Background:  Nucleoside reverse transcriptase inhibitors (NRTI) are often administered in salvage therapy even if genotypic resistance tests (GRT) indicate high-level resistance. The clinical benefit of these additional NRTI is unknown.

Methods:  The effect of <2 compared to 2 NRTI on viral suppression (HIV-1 RNA <50 copies/mL) at week 24 was studied in salvage patients from the Swiss HIV Cohort Study (SHCS) who received raltegravir (RAL). Inclusion criteria were a GRT prior to RAL start and a HIV-1 RNA >500 copies/mL. Intent-to-treat and per-protocol analyses were performed. Because patients with a better resistance profile and more remaining drug options tended to receive fewer NRTI (i.e., factors associated with better outcome), we performed a weighted analysis (marginal structural model). Weights were defined as the inverse of the probability for receiving <2 NRTI. This method creates a pseudo-population, in which the probability for receiving <2 or 2 NRTI is unrelated to baseline factors prognostic for treatment response. To analyze the outcome, a weighted logistic regression was performed adjusted for age, sex, risk group, ethnicity, cumulative genotypic sensitivity score (GSS) of the regimen (without NRTI, calculated with Stanford algorithm 6.0.8), number of drug classes, baseline RNA, and CD4 cell count.

Results:  One hundred and thirty patients were included, of whom 58.5% (n = 76) received <2 NRTI. NRTI were often replaced by other drug classes, percentages with ≤1, 2, 3 additional classes were 68.5%, 27.8%, 3.7% (2 NRTI) and 34.2%, 55.3%, 10.5% (<2 NRTI, p-exact <0.001). The activity of non-NRTI treatment components was lower in the 2 NRTI group (median [IQR] GSS: 2 [1.5 to 2.5]) compared to the <2 NRTI group (2.5 [2 to 3], p-Wilcoxon <0.001). The median contribution of each NRTI to the overall GSS was 0.5 in both groups (p = 0.134).The administration of <2 NRTI was associated with a worse viral suppression rate. Weighted OR were 0.34 (p = 0.027) and 0.19 (p = 0.027) when performing a last observation carried forward (LOCF) and a missing equal failure (m = f) approach to deal with missing information. The per-protocol analysis yielded similar results (OR 0.19, p = 0.023). A subanalysis including only patients with GSS NRTI ≤0.5 in the regimen confirmed our findings.

Conclusions: Our findings suggest that even partially active or inactive NRTI contribute to treatment response, and thus NRTI with partial activity may still play an important role in salvage therapy.