Spectrum and immunovirological determinants of tumours among people living with HIV in Uganda: A retrospective cross sectional study from a specialised HIV centre, 2017-2026.

backgroundSub-Saharan Africa carries a disproportionate burden of HIV-associated malignancies. The contemporary tumour spectrum in the dolutegravir (DTG) era remains incompletely characterised in East African routine HIV-care settings. The immunovirological context at the time of cancer or benign tumor diagnosis is similarly under-described. We sought to describe the spectrum, malignant fraction, and immunovirological correlates of neoplasms diagnosed at a specialised HIV centre in Kampala, Uganda.

methodsWe conducted a retrospective, cross-sectional analysis of 219 tumour records identified within the longitudinal clinical cohort of people living with HIV (PLHIV) in continuous care at Mildmay Hospital, contributed by 216 individual patients (three of whom each had two separate tumour diagnoses on different dates) between November 2017 and January 2026. Tumours were classified by ICD-10 disease group and malignancy status (benign/malignant). Bivariate associations with malignancy status were tested using χ² or Fisher exact tests for categorical variables and Mann-Whitney U tests for continuous variables (two-sided α = 0.05), with Benjamini-Hochberg false discovery rate (FDR) correction applied across each panel of comparisons given the number of variables tested. Crude odds ratios (OR) with 95% confidence intervals (CI) were computed for dichotomous comparisons. A sensitivity analysis restricted to one tumour record per patient (n = 216) was performed to assess the influence of the three patients with two records each.

resultsThe cohort had a median age of 47 years (IQR 39-54) and was 127/219 (58.0%) female. Overall, 138/219 tumours (63.0%) were malignant. Kaposi sarcoma (KS) was the most frequently registered malignancy (n = 65; 47.1% of all malignant tumours; 29.7% of the full cohort). Among 155 records with viral load data, 133 (85.8%) were virally suppressed at tumour diagnosis, yet 70/133 (52.6%) of those suppressed records were malignant. Advanced WHO clinical stage (III/IV) at ART initiation was associated with malignant diagnosis (OR 3.51, 95% CI 1.54-8.81; FDR-adjusted p = 0.003). Median CD4 at ART initiation was lower among malignant compared with benign tumour records (147 vs 476 cells/µL; FDR-adjusted p = 0.001), and median ART duration before diagnosis was shorter (37 vs 101 months; FDR-adjusted p < 0.001). Restricting the analysis to KS versus other malignancies only (n = 138), KS patients had lower CD4 at ART initiation (136 vs 171 cells/µL, p = 0.079) and markedly shorter ART duration before diagnosis (4 vs 63 months, p < 0.001) than patients with other malignancies. All associations were materially unchanged in the patient-level sensitivity analysis.

conclusionsHIV-associated tumours in this Ugandan cohort remain predominantly malignant and are dominated by KS despite high viral suppression rates. Malignancy clusters among patients with low pre-ART CD4 counts, advanced WHO stage, and short ART duration, though these are unadjusted, hypothesis-generating comparisons rather than evidence of causal or temporal gradients. These data establish the analytic platform for a prospective HIV-oncology cohort at Mildmay Hospital and Mildmay Research Centre and inform context-appropriate cancer screening priorities.