Abstract Body

Several risk factors for cerebrospinal fluid HIV-1 escape (CSFE) have been reported: length of HIV infection, cART interruptions, low CD4 nadir/CPE score, persistent low-level viremia and the use of ABC+3TC, boosted PIs or unboosted ATZ. We sought to assess whether the presence of previous plasma RAMs may be a determinant behind the reported risk association between CSFE and ARVs class composing cART

Retrospective cross-sectional study on HIV+ adult patients on cART undergoing lumbar puncture (LP) for any reason (2007-July 2019) at 4 Italian hospitals (Brescia, Torino, Roma, Milano). Inclusion criteria: being on cART for at least 6 months, available coupled plasma and CSF HIV-RNA measurements, available historical cumulative plasma genotypic resistance testing (HGRT) for reverse transcriptase (RT) and protease (PI) genes. Exclusion criteria: secondary CSFE. CSFE was defined as any measurable CSF HIV-RNA coupled with a plasma HIV-RNA <50 copies/mL and any difference ≥0.5 Log10 between CSF and plasma HIV-RNA when the latter was detectable

197 patients were enrolled: 50 years (43-54), current and nadir CD4 count 312 (115-560) and 82 (24-200) cells/μL; median length of cART treatment 54 months (17-171). 126 patients (63.9%) had plasma HIV-RNA<50 cp/mL and 28 (14.2%) showed CSFE. The main reasons for LP were diagnostic assessment in diseases without eventually CNS involvement (25.4%), HIV-associated neurocognitive disorders (28.4%), CNS infections (19.8%) and research purposes (16.2%). CSFE was not associated with PIs use in the whole cohort (16.6% vs 8.6%, p.14) nor in any subgroup identified by cART type (3 different-classes-, 3 drugs-NRTIs- and ≥4 drugs-based cART). Instead, PIs use was more common in patients with a positive HGRT for RAMs in RT (44.9% vs 29.3%, OR 2.0 [1.1-3.8], p.04). Having a cumulative HGRT positive for RAMs in RT associated with a higher risk of CSFE (21.5% vs 9.3%, OR 2.7 [1.2-6.0], p.01), while no such an association was observed for RAMs in PI (17.4% vs 13.8%). Interestingly, at the CSFE diagnosis patients showed higher proportion of positive CSF RAMs in RT compared to patients without CSFE with available CSF GRT (55.6% vs 19.0%, p.04). At multivariable analysis, only RAMS in RT and CD4 nadir were independent predictors of CSFE (tab.1)

In this cohort, CSFE prevalence was slightly higher than what reported in recent studies. Besides low CD4 nadir, the positivity of HGRT for plasma RAMs in the RT gene and not the use of PIs per sé was an independent predictor of CSFE