Pre-Diagnosis Plasma Immune Markers and Risk of Non-Hodgkin Lymphoma and Its Major Histologic Subtypes: A Prospective Analysis in Two Large Cohorts

Abstract

Introduction: Severe immune compromise is a strong risk factor for non-Hodgkin lymphoma (NHL), and serum B-cell activation and inflammation markers predicted AIDS-NHL risk in HIV+ persons. We performed a nested case-control analysis in the prospective Nurses’ Health Study (NHS) and Health Professionals Follow-up Study (HPFS) cohorts to characterize pre-diagnosis plasma immune marker profiles associated with risk of HIV-unrelated NHL and major histologic subtypes.Methods: The NHS formed in 1976 with 121,701 female US registered nurses. The HPFS enrolled 51,529 male health professionals in 1986. Both cohorts are followed via biennial questionnaires and confirm cancer diagnoses by medical record review and National Death Index follow-up. Among participants who contributed peripheral blood samples in 1989-90 (NHS) or 1993-4 (HPFS), we confirmed incident primary NHL diagnoses in 345 women and 255 men with no other cancer history. We matched one control per case on cohort (gender), age, race and time of blood draw among persons with no cancer history as of the index date. We determined plasma concentrations of 13 cytokines, soluble receptors and immune activation markers using multiplexed Luminex assays (R & D Systems). In preliminary cohort-specific analyses we calculated matching factor-adjusted odds ratios (OR) and 95% confidence intervals (CI) using logistic regression to estimate the relative risk of NHL per standard deviation (SD) increase in assay batch-adjusted natural log-transformed biomarker concentration. We examined associations for NHL overall and for more common World Health Organization-defined histologic subtypes, including chronic lymphocytic leukemia (CLL; 78 female/70 male cases), follicular lymphoma (FL; 61/24), diffuse large B-cell lymphoma (DLBCL; 69/37) and T-cell NHL (all histologic types, 18/10).Results: The mean (range) years from blood draw to NHL diagnosis was 11 (0.2-22) in women and 7 (0.1-17) in men. In both cohorts, pre-diagnosis levels of the immune activation markers soluble interleukin-2 receptor-α (sIL-2Rα) and sCD30, the B cell chemokine CXCL13, and the inflammatory marker soluble tumor necrosis factor receptor-2 (sTNF-R2) were significantly positively associated with risk of NHL overall (Table 1). Table 1Association of pre-diagnosis plasma immune marker level with risk of NHL overallBiomarkerWomen, OR [95% CI] per SD, p-valueMen, OR [95% CI] per SD, p-valuesIL-2Rα1.41 [1.23, 1.61], p<0.00011.46 [1.22, 1.75], p<0.0001sCD301.36 [1.18, 1.56], p<0.00011.55 [1.29, 1.86], p<0.0001CXCL131.37 [1.20, 1.56], p<0.00011.28 [1.02, 1.59], p=0.03sTNF-R21.23 [1.06, 1.42], p=0.0071.50 [1.23, 1.82], p<0.0001In subtype analyses, risk of CLL was significantly increased in association with increased levels of sIL-2Rα (OR [95% CI] per SD, women: 1.57 [1.26, 1.96], p<0.0001; men: 1.62 [1.24, 2.11], p=0.003) and sTNF-R2 (men only: 1.79 [1.34, 2.39], p<0.0001). To our surprise, the B-cell stimulatory cytokine BAFF was inversely associated with CLL risk in both cohorts (women: 0.59 [0.47, 0.76], p<0.0001; men: 0.56 [0.43, 0.73], p<0.0001); BAFF levels were not associated with any other NHL endpoint. Risk of FL was significantly positively associated with sIL-2Rα (women only: 1.78 [1.36, 2.35], p<0.0001), CXCL13 (women only: 1.64 [1.28, 2.11], p=0.0001), and sTNF-R2 concentrations (women only: 1.57 [1.21, 2.04], p=0.0007). Risk of DLBCL was positively associated with sIL-2Rα (men only: 1.67 [1.21, 2.31], p=0.002) and sTNF-R2 levels (men only: 1.48 [1.04, 2.11], p=0.03). T-cell NHL risk was positively associated with levels of sIL-2Rα, significantly in women (1.83 [1.13, 2.96], p=0.01) and marginally in men (1.65 [0.98, 2.78], p=0.06). No other markers were associated with NHL overall or any subtype. Preliminary analyses by follow-up interval suggested that some of the aforementioned associations may be evident 10+ years after blood draw.Discussion: These findings support an etiologic role for immune activation and inflammation in HIV-unrelated NHL. Apparent gender differences in associations may be due to chance (sparse data); statistical heterogeneity by gender will be evaluated and, if appropriate, data will be pooled to improve stability of estimates. Further evaluation by follow-up interval may lend additional insights on the immune milieu that promotes lymphomagenesis and on potential markers of longer- or shorter-term NHL risk. DisclosuresNo relevant conflicts of interest to declare.