Background:  Following long-term HAART, many patients show virological failure in plasma. Because of poor penetration of some anti-HIV drugs in brain tissue, there is concern that virological failure might occur earlier and more importantly in cerebrospinal fluid (CSF) than in plasma. The objective of this study was to assess the proportion and magnitude of virologic failure in CSF in relation to plasma in long-term HAART-treated patients.

Methods:  HIV-1 RNA load was prospectively measured in CSF and plasma from 267 patients with neurological symptoms in whom CSF was drawn for diagnostic purposes between January 1997 and December 2003 at the Infectious Diseases Clinic of San Raffaele Hospital, Milan. Only patients who received HAART continuously or were untreated for at least 6 months before sampling were included in the analysis. Univariate analysis was performed by χ² test to compare discrete variables and Wilcoxon 2-sample rank test to compare mean independent values. Multiple logistic regression was used to assess the independent contribution of potential risk factors on CSF virological failure.

Results:  Data from 128 of the 267 subjects were included, of whom 61 were on HAART and 67 were untreated at the time of sampling. Based on cut-off level of 2.60 HIV-1 RNA log copies/mL in plasma, 24 treated patients were classified as HAART-success and 37 as HAART-failure. Median duration of HAART did not differ between HAART-success (median 642 days; IQR, 336 to 1247) and HAART-failure (median, 510 days; IQR, 324 to 1107). CSF HIV-1 RNA levels were > 2.60 log copies/mL in 56 of 67 HAART-untreated (84%), 1 of 23 HAART-success (4%), and 18 of 37 HAART-failure (49%) (HAART-untreated vs HAART-failure, p = 0.001). CSF HIV-1 RNA levels were lower in HAART-failure than HAART-untreated (median:  2.60 vs 4.00 log copies/mL, p < 0.0001). Also plasma HIV-1 RNA levels were lower and CD4 cell counts higher in HAART-failure than HAART-untreated (plasma HIV-1 RNA: median, 4.30 vs 5.36 log copies/mL, p < 0.0001) (CD4:  median, 109 vs 50/µL, p < 0.0001). However, in a multivariate model including only patients with plasma HIV-1 RNA > 2.60 log copies/mL, the risk of virologic failure in CSF was higher in HAART-untreated than HAART-failure (OR=1.86, 95% CI = 1.12 to 3.17, p = 0.018), but it was neither associated with higher HIV-1 RNA levels in plasma nor with lower CD4 cell counts.

Conclusions:  HAART seems to maintain its efficacy on intrathecal virus replication in long-term treated patients irrespective of poor systemic viro-immunological responses.

Keywords: cerebrospinal fluid; haart; virological failure