Open access journal of the Ferrata-Storti Foundation, a non-profit organization
Articles

Pre-diagnosis plasma immune markers and risk of non-Hodgkin lymphoma in two prospective cohort studies

Department of Medicine and the Meyers Primary Care Institute, University of Massachusetts Medical School, Worcester, MA, USA
Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, MA, USA;Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA
UCLA AIDS Institute, Los Angeles, CA, USA;Department of Psychiatry & Biobehavioral Sciences, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA
Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA
Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA;Department of Nutrition, Harvard T.H. Chan School of Public Health, Boston, MA, USA;Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA, USA
UCLA AIDS Institute, Los Angeles, CA, USA;Department of Obstetrics & Gynecology, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA
Department of Pathology, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA
Department of Pathology, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA
Slone Epidemiology Center at Boston University, Boston, MA, USA
Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA;Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA, USA
UCLA AIDS Institute, Los Angeles, CA, USA;Department of Obstetrics & Gynecology, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA;Department of Epidemiology, UCLA Fielding School of Public Health, Los Angeles, CA, USA;Department of Microbiology, Immunology & Molecular Genetics, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA;Jonsson Comprehensive Cancer Center, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA
Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA
Vol. 103 No. 10 (2018): October, 2018 https://doi.org/10.3324/haematol.2017.183236

Abstract

Inflammation and B-cell hyperactivation have been associated with non-Hodgkin lymphoma development. This prospective analysis aimed to further elucidate pre-diagnosis plasma immune marker profiles associated with non-Hodgkin lymphoma risk. We identified 598 incident lymphoma cases and 601 matched controls in Nurses’ Health Study and Health Professionals Follow-up Study participants with archived pre-diagnosis plasma samples and measured 13 immune marker levels with multiplexed immunoassays. Using multivariable logistic regression we calculated Odds Ratios (OR) and 95% Confidence Intervals (CI) per standard deviation unit increase in biomarker concentration for risk of non-Hodgkin lymphoma and major histological subtype, stratifying additional models by years (<5, 5 to <10, ≥10) after blood draw. Soluble interleukin-2 receptor-α, CXC chemokine ligand 13, soluble CD30, and soluble tumor necrosis factor receptor-2 were individually positively associated, and B-cell activating factor of the tumor necrosis factor family inversely associated, with all non-Hodgkin lymphoma and one or more subtypes. The biomarker combinations associated independently with lymphoma varied somewhat by subtype and years after blood draw. Of note, the unexpected inverse association between B-cell activating factor and chronic lymphocytic leukemia/small lymphocytic lymphoma risk (OR: 95%CI: 0.51, 0.43-0.62) persisted more than ten years after blood draw (OR: 0.70; 95%CI: 0.52-0.93). In conclusion, immune activation precedes non-Hodgkin lymphoma diagnosis by several years. Decreased B-cell activating factor levels may denote nascent chronic lymphocytic leukemia many years pre-diagnosis.

Introduction

Severe immune compromise is a strong risk factor for non-Hodgkin lymphoma (NHL), and B-cell activation and inflammation have been associated with an increased risk of AIDS-related NHL. Elevated pre-diagnosis plasma levels of markers of B-cell stimulation including CXC chemokine ligand 13 (CXCL13; a B-cell attracting chemokine),1 interleukin (IL)-6 (a B-cell stimulatory cytokine), and soluble (s) CD30 (sCD30; a soluble receptor indicative of B- and T-cell activation) predicted risk of an AIDS-NHL diagnosis in HIV-positive persons,42 in some instances as early as five years pre-diagnosis.5 Several of these markers have also demonstrated an association with NHL risk in immunocompetent people in prospective studies.136 Of interest, plasma sCD30 levels were positively associated with NHL risk at 6-10 years9 and even 15-23 years pre-diagnosis.11 Another small nested case-control study reported a significant 2.5-fold increase in NHL risk in women with elevated soluble IL-2 receptor-α levels (sIL-2Rα; a marker of T-cell activation and IL-2 upregulation), and marginally significant increases in NHL risk in women with higher pre-diagnosis tumor necrosis factor (TNF)-α and soluble TNF-receptor-2 (sTNF-R2) levels.14 These findings collectively suggest that chronic B-cell stimulation has a role in lymphomagenesis in immunocompetent persons.

