Abstract Body

Background: Viremia copy-years (VCY), a measure of cumulative HIV burden approximated by the area under a patient’s viral load (VL) curve (AUC), was shown to predict mortality independently of VL and current CD4 count in ART-treated patients. For a given AUC, its shape (e.g. a VL of 10,000 copies/mL for 1 year) might help identifying patients (pts) whose VL/ART history differ from that of pts with other shapes (e.g. a VL of 1,000 copies/mL for 10 years).

Methods: All pts who were enrolled in the Icona Foundation study and who started cART after January 1, 2000 were included. VCY (log10 scale) was calculated using the trapezoidal rule. Cox proportional hazards regression model was used to estimate the relationship between VCY and two endpoints (AIDS/death and severe non-AIDS[SNAE]/death) with time accruing from ART initiation. Multivariable models were constructed both controlling for CD4 count as time-updated covariate and using inverse probability weights (IPW). Pts were classified according to the quartiles of the proportion of AUC over the maximum rectangular with base the length of VL follow-up and height pts’ ever observed VL peak under ART. Roughly, a proportion of 100% identifies patients with stable VL trajectory at peak level while low percentages people with dips and spikes in VL.

Results: We included 5,376 pts in Icona with the following characteristics: 36% MSM, 84% of Italian nationality with median (range) age of 37 (18-78) years, who started cART on average in 2010, 54% started PIr-based cART. By the end of follow-up median (IQR) of VCY was 5.27 (2.69-11.19 log10 copies/mL). The quartiles of the VCY distribution identified the following groups: A: 0-35%, B: 36-40% C: 41-55% and D: 56+% of maximum rectangular. We observed 175 AIDS, 187 SNAE and 69 deaths. The magnitude of risk associated with 1 log10 higher VCY was under-estimated when controlling for CD4 as time-updated covariate vs. IPW (e.g. for AIDS/death HR=1.10 vs. HR=1.19). Results of the stratified analysis are shown in Table.

Conclusions: The degree of association between VCY and risk of clinical events varied across different AUC shapes, especially for AIDS/death. In particular, for a given difference in AUC, in people showing stable VL trajectories rather than periods of VL dips and spikes, VCY appear to better discriminate the risk. Strategies to maximize the chance of viral suppression should be considered for patients with suboptimal viral response even at low-level detectable viremia.

viewer?TAG_ACTION=SECURE_FILE_DOWNLOAD&V