Analyses using COHERE data previously suggested (Clin Infect Dis 2010:51:611) that primary Pneumocystis Pneumonia (PcP) prophylaxis could be withdrawn in patients with CD4 counts of 100-200 cells/µL if HIV-RNA is suppressed, suggesting HIV replication as major risk factor for PcP. Given the wealth of new data available in COHERE we investigated whether prophylaxis might be withheld or stopped in all patients on antiretroviral therapy with suppressed plasma HIV RNA (<400c/mL) irrespective of CD4 count.
We estimated the risk of primary PcP in COHERE patients on cART including time-updated CD4 counts, HIV-RNA and use of PcP prophylaxis. We emulated a hypothetical randomised trial using established causal inference methods in which inverse probability (IP) weighting adjusts for censoring selection bias. Eligibility criteria were plasma HIV RNA (<400c/mL) and CD4 counts ≤200 cells/µL. We emulated three trials comparing the effect of A.) continue PCP prophylaxis versus stop prophylaxis, B.) start and then continue PCP prophylaxis versus not starting prophylaxis, and C.) taking PCP prophylaxis versus not taking PCP prophylaxis, irrespective of PCP prophylaxis status at baseline. In each case, we estimated the hazard ratio (HR) fitting a pooled logistic model which included baseline characteristics (CD4, RNA, gender, age, transmission, geographical origin calendar year), used restricted cubic splines to capture CD4/RNA trajectories, and included polynomial time for modelling the baseline hazard.
There were 9,743 patients eligible for the emulated trials with a total of 18,550 person years followed-up during 1998-2015. The unadjusted incidence rate of PCP diagnosis was 1.5 per 1000py on PCP prophylaxis compared to 2.8 off PCP prophylaxis. The HR estimates for the PCP outcome from the 3 emulated trials were 2.0 ([0.61 6.4], p=0.3) for Trial A, 2.8 ([0.8 9.9], p=0.1) for Trial B, and 1.2 ([0.5, 3.2], p=0.8) for Trial C (see Figure).
In virologically suppressed patients, irrespective of CD4 levels, the risk of PcP appears to be low, and similar for individuals on and off prophylaxis, although the precision of the results was limited due to the overall low incidence of PCP. This suggests that primary PcP prophylaxis might be withheld in this patient group.