Immunological and Virological Response to Initial Antiretroviral Therapy Among Older People Living With HIV in the Canadian Observational Cohort (CANOC)

A Zhabokritsky; L Szadkowski; AN Burchell; C Cooper; RS Hogg; M Hull; DV Kelly; M Klein; M Loutfy; A McClean; J Montaner; SL Walmsley


HIV Medicine. 2021;22(8):759-769. 

In This Article


Study Population

This study was conducted using data from the Canadian Observational Cohort (CANOC), an interprovincial usual care cohort of ART-naïve PLWH across Canada who initiated treatment with at least three antiretrovirals on or after 1 January 2000. Complete details of the CANOC design and data collection have been described elsewhere.[17] In brief, individuals were enrolled from 11 participating cohorts in British Columbia, Saskatchewan, Ontario, Quebec and Newfoundland and Labrador. CANOC participants were 18 years of age or older and had at least one pre-treatment CD4 count and HIV RNA measured within 12 months prior to ART initiation. This analysis uses the most recent data available up to 31 December 2016, and includes all participants starting ART with at least two nucleoside reverse transcriptase inhibitors (NRTIs) and at least one antiretroviral agent from another class [i.e. INSTI-, non-NRTI (NNRTI)- or protease inhibitor (PI)-based regiments] who had at least one post-treatment measurement of CD4 count or viral load.


The immunological outcome of interest was CD4 recovery which was assessed using three separate outcomes: (1) absolute CD4 counts over time, (2) time from ART initiation to CD4 count ≥ 200 cells/μL, and (3) time from ART initiation to CD4 count ≥ 500 cells/μL. Virological outcomes included time from ART initiation to virological suppression (viral load ≤ 50 copies/mL) and time from virological suppression to virological failure (viral load > 200 copies/mL), with the higher threshold selected to avoid detection of transient low-level viraemia. All time-to-event outcomes required two consecutive measurements at least 30 days apart.

Statistical Analysis

Demographic and clinical characteristics were compared by age group at the time of treatment initiation (≤ 30, 31–40, 41–50, 51–60 and > 60 years) using the median, interquartile range (IQR) and Kruskal–Wallis rank sum test for continuous variables and the frequency, percentage and χ2 test for categorical variables. The following variables were chosen a priori to be considered for association with immunological or virological response in all analyses described below: age at treatment initiation, sex at birth, men who have sex with men (MSM), injection drug use (IDU), province, pre-treatment CD4 count, pre-treatment viral load, calendar year of treatment initiation, initial treatment regimen, and history of hepatitis C. Age was investigated as both a continuous and a categorical variable. The effect of age on immunological outcomes was not linear and so we have presented age as a categorical variable. Missing data were considered non-informative; multivariable models included complete-case data. As the primary objective of this study was to assess the effect of age on CD4 and virological outcomes (as opposed to the overall incidence of achieving these outcomes), we used cause-specific proportional hazards (PH) models for all time-to-event outcomes.[18]

Longitudinal absolute CD4 count response to treatment was modelled using generalized estimating equations (GEE) with an exchangeable correlation matrix. Time from treatment initiation was included as a natural cubic spline with three knots in order to capture non-linear changes over time. Differences in the rate of change over time by age group were investigated using an age–time interaction.

Only participants with a pre-treatment CD4 count less than the outcome were included in each time-to-event analysis (time to CD4 count ≥ 200 cells/μL and CD4 count ≥ 500 cells/μL). Participants were censored if there was a gap between CD4 count measures lasting 2 years or longer or at their last CD4 count measurement. Death within 1 year of the last CD4 count was considered a competing risk. Time to CD4 cut-off was estimated and compared by age group using the cumulative incidence functions for competing risks data.[19] Cause-specific PH models were used to identify factors associated with CD4 recovery.

For the virological outcome analyses, time from treatment initiation to suppression and time from suppression to failure were estimated and compared by age using cumulative incidence functions with a competing risk of death. Participants were censored if there was a 2-year gap between viral load measurements or at the last viral load. Univariable and multivariable cause-specific PH models of time to suppression and time to virological failure were used to identify variables associated with each outcome.


The human subjects activities of the CANOC have been approved by the Simon Fraser University Research Ethics Board and the University of British Columbia Research Ethics Board as well as the local institutional review boards at each of the participating cohorts.