Our study aimed to further characterize pre-diagnosis plasma immune marker profiles associated with risk of HIV-unrelated NHL and its major histological subtypes in two large US cohorts. This study represents one of the largest populations with prospectively collected pre-diagnosis blood samples to investigate the association between numerous immune markers and NHL risk, including those with specific NHL subtypes that are often precluded due to small sample size, and to assess the independence of biomarker-NHL associations for multiple immune markers.1211 The long-term follow up of the study population also allowed for examination of the influence of time since blood draw on observed immune marker-NHL associations, including an assessment of potential early markers of lymphomagenesis present ten years or more prior to diagnosis. The choice of immune markers was guided in part by the immune deregulation we sought to characterize and by reported findings in AIDS- or HIV-unrelated NHL. We hypothesized that pre-diagnosis levels of immune markers indicative of B-cell activation or inflammation would be positively associated with risk of developing NHL and major NHL subtypes, and that the use of multi-marker models will enhance characterization of the immune milieu associated with NHL risk and suggest subtle differences by histological subtype.

Methods

Study population

The study population comprised Nurses’ Health Study (NHS, all female) and Health Professionals Follow-up Study (HPFS, all male) participants with archived plasma (Online Supplementary Methods).1615 Cancer diagnoses were identified via routine questionnaires or follow up after death1817 and confirmed by medical record review or tumor registry linkage.

Participants provided written informed consent at blood collection. The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Boards of the Brigham and Women’s Hospital and Harvard T.H. Chan School of Public Health.

Case and control selection

We included all participants with confirmed incident NHL diagnosed three months or more after blood draw through 31 December 2010 with no other cancer history. Study pathologists (JCA, SJR) classified NHL histological subtype19 according to World Health Organization2120 and International Lymphoma Epidemiology (InterLymph) Consortium guidelines.2322 We analyzed common B-cell (B-)NHL subtypes individually [diffuse large B-cell lymphoma (DLBCL), follicular lymphoma (FL), and chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL/SLL)], combined less common B-NHLs (“other B-NHL”) and defined additional categories by cell type (T-NHL, B-NHL). We matched one control per case by sex (cohort), age, race, and blood draw details (Online Supplementary Methods).

Biomarker assessment

Assays were performed at the University of California, Los Angeles (LM, OMM), using multiplexed kits (Fluorokine MAP, R & D Systems, Minneapolis, MN, USA), a Bio-Plex 200 Luminex instrument and Bio-Plex analysis software (Bio-Rad, Hercules, CA, USA). Blinded laboratory personnel measured sCD30, sIL-2Rα, B-cell activating factor of the TNF family (BAFF, a B-cell stimulatory cytokine), CXCL13, sIL-6Rα, sGP130, sCD14, sTNF-R2, C-reactive protein (CRP), IL-6, IL-8, IL-10, and TNF-α concentration according to the manufacturer’s directions (Online Supplementary Methods). We set TNF-α, IL-8 and CXCL13 values to missing for samples with >24-hour processing delays (NHS: n=35; HPFS: n=23). Analyte concentrations were natural log-transformed for all analyses. We observed similar measured biomarker concentrations for the NHS and HPFS (Online Supplementary Table S1) and pooled the data.

Statistical analysis

We conducted batch calibration to diminish the potential influence of laboratory batch-related variability on biomarker-NHL associations.24 Outlying biomarker values were identified using the Rosner extreme Studentized deviate method25 and omitted from analyses of the marker.

The primary analysis assessed batch effect-corrected, log-transformed biomarker values continuously per Standard Deviation (SD) increase in concentration, with SD units calculated for log-transformed values in the pooled controls. We calculated Odds Ratios (OR) and 95% Confidence Intervals (CI) for the association of each biomarker with NHL risk (overall and for DLBCL, FL, CLL/SLL, other B-NHL, all B-NHL and all T-NHL) using unconditional logistic regression. Models adjusted for all matching factors unless small cell counts precluded adjustment for race. We evaluated but did not observe confounding by body mass index (BMI) and autoimmune disease history.

We intended a priori to identify multi-marker profiles associated with NHL risk via mutual adjustment of models for biomarkers that were individually associated. We also examined models stratified by follow-up interval (0 to <5, 5 to <10, ≥10 years) and assessed heterogeneity by time period using the contrast test.26 The Online Supplementary Methods describe additional analyses designed post hoc.

