Objectives: To assess prevalence and predictive factors of viro-immunological discordant trends in a cohort of heavily pretreated patients.
Methods: Factors associated with viro-immunological discordant trends either as categorical or continuous measures have been studied in 159 heavily pretreated HIV-positive patients from a multicentre prospective study of real- vs. virtual-phenotype. Univariate and multivariate logistic regressions were used to assess risk factors for categorical discordant responses, ceasing follow-up at week 32 since enough patients had been on the original drug combination for a sufficient amount of time to evaluate their immune response. Complementary linear regression analysis was performed over the entire 48 weeks' follow-up considering CD4 and plasma viral load (pVL) as continuous measures.
Results: Among 58 virological responder patients (> or =1 log10 HIV-1 RNA copies/mL decrease) and 101 virologically non-responders, immunological discordances (increase in CD4 count of< or > or =100 cells/microL) were observed in 58.6% and 38.6%, respectively. Baseline CD4 count was associated with discordant responses in both groups. Multivariable linear regression over the entire 48 weeks' follow-up demonstrated significant correlation between absolute decrease in pVL and increase in CD4 count (HR 28.06, 95%CI 35.32-20.79; P<0.001), also the use of protease inhibitors (PIs) in the salvage regimen (HR 36.57, 95%CI 15.45-57.68; P<0.001) and >8 months on treatment (HR 41.64, 95%CI 19.27-64.01; P<0.001) correlated with highly significant immune recovery.
Conclusions: These data confirm that therapy, possibly including PIs, should be continued in heavily pretreated patients and that hard-to-reach pVL undetectability is not essential to obtain immunologic recovery; however, this is strongly increased by the degree of pVL reduction that should be achieved.