those in whom life expectancy is limited. Cause of death was difficult to ascertain in most cases given the majority of patients died at home. There was a trend towards a lower baseline CD4 count in those who died during the study. Given no clear association between lower baseline CD4 counts and KS progression, it is possible that opportunistic infections may have contributed to deaths more than progressive KS during the study period. Immune reconstitution inflammatory syndrome (IRIS) is well described in KS [25
] and is another potential cause for deterioration in patients with KS receiving HAART. It is of interest that the baseline CD4 count of 263 in our study is somewhat higher than previously published. In similar cohorts [17
], further study regarding the causes of death in this group would be of interest.
In this study we rigorously attempted to contact patients who failed to attend appointments firstly by telephone and then by home visits. It is common cultural practice in Malawi to travel home to the family village as people near the end of their life. This may have contributed to at least 3 patients being lost to follow up. 3 others were known to have potentially life threatening conditions at the point of last contact—one with a large cerebrovascular accident, one with Stevens-Johnsons-syndrome thought secondary to HAART, and one with proven gastrointestinal KS. A prior study of patients on an antiretroviral programme in Lilongwe, Malawi, found that of those lost to follow-up, 41% were later found to have died [27
]. It is likely that death was the cause of several defaults in this study.
Due to lack of access to prompt histopathological services, the diagnosis of KS in this study was made clinically by two experienced physicians. Histological confirmation would have been ideal if feasible.
The APCA POS tool has been validated for use in an African setting. We utilized it in this survey in an attempt to measure palliative care outcomes. Whilst some trends could be elucidated, we observed that the patients and families in our study found the questionnaire difficult to complete. They struggled to use numerical rating scales to express the magnitude of a subjective experience, even with the assistance of a research nurse. More refinement in such assessment tools is needed before they can be used reliably in our population group.