Results

In total, 601 cases of NHL (345 NHS and 256 HPFS) were identified and individually matched to controls. Three cases were later excluded due to unconfirmed lymphoma status. The final analysis thus included 598 cases, including 114 DLBCL, 92 FL, 165 CLL/SLL, 132 other B-NHL (4 Burkitt lymphoma, 19 lymphoplasmacytic lymphoma, 20 mantle cell lymphoma, 44 marginal zone lymphoma, 20 other B-NHL, and 25 unclassified B-NHL) and 30 T-NHL, and 601 controls. The study population was 96% Caucasian and 58% female. Cases and controls had similar covariable distributions, due in part to the matched design (Table 1).

Table 1.Characteristics of non-Hodgkin lymphoma cases and matched controls from two prospective cohort studies.

We omitted 109 individual biomarker measurements (<1% of all measurements) with implausible outlying values (NHS: 72; HPFS: 37), the majority (90%) of which were implausibly high for the particular marker. Omitted values ranged from one measure of IL-10 to 17 measures of IL-8. Spearman correlation coefficients ranged from −0.03 (IL-10 and CXCL13) to 0.58 (sIL-2Rα and sCD30) (Online Supplementary Table S2).

Individual immune marker models

Multivariable analyses of individual log-transformed immune markers revealed significant associations for all NHL per SD increment of log-transformed sTNF-R2, sIL-2Rα, CXCL13, sCD30 (all positive) and BAFF (inverse; Table 2). In subtype-specific analyses, sTNF-R2 levels were also positively associated with risk of all B-NHL, FL and CLL/SLL, while CXCL13 was positively associated with risk of all B-NHL, DLBCL and FL (Table 2). Levels of sIL-2Rα and sCD30 were positively associated with every NHL subtype, including T-NHL. Of interest, the association of BAFF with a 17% decreased risk of all NHL appeared to be driven by CLL/SLL, for which risk decreased by 49% per SD increase in log-transformed BAFF levels (OR: 0.51; 95%CI: 0.43, 0.62; P<0.001); BAFF was not associated with other NHL subtypes in single-marker models. We did not observe significant or consistent associations for the remaining immune markers with risk of any NHL end point. Results from cohort-specific models did not suggest marked differences by sex for these associations (Online Supplementary Table S3).

Table 2.Associations between pre-diagnosis concentrations of 13 individual immune markers and risk of non-Hodgkin lymphoma (NHL), overall and by major histological subtype.

Multi-marker profiles

In the model that mutually adjusted for the five log-transformed immune markers that had significant individual associations with NHL end points (sTNF-R2, sIL-2Rα, CXCL13, sCD30, BAFF), sIL-2Rα, CXCL13 and sCD30 remained significantly associated with a 17-24% increased risk, and BAFF with a 26% decreased risk of all NHL per SD increase in log concentration, while sTNF-R2 was no longer significantly associated (Table 3). Results for all B-NHL risk were similar to those for all NHL, whereas mutual adjustment attenuated all the immune marker associations with DLBCL. In the multi-marker model of FL risk, sCD30 and BAFF remained independently associated, with a borderline association noted for CXCL13 (Table 4). In the multi-marker model of CLL/SLL risk, sIL-2Rα was significantly associated with a 50% increase (95%CI: 1.18-1.90), and BAFF with a significant 53% reduction (95%CI: 0.38, 0.58), per SD increase in log concentration. Lastly, only sIL-2Rα was independently associated with T-NHL risk (OR per SD increase in log concentration: 1.96; 95%CI: 1.22, 3.13) in mutually adjusted models.

Table 3.Independent associations of multiple pre-diagnosis plasma immune markers with risk of non-Hodgkin lymphoma (NHL), overall and by B- or T-cell type of origin, for the complete follow-up period and stratified by years of follow up.

Table 4.Independent associations of multiple pre-diagnosis plasma immune markers with risk of non-Hodgkin lymphoma (NHL) by major histological subtype of B-cell NHL, for the complete follow-up period and stratified by years of follow up.

The 5-marker models using the polytomous logistic regression (PLR) approach yielded essentially the same effect estimates as described above for biomarker associations with the NHL end points for the full follow-up period (Online Supplementary Tables S4 and S5). sTNF-R2 had significantly different associations with B-NHL and T-NHL (P-value for heterogeneity by subtype=0.04) (Online Supplementary Table S4); the associations of CXCL13 and BAFF with individual B-NHL subtypes also showed evidence of significant heterogeneity (P-values for heterogeneity by subtype =0.007 and <0.0001, respectively) (Online Supplementary Table S5).

