Skip to main content
eScholarship
Open Access Publications from the University of California

Circulating Markers of Immune Activation and Inflammation and AIDS-Associated Non-Hodgkin Lymphoma in the Multicenter AIDS Cohort Study (MACS)

  • Author(s): Makgoeng, Solomon B.
  • Advisor(s): Arah, Onyebuchi A
  • Hussain, Shehnaz K
  • et al.
Abstract

Background: AIDS-associated non-Hodgkin lymphoma (AIDS-NHL) remains a significant public health challenge among HIV-infected individuals. Chronic inflammation and immune activation have been documented in the literature to play a crucial role in the etiology of AIDS-NHL. We summarized results from prior work in a meta-analysis of the associations between prospectively measured circulating levels of immune biomarkers and the risk of NHL among both HIV-infected and HIV-uninfected populations. Our second study characterized the temporal variation in 24 pre-AIDS-NHL diagnosis circulating markers of inflammation and immune activation. Finally, we assessed the predictive ability of a set of 13 biomarker levels and AIDS-NHL diagnosis.

Methods: Meta-analysis: Our meta-analysis identified 17 relevant studies from inception of major biomedical databases (PubMed, EMBASE, and Web of Science) until January 1, 2017. We summarized published results using random-effects models for NHL and several histological subtypes of NHL. Longitudinal study: we summarized the slopes and means (intercepts) of biomarker trajectories using linear mixed models. Prediction Models: we calculated incremental discrimination ability (AUC) of models including biomarkers, individually and concurrently, relative to models including only participant characteristics and known risk factors of NHL.

Results: Meta-analysis: Summarizing 17 nested case–control studies, we found elevated levels of several biomarkers associated with increased odds of NHL overall: TNF-α, OR=1.18 [95% CI: 1.04, 1.34]; CXCL13, OR=1.47 [95% CI: 1.03, 2.08]; sCD23, OR=1.57 [95% CI: 1.21, 2.05]; sCD27, OR=2.18 [95% CI: 1.20, 3.98]; sCD30, OR=1.65 [95% CI: 1.22, 2.22]; AIDS-NHL showing stronger associations with IL-6, TNF-α, sCD27, and sCD30. Longitudinal study: prior to HAART, geometric mean biomarkers are elevated for cases relative to controls for IL-2, TNF-α, IL-6, sCD27, sIL-2Rα, IP-10, CXCL13, CRP, and pre-HAART slopes were observed to be higher for cases relative to controls for sIL-6R, sTNFR2, IL-10. Following HAART initiation, geometric mean levels are elevated to a higher degree than pre-HAART for cases relative to controls for BAFF, TNF-α, sIL-2Rα, sTNFR2, IP-10, MCP-1, CRP. Prediction Modeling: Models including individual biomarkers yielded modest improvements in AUC statistics above a base-case model that comprised NHL risk factors and other participant characteristics. A model including IL-6, IL-10, TNF-α, IP10, and CXCL13, concurrently performed better than all other models including individual biomarkers: 0-1: AUC=0.943 95% CI: 0.910, 0.975; 1-3: AUC=0.895 95% CI: 0.856, 0.934; >3: AUC=0.836 95% CI: 0.787, 0.885. Increments in AUC above the risk factors only model were also improved relative to individual biomarker models: 0-1: difference in AUC=0.056 95% CI: 0.021, 0.091; 1-3: difference in AUC=0.032 95% CI: 0.007, 0.057; >3: difference in AUC=0.074 95% CI: 0.030, 0.118.

Conclusion: Each study provided further novel evidence of the association between circulating biomarkers and AIDS-NHL risk, as well as the utility of biomarkers in risk prediction. Our meta-analysis provides an overarching summary of evidence that elevated circulating levels of several markers are associated with an increased risk of NHL. Longitudinal analyses illustrate novel differences between cases and controls in aspects of the trajectories of 24 markers. Our prediction models elucidate the ability of a set of 13 marker levels to discriminate between AIDS-NHL cases versus controls. The totality of new evidence we provide supports the notion that chronic inflammation and immune activation is associated with increased AIDS-NHL risk, and that these biomarkers may have utility in the development of clinical risk prediction models.

Main Content
Current View