In our review of outcomes of adult and adolescent NHL cases in Uganda, we found that NHL is a morbid condition with discrete predictors of survival in this resource-limited setting. Specifically, HIV infection in the absence of ART, female gender, anemia, and constitutional (“B”) symptoms at presentation were all independent predictors of mortality.
In general, we observed a low one-year survival among patients in our population with NHL. We found that several of the previously described prognostic factors for NHL in resource-rich settings were not predictive of survival in our cohort.39–41
Factors such as tumor histopathology, international prognostic index, presence of extra nodal disease, immunohistochemistry, or cytokine analyses were not routinely or reliably assessed in our cohort or other resource-limited settings,9
and this could explain why they were not found to be predictive of survival. Additionally, the amount of missing data in our cohort may have precluded finding significant associations between these factors and survival. Finally, the diagnostic and supportive care, as well as treatment regimens, available to patients in Uganda differs substantially from what is available in the United States. The unavailability of rituximab, G-CSF, and extended spectrum antimicrobials may result in different factors being more important to patient survival in Uganda. Alternatively, it is possible that the pathogenesis of NHL in Uganda differs to some extent from the disease in the United States, a hypothesis which is currently being investigated in translational studies of NHL tumors in Uganda. Differences in prognostic factor measurements are unlikely to completely explain the observed differences, as risk factors which are simple to measure, such as LDH and age, were not predictive in this cohort.
Future prospective studies are needed to assess the factors predictive of NHL survival in resource-limited settings more definitively, and examine others such as late presentation, nutritional status, access to supportive care, and socioeconomic status.
Our cohort had limited access to chemotherapy, ART, and other supportive therapies and had poor survival. The median survival was only 2 months (61days) (IQR 25–203) in those determined to have died. This is significantly lower than for HIV-infected cohorts in the pre-ART era in resource-rich settings era39,41
and even lower compared to more recent cohorts in Europe.25
We also found that the receipt of chemotherapy, which has been shown to be associated with a survival advantage in both HIV-positive and HIV-negative patients in resource-rich settings, was not significantly associated with survival among patients in Uganda. However, these data should be interpreted with caution in this retrospective study. In such a design, patients needed to survive and remain in care for more than 4 months in order to receive all courses of chemotherapy. Patients with advanced NHL, or NHL and HIV, may have been too ill at presentation to receive, or tolerate, chemotherapy. To this point, we found that more than a quarter of patients did not receive chemotherapy. And even among those who received some, less than three quarters (72.6%) received more than 2 courses.
We acknowledge that our chosen method of treating chemotherapy as a time-varying covariate may have minimized but did not fully eliminate the bias of assuming constant harzards and using the Cox regression. Patients who received more courses of chemotherapy may have been healthier at the time of initiating chemotherapy than patients who received less, who were likely to die prior to scheduled treatment. Additional studies are needed to determine the optimal treatment strategy for patients presenting with advanced NHL in resource-limited settings.
HIV was a common co-morbidity to NHL in this cohort, with a prevalence of 31.9% among study subjects with known HIV serostatus. This prevalence is much higher than the national prevalence of HIV in Uganda (7%)42,43
reflecting observations else where37
and in Uganda and other places.8
Given the high prevalence of HIV among patients with NHL in Uganda, our study made important observations about the role of HIV and its treatment in the prognosis of NHL. We found that HIV infection alone may not be associated with a poorer outcome in Ugandans with NHL. More to the point, we found that patients with HIV infection who were receiving ART had survival rates which approximated those of HIV-negative persons. These results have important ramifications for the management of HIV-associated malignancies in resource-limited settings, showing that the appropriate co-management of HIV and cancer can result in improved outcomes. These results, however, require additional evaluation, as a number of factors could have potentially confounded the observed beneficial effects of HIV therapy. For example, it is possible that persons who received ART were in better health at baseline, were better able to start and tolerate chemotherapy, or differed from HIV-infected patients who failed to receive ART based on their socioeconomic status. Also, as noted earlier, we cannot assume that the association between ART and survival was the same for patients who remained in care and patients who were lost to follow-up.
This study has other limitations, in addition to missing baseline data and loss to follow-up. Construction of cause-specific mortality estimates was not possible based on available data, and we cannot assume that all deaths among cancer patients were entirely due to cancer. That assumption would be particularly faulty in patients infected with HIV, because they are also at risk of death from opportunistic infection or ART side effects. Diagnostic challenges precluded identification of some opportunistic infections in HIV-infected patients and in many study subjects, the characterization of NHL into types and subtypes. Similarly, missing data precluded description of reasons for causes of treatment delays, information which is probably of prognostic significance. Also, information on adherence to ART, ART interruption during chemotherapy administration, and the incidence or severity of chemotherapy- or ART-related side effects was not consistently collected or graded.
Despite these limitations, our data, although retrospectively collected from a single institution, describe a cohort of patients from an area that has one of the best cancer registries in sub-Saharan Africa, and represents the first attempt to study HIV and survival rates of patients with NHL in this setting.
The study findings have several implications both for research and for practice. Future studies should prospectively describe factors associated with late presentation for AIDS and NHL treatment, failure to initiate chemotherapy or to complete chemotherapy once initiated, and loss to follow-up. Such studies would additionally determine optimal treatment strategies for AIDS-NHL in the era of increasing but still limited access to ART, and identify prognostic factors that are unique to these resource-limited settings.
Clinical practice should be improved through the design and implementation of comprehensive cancer plans, including plans for the management of concurrent cancer and HIV/AIDS. Such plans would allow for the more effective use of extremely limited funds for cancer control and treatment. Better strategies to improve the awareness, early diagnosis, referral and treatment of cancer in resource-limited settings will make meaningful impacts on the health of the overall population.