In the covariable-adjusted multi-marker models containing restricted cubic splines, there was evidence of non-linearity for two biomarkers, CXCL13 and BAFF, in their associations with risk of aggregated end points (all NHL, B-NHL and other B-NHL; P-value tests for significance of the curve <0.05), but not for biomarker associations with individual B-NHL subtypes or T-NHL.

In alternative models using semi-automatic stepwise selection, the final models for all NHL and all B-NHL included sIL-2Rα, CXCL13 and sCD30, which were positively associated with risk, as well as BAFF, which was inversely associated (Online Supplementary Table S6). In comparison, for DLBCL and FL, the stepwise procedure selected only sCD30 (P=0.004 and <0.0001, respectively), and for T-NHL the procedure selected only sIL-2Rα (P=0.002) as independently (positively) associated with risk. Of interest, the stepwise procedure identified four immune markers independently associated with risk of CLL/SLL, including BAFF and IL-10 with significant inverse associations and sIL-2Rα with a significant positive association. The stepwise procedure identified three immune markers associated with the combined category of other B-NHL subtypes, including significant positive associations for CXCL13 and sIL-2Rα, and a significant inverse association for BAFF.

In the model that included all 13 immune markers, sIL-2Rα, CXCL13, and sCD30 again had strong positive associations with risk of all NHL and all B-NHL (Table 5). In the CLL/SLL-specific model, we observed a significant inverse association with risk for BAFF and also for sCD14 and IL-10, and a positive association with sIL2-Rα. BAFF was also significantly inversely associated with FL risk, while sCD30 was significantly positively associated with FL risk. Only sIL-2Rα was significantly associated with an increased risk of T-NHL. We observed suggestive positive associations of DLBCL risk with IL-6, CXCL13 and sCD30 in this 13-marker model.

Table 5.Associations of pre-diagnosis plasma immune markers with risk of non-Hodgkin lymphoma (NHL), with mutual adjustment for all thirteen immune markers, for all NHL and by major histological subtype.

Time-stratified analyses

The analyses stratified by time between blood draw and diagnosis/index date suggested that the individual biomarker associations with all NHL (Online Supplementary Table S7) and with NHL subtypes (Online Supplementary Table S8) varied somewhat by length of time after blood draw but did not strongly implicate any additional immune marker-NHL associations. The time-stratified 5-marker models (Table 3) also suggested variability by follow-up interval in the independent associations of those immune markers with future NHL risk. For example, the association of sIL-2Rα with risk of all NHL appeared to be restricted to a shorter-term interval, specifically within five years of blood draw (OR: 1.52, 95%CI: 1.09, 2.11) (Table 3), whereas significant associations of CXCL13 with risk of all NHL were evident only five or more years after blood collection (5-<10 years; OR: 1.23, 95%CI: 1.00, 1.52; and ≥ 10 years; OR: 1.21, 95%CI: 1.01, 1.45). sCD30 was most strongly associated with all NHL risk within ten years of blood draw, while BAFF was consistently inversely associated with all NHL across time periods. Of note, in subtype-specific time-stratified analyses, sCD30 levels were strongly positively associated with risk of FL within five years of blood draw (OR: 4.85, 95%CI: 2.02, 11.61), and the association decreased in magnitude with increasing follow-up time. In CLL/SLL-specific models, elevated sIL-2Rα was associated with a nearly 4-fold increased risk within five years of blood draw (OR: 3.71, 95%CI: 1.77, 7.76) but had no clear association with longer-term CLL/SLL risk. In contrast, BAFF had significant inverse associations with risk of CLL/SLL in all pre-diagnosis time periods, albeit with particularly strong associations with risk of CLL/SLL within five or ten years of blood draw (Table 4). When modeled using PLR, the effect estimates were virtually the same for time period-specific biomarker associations, both for the aggregated and the individual NHL end points (Online Supplementary Tables S4 and S5). The most prominent differences between the two approaches for assessing heterogeneity by time period (PLR with interaction terms vs. time-stratified unconditional logistic regression) pertained to the statistical significance of apparent heterogeneity by follow-up period for the associations of sTNF-R2 with all B-NHL and FL. For example, for the association of sTNF-R2 with all B-NHL, the P-value for heterogeneity by follow-up time was 0.04 for the cross-product term in PLR (Online Supplementary Table S4) and 0.40 for the main model contrast test (Table 3). For the association of sTNF-R2 with FL, the P-value for heterogeneity by time period was 0.0007 for the cross-product term in PLR (Online Supplementary Table S5) and 0.11 for the main model contrast test (Table 4). Time-stratified results for the multi-marker models identified with stepwise selection largely agreed with the main results described above (Online Supplementary Table S6).

