HIV Brain Conditions Split Into Two Distinct Patterns in Ugandan Cohort

People living with HIV who attended a specialist neuropsychiatric clinic in Kampala showed two clearly separable brain health profiles, according to a comorbidity network analysis published in AIDS Research and Therapy. In an acute phenotype, opportunistic infections of the central nervous system dominated, while a chronic phenotype brought together depressive, bipolar and psychotic presentations. The study, based on routinely collected diagnostic records from 2017 to April 2026, also found that one of the most widely assumed comorbidities in HIV, the pairing of cognitive impairment with major depressive disorder, showed no evidence of an association in this cohort.

The research team analyzed 1,137 ICD-10-coded records from 581 patients in the specialist HIV neuropsychiatric cohort in Kampala. With expert clinical consultation, every ICD-10 code was assigned to one of eight brain health domains and to a two-level phenotype distinguishing acute from chronic presentations. Among 573 patients whose records could be classified, the investigators tested pairwise co-occurrence between domains using phi coefficients, bootstrap confidence intervals and Benjamini-Hochberg correction for multiple testing. Latent class analysis was pre-specified as a sensitivity analysis.

Multimorbidity was the exception rather than the rule in the classified records. Of the 573 classifiable patients, 476, or 83.1 percent, carried a single brain health domain, and only 12 patients, or 2.1 percent, carried diagnoses from both the acute and chronic phenotypes. This limited recorded overlap suggests that, at least as documented in this clinical setting, the two phenotypes rarely coexist in the same patient record.

The network structure itself was strikingly polarized. Depression, bipolar and psychosis spectra formed a connected cluster, with phi coefficients ranging from 0.11 to 0.25, indicating modest positive co-occurrence among these chronic conditions. In contrast, opportunistic central nervous system infection was negatively associated with each chronic domain tested, with phi coefficients as low as -0.734. A negative association of this magnitude means that patients recorded with an opportunistic CNS infection were very unlikely to also carry diagnoses in the chronic domains, and vice versa.

Cognitive impairment told a different story. It showed no evidence of an association with major depressive disorder: only 12 patients carried both domains, and the phi coefficient was -0.064 with a corrected q value of 0.203, well above any conventional threshold for statistical significance. The authors note that this depression-cognitive impairment comorbidity is widely assumed to be common in HIV, yet their data from this Ugandan cohort provided no support for that assumption.

The pre-specified sensitivity analysis partly qualified the picture. Latent class analysis identified a multimorbid chronic phenotype accounting for 24.6 percent of patients, indicating that a substantial subgroup did carry multiple chronic brain health conditions even though most classified patients had a single domain. The convergence of the network and latent class approaches on a chronic, multimorbid group strengthens the case that the chronic phenotype is internally cohesive.

The authors conclude that brain health conditions in this cohort show limited recorded overlap between two clinically defined phenotypes: an acute opportunistic-infection phenotype and a chronic phenotype in which depressive, bipolar and psychotic presentations co-aggregate.

Several limitations frame these findings. The analysis reflects diagnoses recorded in a single specialist clinic in Kampala, so the results may not generalize to other settings or to care delivered outside specialist neuropsychiatric services. The co-occurrence measures capture recorded diagnoses rather than systematically screened symptoms, and domains with fewer than ten patients were excluded from pairwise testing. The article is being shared early as a peer-reviewed, accepted manuscript carrying a permanent DOI, and this version remains subject to further edits before the final Version of Record replaces it.

For clinicians and researchers in sub-Saharan Africa, where individual brain health conditions among people living with HIV are increasingly well documented but multimorbidity has remained poorly characterized, the study offers a data-driven map of how these conditions actually cluster.

Subject of Research: Comorbidity network analysis of brain health conditions among people living with HIV in a Ugandan clinical cohort

Article Title: Distinct acute and chronic brain health phenotypes among people living with HIV in a Ugandan clinical cohort: a comorbidity network analysis

Article References: Mwaka, J., Dastgheyb, R. M., Sserunjogi, C., Nataliya, J., Mulebeke, R., Karamagi, Y., & Rubin, L. H. (2026). Distinct acute and chronic brain health phenotypes among people living with HIV in a Ugandan clinical cohort: a comorbidity network analysis. AIDS Research and Therapy. https://doi.org/10.1186/s12981-026-00956-4

Image Credits: AI Generated

DOI: 10.1186/s12981-026-00956-4

Keywords: HIV, brain health, comorbidity, network analysis, Uganda, neuropsychiatry, depression, cognitive impairment, opportunistic CNS infection, latent class analysis, ICD-10, multimorbidity