Combined Effect of Interleukin-6 and D-dimer on the Risk of Serious Non-AIDS Conditions: Data from 3 Prospective Cohorts
Birgit Grund*1, J Baker2, S Deeks3, J Wolfson1, D Wentworth1, A Cozzi-Lepri4, C Cohen5, A Phillips4, J Lundgren6, J Neaton1, and INSIGHT SMART/ESPRIT/SILCAAT Study Group
1Univ of Minnesota, Minneapolis, US; 2Hennepin County Med Ctr, Minneapolis, MN, US; 3San Francisco Gen Hosp and Univ of California, San Francisco, US; 4Univ Coll London, UK; 5Univ of California, San Francisco, US; and 6Copenhagen Univ Hosp and Univ of Copenhagen, Denmark
Background: Despite effective ART, HIV+ patients are at an increased risk of serious non-AIDS (SNA) conditions (cardiovascular disease [CVD], liver and renal disease, and cancers), perhaps due in part to ongoing inflammation and/or coagulation. We use the relationship of interleukin-6 (IL-6) and D-dimer levels with a composite endpoint of SNA morbidity or all-cause death (SNA/death) in 3 large cohorts to predict the effect of reducing IL-6 and/or D-dimer.
Methods: Associations of log-transformed IL-6 and D-dimer levels at study entry with SNA/death were studied using Cox regression in 3766 patients on suppressive ART in the control arms of the SMART (n = 1748), ESPRIT (1446), and SILCAAT (572) trials. All models include age and sex; baseline CD4 T cell count was excluded by model selection. To correct for random fluctuations in IL-6 and D-dimer that attenuate associations between usual levels of biomarkers and disease risk (regression dilution), 3-year longitudinal biomarker data for 235 patients on stable ART were used. Hazard ratios (HR) for lower baseline levels of IL-6 and D-dimer are used to predict risk reductions.
Results: Median (interquartile range [IQR]) for IL-6 and D-dimer were 1.7 (1.1-2.7) pg/mL and 0.22 (0.15-0.35) μg/mL, correlation 0.31. Over mean follow-up of 4.9 years, 262 patients developed an event, primarily non-AIDS cancer and CVD. In models with only one biomarker, IL-6 and D-dimer were each associated with risk of SNA/death (hazard ratio [HR] = 0.86 and 0.91 per 25% lower levels, respectively; p <0.001 for both). In joint models, both biomarkers were independently associated with SNA/death (p <0.001 for IL-6, p = 0.02 for D-dimer, HR = 0.82 per 25% lower in both), with consistent results across the 3 cohorts (HR 0.80-0.84) and across SNA event types (0.64-0.88). The resulting IL-6/D-dimer score (0.77*log2 IL-6 + 0.33*log2 D-dimer, after adjustment for regression dilution; reliability coefficient = 0.43) was used to predict HR of SNA/death for any combination of 0%-40% reductions in IL-6 and/or D-dimer (Figure).
Conclusions: An IL-6/D-dimer score estimates effects of inflammation and coagulation on SNA/death among treated patients. Findings are consistent across the 3 cohorts. Use of both markers improves precision, suggesting that both pathways contribute. The IL-6/D-dimer score can be used as an outcome for early phase trials aimed at evaluating and comparing treatments with different mechanisms of action for their potential to reduce SNA/death.