Discussion

In this pooled analysis within the NHS and HPFS cohorts, we observed significant associations between NHL risk and pre-diagnosis levels of specific plasma immune markers, including a novel, inverse association between levels of BAFF and risk of CLL/SLL. Positive associations between levels of sIL-2Rα, CXCL13, and sCD30 and risk of all NHL and all B-NHL, as well as the inverse association of BAFF with risk of all NHL and CLL/SLL, were consistent and independent across several analytical approaches to constructing a multi-marker profile associated with risk. In contrast, the individual positive associations noted for sTNF-R2 with risk of all NHL and some B-NHL end points were attenuated upon adjustment for other immune markers, suggesting a lack of independence in the association between sTNF-R2 levels and NHL risk. Manual selection and automated stepwise selection of multi-marker profiles yielded fairly consistent results for all NHL, but also some differences for individual histological subtypes, particularly for CLL/SLL. We also observed some variation in the associations between NHL risk and immune markers by time between blood draw and diagnosis.

Our findings are in agreement with previous studies reporting associations between elevated CXCL13 and/or sCD30 levels and increased NHL risk in HIV-positive and immunocompetent populations, including several reports analyzing blood samples taken many years prior to NHL diagnosis.13832 In the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study, Purdue et al.11 prospectively investigated multi-marker models similar to those in our analysis and observed independent positive associations for sCD30 with risk of NHL and DLBCL when adjusted for other biomarkers. Those observations were detectable more than 15 years prior to diagnosis. Also similar to our findings, positive associations observed for sTNF-R2 with NHL did not persist upon adjustment for other immune markers.11 In a prospective analysis in the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial, individual associations of CXCL13 and sTNF-R2 with NHL both remained significant with mutual adjustment, with correction for multiple comparisons and with restriction to samples collected 8-13 years prior to diagnosis.10

We observed an unexpected yet consistently strong inverse association between BAFF levels and CLL/SLL risk. BAFF is a member of the TNF family involved with B-cell survival and maturation.27 Pre-diagnosis serum BAFF concentrations were positively associated with AIDS-NHL, and BAFF overproduction has been associated with systemic autoimmune diseases, including systemic lupus erythematosus and Sjögren syndrome,2928 which are associated with an increased risk of NHL in HIV-negative persons.3130 However, systemic autoimmune disorders in HIV-negative individuals appear to be preferentially associated with NHL subtypes with a different natural history than CLL/SLL.3230 Nonetheless, CLL cells are known to express multiple BAFF receptors (including TNFRSF13B, TNFRS13C and TNFRSF17),33 and the inverse association that we observed is biologically plausible if considered indicative of rapid uptake of circulating BAFF by nascent CLL/SLL clones,34 reflecting subclinical progression of an indolent tumor whose natural course may extend multiple decades. Consistent with this interpretation, several clinical studies have observed lower levels of BAFF in sera from CLL/SLL patients than in healthy controls.3735 The mechanism for the latter findings is unknown; our observation suggests that those underlying physiological processes may commence early in CLL pathogenesis, even ten or more years pre-diagnosis. Concurrent measurement of soluble BAFF receptors and study of cell surface expression of those molecules and classification of cases into prognostic subgroups were not feasible for the present study. Confirmation of the present findings is warranted in larger populations with specimens suitable for determining cell surface marker or gene/protein expression. Additionally, prospective studies in patients with monoclonal B-lymphocytosis would be informative to evaluate whether circulating BAFF levels can enhance risk stratification for progression to malignancy.38

We also observed significant associations of elevated sIL-2Rα levels with increased risks of all NHL, B-NHL, DLBCL, CLL/SLL and T-NHL, primarily within five years of blood draw. One other study reported a positive association between sIL-2Rα levels and NHL risk in an HIV-negative population with prospective blood collection that persisted after incorporating lag-times greater than two years.14 Of note, comparatively high sIL-2Rα levels at diagnosis were also associated with poor prognosis in patients with NHL.4139 Biologically, sIL-2Rα and sCD30 are highly correlated (r=0.58 in this study), and both can indicate B-and T-cell activation;42 in the present analysis, both markers remained independently associated with a significant increased risk of all NHL and all B-NHL after mutual adjustment. In contrast, only sIL-2Rα was significantly associated with an increased risk of T-NHL in the multi-marker models, although small sample size (n=30 cases) limited statistical power to detect significant independent associations for more strongly correlated biomarkers. Of interest, we observed the strongest positive associations of sIL-2Rα with T-NHL risk within ten years of blood draw, a novel observation that requires confirmation in other populations.

