Null Results in Brief Prediagnostic Circulating Polyomavirus Antibody Levels and Risk of Non-Hodgkin Lymphoma Lauren R. Teras 1 , Dana E. Rollison 2 , Michael Pawlita 3 , Angelika Michel 3 , Jennifer L. Blase 1 , Martina Willhauck-Fleckenstein 3 , and Susan M. Gapstur 1 Abstract Background: Three human polyomaviruses have been classified as probable (Merkel cell polyomavirus) or possible (BK and JC polyomaviruses) carcinogens, but few epidemiologic studies have examined associations between this growing class of viruses and risk of non-Hodgkin lymphoma (NHL). Methods: Associations between polyomavirus antibodies and NHL incidence were examined using data from the American Cancer Society Cancer Prevention Study-II. This nested case–con- trol study included 279 NHL cases and 557 controls. Prediagnostic antibodies to the major capsid protein of polyomaviruses BKV, JCV, MCV, TSV, WUV, KIV, HPy6, and HPy7 were measured by fluo- rescent bead-based multiplex serology, and associations with NHL were estimated using conditional logistic regression (NHL overall) and unconditional polytomous logistic regression (NHL subtypes). Results: Although an inverse trend was suggested for TSV antibody levels and NHL risk, the HRs were not statistically significant. There were no other observed associations between polyomaviruses and NHL risk. For NHL subtypes, TSV antibody level above the median was associated with a lower risk of CLL/SLL; however, this association was based on 19 cases in the high antibody group and may be due to chance. Conclusions: Our results do not support associations of poly- omaviruses BKV, JCV, WUV, KIV, HPyV6, HPyv7, MCV, or TSV with risk of NHL. Impact: Human polyomavirus antibody levels do not appear to predict a higher NHL risk in immunocompetent individuals. Cancer Epidemiol Biomarkers Prev; 24(2); 477–80. Ó2014 AACR. Introduction Polyomaviruses cause common, asymptomatic infections, and antibodies persist throughout life (1). To date, 12 polyomaviruses have been identified, 10 since 2007. A recent International Agency for Research on Cancer panel classified Merkel cell polyomavirus (MCV) as a probable carcinogen and BK and JC polyomaviruses (BKV, JCV) as possible carcinogens (1). In the only three epide- miologic studies (2–4) to examine associations of BKV and JCV antibodies with risk of non-Hodgkin lymphoma (NHL) or NHL subtypes, no associations were found for BKV, but for JCV results were inconsistent. The only study to examine MCV and risk of NHL showed positive associations with some NHL subtypes (4). Because of the limited number of studies to date, we examined associations of plasma antibody levels of JCV, BKV, and MCV—as well as several other polyomaviruses [WU polyomavirus (WUV); KI polyomavirus (KIV); human polyomavirus 6 (HPy6); human polyomavirus 7 (HPy7); Trichodysplasia spinulosa-associated polyomavirus (TSV)]—with risk of NHL in the American Cancer Society (ACS) Cancer Prevention Study-II (CPS-II) Nutrition Cohort, a large prospective study of U.S. men and women. Materials and Methods Participants in this nested case–control study were selected from the subgroup of CPS-II participants who provided a blood sample between 1998 and 2001 (n ¼ 39,371; ref. 5). Lympho- mas (n ¼ 279) were reported on biennial questionnaires and verified through medical records (n ¼ 209) or registry linkage (n ¼ 70). NHL subtypes were categorized using INTERLYMPH guidelines (6). For each case, two controls were incidence density matched on sex, race, birth, and blood draw dates (6 months). Seroreactivity against BKV, JCV, MCV (isolate 344), TSV, WUV, KIV, HPy6, HPy7 viral capsid protein 1 (VP1) was measured by fluorescent bead-based multiplex serology (1:1000 dilution) at the German Cancer Research Center. In this method, each antigen is carried by a uniquely colored bead. Mixtures of bead sets carrying different antigens are reacted with plasma. A Luminex 100 analyzer quantifies the bead-bound fluorescence-stained human antibodies for each plasma sample and antigen as median fluorescence intensity (MFI; refs. 7, 8). Coefficients of variation for quality control samples were between 1.8% and 5.1%; intraclass correlation coefficients were between 93.6% and 99.7%. Odds ratios (ORs) and 95% confidence intervals (CI) were calculated using conditional logistic regression for all NHL, and uncondi- tional polytomous logistic regression (controlling for the match- ing factors) for the NHL subtypes. Median MFI values for cases and controls were compared using the Wilcoxon rank-sum test. Sero- positive MFI values were 250. Associations with seropositivity and antibody levels were analyzed for each virus. Cutpoints for the latter analysis were based on quartiles of MFI among seropositive control participants. Cubic splines were used to assess potential nonlinear associations. 1 Epidemiology Research Program, American Cancer Society, Atlanta, Georgia. 2 Department of Cancer Epidemiology, Moffitt Cancer Center, Tampa, Florida. 3 Infection and Cancer Program, German Cancer Research Center, DKFZ, Heidelberg, Germany. Corresponding Author: Lauren R. Teras, Epidemiology Research Program, American Cancer Society, 250 Williams Street, NW, Atlanta, GA 30303. Phone: 404-329-5785; Fax: 404-327-6450; E-mail: lauren.teras@cancer.org doi: 10.1158/1055-9965.EPI-14-1125 Ó2014 American Association for Cancer Research. Cancer Epidemiology, Biomarkers & Prevention www.aacrjournals.org 477 on May 28, 2020. © 2015 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from Published OnlineFirst December 8, 2014; DOI: 10.1158/1055-9965.EPI-14-1125