Description of the cohorts
Of the 3,380 NHS participants eligible for this analysis, 564 (16.7%) had not achieved sustained VS by August 31, 2017 and 743 (21.9%) had less than two years of viral load measurements because they were either no longer eligible for DoD benefits (n=244), lost to follow-up (n=226), on ART for less than two years at the time of data collection (n=160), deceased (n=64) or transferred/withdrawn from the study (n=49). Out of 1,784 NHS participants with at least 2 years VS and an available CD4 measure who were eligible for this study, 193 (10.8%) met the definition for INR.
Of the 2,625 AFRICOS participants eligible for this analysis, 313 (12.0%) had not achieved sustained VS by August 31, 2017 and 1,316 (50.1%) had less than two years of viral load measurements because they were either on ART for less than two years (n=1,148), lost to follow-up (n=115) or deceased. (n=53). Out of 984 AFRICOS participants with at least two years VS and an available CD4 measure who were eligible for this study, 254 (25.8%) experienced INR.
Table 1 summarizes the demographic and clinical characteristics of each study population by immune response status. Compared to the NHS, the AFRICOS population was older (37.2 years vs. 34.2 years), with a larger proportion of females (57.0% vs. 6.8%) and a smaller proportion of smokers (2.0% vs. 19.7%) and depression diagnoses (8.0% vs. 20.0%). Median CD4 nadir was higher in the NHS (302 cells/µl vs. 190.0 cells/µl), and while the median time from HIV diagnosis to ART initiation was longer in the NHS than AFRICOS (1.7 years vs. 0.4 years), the median time from ART initiation to VS was shorter (2.5 years vs. 4.8 years). In both cohorts, immune non-responders were significantly older, had a significantly lower CD4 nadir and a significantly longer time from ART initiation to VS, compared to those with GIR. Time from HIV diagnosis to ART initiation was also significantly different by immune response status in both cohorts, though it was longer for immune non-responders in the NHS and shorter in the AFRICOS.
Odds of immune non-response
Table 2 reports the crude and adjusted odds ratios (aORs) for factors found to be significantly different by immune response status in each cohort. In both cohorts, CD4 nadir was significantly associated with INR in the adjusted models, such that an increase in CD4 was associated with decreased odds of INR (aOR=0.31 [95% CI: 0.26, 0.37] and aOR=0.50 [95% CI: 0.43, 0.58] in NHS and AFRICOS, respectively). Hypertension was also associated with INR in the NHS model, such that odds of INR increased by 71% for those with a hypertension diagnosis (p<0.005). In the AFRICOS analysis, age at ART initiation and sex were also significantly associated with odds of INR in the adjusted model, such that odds of INR increased 4% for every year increase in age (p<0.001) and decreased by 48% for women versus men (p=0.001). Table 3 reports the adjusted multiple logistic regression models excluding participants with CD4 >350 cells/µ at ART initiation. In both cohorts the models remained similar to the primary models, though instead of hypertension, diabetes became independently associated with INR among the restricted sensitivity analysis population of the NHS. The Hosmer-Lemeshow test for model fitness found the adjusted models in the NHS and AFRICOS to provide adequate fit (chi-square=13.067; p=0.12 and chi-square=4.58; p=0.80, respectively).
Relative risk of incident serious non-AIDS event
Table 4 reports the Cox proportional-hazards modeling results for hazard of incident SNAE in the NHS, with INR as the primary exposure of interest. In addition to INR, the final adjusted model included age, BMI and history of smoking at ART initiation, as well as history of hypertension, depression, diabetes and “at-risk” drinking. Crude hazards ratios stratified by immune response status did not provide enough evidence for effect modification and interaction terms were therefore not included in the model.
After adjustment, age and INR remained statistically significant, such that a one-year increase in age was associated with a 7% increase in the relative risk of incident SNAE (p<0.001), and INR was associated with a 61% increased relative risk of SNAE compared to GIR (p=0.0098).
Time to incident serious non-AIDS event
The Kaplan-Meier curves in Figure 1 depict the time to incident SNAE after sustained VS as survival probabilities stratified by immune response status, age at ART initiation and nadir CD4. Median time to incident SNAE was approximately six months longer for NHS participants with GIR compared to INR (13.74 years vs. 13.19 years, p<0.001). The probability of "SNAE-free" survival at 15 years since sustained VS was approximately 20% lower comparing those with and without INR (p<0.001); nearly equal to the differences observed by 15-year age groups. Median time to new SNAE was not significantly different by CD4 nadir (14.56 years for CD4<350 vs. 13.68 years for CD4≥350, respectively; p=0.51).