We observed significant positive associations between CXCL13 and risk of all NHL, B-NHL, and FL, as well as borderline associations with DLBCL and other B-NHL, more than ten years after blood draw, suggesting an early role for an immune environment characterized by B-cell stimulation and aberrant B-cell trafficking. Consistent with this interpretation, a recent, large-scale genome-wide association study of FL identified CXCR5, which is the receptor for CXCL13, as a potential FL susceptibility locus.43 Further, genetic variation in CXCR5 and CXCL13 was associated with serum CXCL13 levels in a study of AIDS-NHL, and elevated serum CXCL13 levels were observed in AIDS-NHL cases more than three years prior to diagnosis.2 In contrast, elevated sCD30 levels were more strongly associated with increased risk of all NHL, B-NHL and FL within ten years of blood draw, with a particularly strong association with FL within five years of blood draw. These findings suggest sCD30 may be capturing a more proximal pre-diagnosis increase in immune activation.

When assessed with multivariable PLR models rather than the main unconditional logistic regression analysis, the associations between immune markers and NHL end points did not change substantially, whether for aggregated end points or the individual B-NHL end points. The minor discrepancies suggested somewhat improved precision in the PLR models, which yielded slightly narrower confidence intervals and slightly stronger P-values for heterogeneity by follow-up period for a few of the comparisons. None of the discrepant findings would suggest a different interpretation of the time- or subtype-specific model findings, however, and thus we retained the unconditional logistic regression models as our primary analysis for methodological consistency across the full series of analyses we conducted.

In the analyses with restricted cubic splines, we observed evidence of significant non-linearity for associations of CXCL13 and BAFF with aggregated NHL end points. Of note, those end points comprise small numbers of diverse histological subtypes of NHL which may have different etiologies. Thus, we believe that the observed non-linear associations more likely reflect sampling variability and/or an artifact of potentially heterogeneous subtype-specific associations for the subtypes in the end point groups than a true biological effect.

Together, our findings add new insight to previous publications on both AIDS-NHL and HIV-unrelated NHL risk, collectively suggesting that higher levels of immune activation, and in particular heightened B-cell stimulation, may affect B-cell lymphomagenesis. Interestingly, several markers of immune activation appear to be elevated many years prior to NHL diagnosis and thus could help identify populations at higher risk for developing NHL. It is important to note that some reported associations between immune markers and all NHL risk were not replicated in analyses of individual histological subtypes; this may be due in part to subtype-specific sample sizes that limited statistical power. Significant associations between immune markers and risk of all NHL may reflect commonalities in subtype etiologies; however, these findings may also conceal a more specific association with one or more of the less common subtypes, as illustrated by the present findings for BAFF and CLL/SLL.

This analysis of immune markers measured from prospectively collected blood specimens from two large US cohorts with lengthy follow up identified several statistically significant associations with the risk of developing NHL, including associations that remained statistically significant for blood samples collected five or more years prior to diagnosis.

Although our main results were fairly consistent across analytical approaches, slight variations in markers chosen by a priori and secondary analyses emphasize the importance of utilizing diverse panels of immune markers in future studies seeking to characterize conditions conducive to NHL development. Furthermore, our findings suggest that even though an activated immune milieu may contribute to the development of multiple types of NHL, there is evidence of subtle differences in the pathogenesis of individual NHL subtypes, some of which had not been previously reported. Larger pooled studies will be important to more accurately identify homogeneous and heterogeneous biomarkers of risk or early disease by NHL subtype and to elucidate which are more indicative of earlier or later pathogenic changes to the immune environment.

Acknowledgments

The authors would like to thank the participants in the Nurses’ Health Study and Health Professionals Follow-up Study for their ongoing participation in the cohort studies. We thank Laura Burns for assistance with manuscript preparation, and wish to recognize the technical contributions of the Dana Farber/Harvard Cancer Center Specialized Histopathology Core Laboratory. We thank the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, and WY. The authors assume full responsibility for analyses and interpretation of these data.

Footnotes

  • Received November 7, 2017.
  • Accepted June 15, 2018.

