Background: Current CD4 count predicts mortality in patients treated with combination ART (cART). It is unclear whether, in patients treated for some years, there is additional prognostic value in CD4 measurements at initiation of treatment.

Methods:  Data from 15 HIV cohort studies in Europe and North America were combined. Eligible patients were aged >15 and ART-naïve when they started cART before 2002. We used Cox models to estimate mortality hazard ratios (HR) from 36 months after initiation of cART according to CD4 at 0, 6, and 36 months, among patients still followed up at this time.

Results:  CD4 at 36 months was measured in 17,159 patients (median 482/mm³, IQR 320 to 683). Median 36-month CD4 was 279, 319, 334, 400, 512, 639, and 780 in patients starting cART with 0 to 24, 25 to 49, 50 to 99, 100 to 199, 200 to 349, 350 to 499, and ≥500/mm³, respectively. The table shows mortality HR from 36 months, according to CD4 at the 3 time points. CD4 at 36 months was highly prognostic for subsequent mortality. In unadjusted models, patients with baseline CD4 <200 cells/mL had higher mortality from 36 months, and CD4 at 6 months was moderately prognostic. In the adjusted model, CD4 at 6 months was no longer prognostic. Given patients’ CD4 at 36 months, there was a negative association with CD4 at initiation because, given CD4 at 36 months, patients with higher baseline CD4 will tend to be those whose rise in CD4 over 36 months is lower because of non-adherence or treatment failure.

Conclusions:  CD4 count at 36 months is strongly related to CD4 count at initiation of cART. At 36 months after starting cART, current CD4 count is strongly prognostic for subsequent mortality. Additionally, patients whose CD4 increase is below average have worse prognosis.