Background:  Progressive multifocal leukoencephalopathy (PML) is still associated with high mortality despite the use of HAART. It is not clear whether poor PML outcome is associated with failure to respond to HAART, HAART-induced immune reconstitution, or both. To address this question, we investigated the relationship of PML survival with type and response to HAART in a selected group of patients who started HAART after PML diagnosis.

Methods:  We evaluated 48 patients with PML diagnosed between 1996 and 2006 (confirmed by brain biopsy or JC virus [JCV] DNA in CSF = 42; probable, based on typical clinical and neuroimaging findings = 6) who were either ART naive (n = 28), ART untreated for ≥6 months (n = 11) or taking 1 or 2 nucleoside analogs at onset of PML (n = 9). Patients started HAART 42 (14 to 42) (median [IQR]) days following PML onset (baseline). Survival curves (Kaplan-Meier method) were estimated according to PML onset variables (age, sex, CD4 and CD8 cell counts and percentage, CD4/CD8 ratio, plasma and CSF HIV RNA level, CSF JCV DNA level, HCV antibody status) and during follow-up (3 months after PML onset, use of protease inhibitor, changes in CD4 and CD8 cell counts and percentage, CD4/CD8 ratio, and plasma HIV RNA level). In this analysis, quantitative variables were stratified according to their median values. Multivariate analysis (Cox model) was performed including as covariates the baseline values of CD4 (< vs ≥86/µL), CD4% (< vs ≥10.5), CD8 (< vs ≥ 630/µL), CD8% (< vs ≥ 62.3), viral load (< vs ≥140,000 copies/mL), and protease inhibitor use (yes vs no). Because all PML-related deaths occurred within 18 months from PML onset, all the variables were correlated to the 2-year survival.

Results:  Survival from onset of PML was 59% at 2 years and 34% at 8.7 years. The 2-year survival probability was lower in patients with baseline CD4% <10.5 (28% vs 75%, p = 0.04). Higher survival was associated with larger increases in both CD4 cell counts (92% vs 50% for ≥ vs <32/µL, p = 0.012) and CD4% (100% vs 62% for ≥ vs <3.25%, p = 0.010), respectively, at 3 months compared to baseline. No association was found with the other variables either considered at baseline or during follow-up. At multivariate analysis only protease inhibitor turned out to independently reduce the risk of mortality at 2 years (HR = 0.410, 95%CI = 0.165 to 1.017, p = 0.041), whereas baseline CD4% <10.5 was associated with higher risk showing a trend toward significance (HR = 2.978, 95%CI = 0.927 to 9.568, p = 0.066).

Conclusions:  These findings indicate that favorable PML outcome is mostly associated with good response to HAART. Use of protease inhibitor in combination regimens might influence this effect.