References

  1. Takagi R, Higashi T, Hashimoto K. B cell chemoattractant CXCL13 is preferentially expressed by human Th17 cell clones. J Immunol. 2008; 181(1):186-189. PubMedhttps://doi.org/10.4049/jimmunol.181.1.186Google Scholar
  2. Hussain SK, Zhu W, Chang SC. Serum levels of the chemokine CXCL13, genetic variation in CXCL13 and its receptor CXCR5, and HIV-associated non-hodgkin B-cell lymphoma risk. Cancer Epidemiol Biomarkers Prev. 2013; 22(2):295-307. PubMedhttps://doi.org/10.1158/1055-9965.EPI-12-1122Google Scholar
  3. Breen EC, Fatahi S, Epeldegui M, Boscardin WJ, Detels R, Martinez-Maza O. Elevated serum soluble CD30 precedes the development of AIDS-associated non-Hodgkin’s B cell lymphoma. Tumour Biol. 2006; 27(4):187-194. PubMedhttps://doi.org/10.1159/000093022Google Scholar
  4. Breen EC, van der Meijden M, Cumberland W, Kishimoto T, Detels R, Martinez-Maza O. The development of AIDS-associated Burkitt’s/small noncleaved cell lymphoma is preceded by elevated serum levels of interleukin 6. Clin Immunol. 1999; 92(3):293-299. PubMedhttps://doi.org/10.1006/clim.1999.4760Google Scholar
  5. Vendrame E, Hussain SK, Breen EC. Serum levels of cytokines and biomarkers for inflammation and immune activation, and HIV-associated non-Hodgkin B-cell lymphoma risk. Cancer Epidemiol Biomarkers Prev. 2014; 23(2):343-349. PubMedhttps://doi.org/10.1158/1055-9965.EPI-13-0714Google Scholar
  6. Edlefsen KL, Martinez-Maza O, Madeleine MM. Cytokines in serum in relation to future non-Hodgkin lymphoma risk: evidence for associations by histologic subtype. Int J Cancer. 2014; 135(4):913-922. PubMedhttps://doi.org/10.1002/ijc.28724Google Scholar
  7. Saberi Hosnijeh F, Krop EJ, Scoccianti C. Plasma cytokines and future risk of non-Hodgkin lymphoma (NHL): a case-control study nested in the Italian European Prospective Investigation into Cancer and Nutrition. Cancer Epidemiol Biomarkers Prev. 2010; 19(6):1577-1584. PubMedhttps://doi.org/10.1158/1055-9965.EPI-09-1237Google Scholar
  8. De Roos AJ, Mirick DK, Edlefsen KL. Markers of B-cell activation in relation to risk of non-Hodgkin lymphoma. Cancer Res. 2012; 72(18):4733-4743. PubMedhttps://doi.org/10.1158/1538-7445.AM2012-4733Google Scholar
  9. Purdue MP, Lan Q, Bagni R. Prediagnostic serum levels of cytokines and other immune markers and risk of non-hodgkin lymphoma. Cancer Res. 2011; 71(14):4898-4907. PubMedhttps://doi.org/10.1158/1538-7445.AM2011-4898Google Scholar
  10. Purdue MP, Hofmann JN, Kemp TJ. A prospective study of 67 serum immune and inflammation markers and risk of non-Hodgkin lymphoma. Blood. 2013; 122(6):951-957. PubMedhttps://doi.org/10.1182/blood-2013-01-481077Google Scholar
  11. Purdue MP, Lan Q, Kemp TJ. Elevated serum sCD23 and sCD30 up to two decades prior to diagnosis associated with increased risk of non-Hodgkin lymphoma. Leukemia. 2015; 29(6):1429-1431. PubMedGoogle Scholar
  12. Purdue MP, Lan Q, Martinez-Maza O. A prospective study of serum soluble CD30 concentration and risk of non-Hodgkin lymphoma. Blood. 2009; 114(13):2730-2732. PubMedhttps://doi.org/10.1182/blood-2009-04-217521Google Scholar
  13. Vermeulen R, Hosnijeh FS, Portengen L. Circulating soluble CD30 and future risk of lymphoma; evidence from two prospective studies in the general population. Cancer Epidemiol Biomarkers Prev. 2011; 20(9):1925-1927. PubMedhttps://doi.org/10.1158/1055-9965.EPI-11-0396Google Scholar
  14. Gu Y, Shore RE, Arslan AA. Circulating cytokines and risk of B-cell non-Hodgkin lymphoma: a prospective study. Cancer Causes Control. 2010; 21(8):1323-1333. PubMedhttps://doi.org/10.1007/s10552-010-9560-3Google Scholar
  15. Colditz GA, Hankinson SE. The Nurses’ Health Study: lifestyle and health among women. Nat Rev Cancer. 2005; 5(5):388-396. PubMedhttps://doi.org/10.1038/nrc1608Google Scholar
  16. Hankinson SE, Willett WC, Manson JE. Alcohol, height, and adiposity in relation to estrogen and prolactin levels in postmenopausal women. J Natl Cancer Inst. 1995; 87(17):1297-1302. PubMedhttps://doi.org/10.1093/jnci/87.17.1297Google Scholar
  17. Rich-Edwards JW, Corsano KA, Stampfer MJ. Test of the National Death Index and Equifax Nationwide Death Search. Am J Epidemiol. 1994; 140(11):1016-1019. PubMedGoogle Scholar
  18. Stampfer MJ, Willett WC, Speizer FE. Test of the National Death Index. Am J Epidemiol. 1984; 119(5):837-839. PubMedhttps://doi.org/10.1093/oxfordjournals.aje.a113804Google Scholar
  19. Bertrand KA, Giovannucci E, Zhang SM, Laden F, Rosner B, Birmann BM. A prospective analysis of body size during childhood, adolescence, and adulthood and risk of non-Hodgkin lymphoma. Cancer Prev Res (Phila). 2013; 6(8):864-873. PubMedhttps://doi.org/10.1158/1940-6207.CAPR-13-0132Google Scholar
  20. Swerdlow SH, Campo E, Harris NL. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues. IARC Press: Lyon, France; 2008. Google Scholar
  21. Jaffe ES, Harris NL, Stein H, Vardiman JW. WHO/IARC Classification of Tumours. International Agency for Research on Cancer: Lyon; 2001. Google Scholar
  22. Morton LM, Turner JJ, Cerhan JR. Proposed classification of lymphoid neoplasms for epidemiologic research from the Pathology Working Group of the International Lymphoma Epidemiology Consortium (InterLymph). Blood. 2007; 110(2):695-708. PubMedhttps://doi.org/10.1182/blood-2006-11-051672Google Scholar
  23. Turner JJ, Morton LM, Linet MS. InterLymph hierarchical classification of lymphoid neoplasms for epidemiologic research based on the WHO classification (2008): update and future directions. Blood. 2010; 116(20):e90-98. PubMedhttps://doi.org/10.1182/blood-2010-06-289561Google Scholar
  24. Rosner B, Cook N, Portman R, Daniels S, Falkner B. Determination of blood pressure percentiles in normal-weight children: some methodological issues. Am J Epidemiol. 2008; 167(6):653-666. PubMedhttps://doi.org/10.1093/aje/kwm348Google Scholar
  25. Rosner B. Percentage Points for a Generalized ESD Many-Outlier Procedure. Technometrics. 1983; 25(2):165-172. https://doi.org/10.2307/1268549Google Scholar
  26. Wang M, Spiegelman D, Kuchiba A. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016; 35(5):782-800. PubMedGoogle Scholar
  27. Ng LG, Sutherland AP, Newton R. B cell-activating factor belonging to the TNF family (BAFF)-R is the principal BAFF receptor facilitating BAFF costimulation of circulating T and B cells. J Immunol. 2004; 173(2):807-817. PubMedhttps://doi.org/10.4049/jimmunol.173.2.807Google Scholar
  28. Rihacek M, Bienertova-Vasku J, Valik D, Sterba J, Pilatova K, Zdrazilova-Dubska L. B-Cell Activating Factor as a Cancer Biomarker and Its Implications in Cancer-Related Cachexia. Biomed Res Int. 2015; 2015:792187. Google Scholar
  29. Thompson N, Isenberg DA, Jury EC, Ciurtin C. Exploring BAFF: its expression, receptors and contribution to the immunopathogenesis of Sjogren’s syndrome. Rheumatology (Oxford). 2016; 55(9):1548-1555. PubMedhttps://doi.org/10.1093/rheumatology/kev420Google Scholar
  30. Smedby KE, Hjalgrim H, Askling J. Autoimmune and chronic inflammatory disorders and risk of non-Hodgkin lymphoma by subtype. J Natl Cancer Inst. 2006; 98(1):51-60. PubMedhttps://doi.org/10.1093/jnci/djj004Google Scholar
  31. Morton LM, Slager SL, Cerhan JR. Etiologic heterogeneity among non-Hodgkin lymphoma subtypes: the InterLymph Non-Hodgkin Lymphoma Subtypes Project. J Natl Cancer Inst Monogr. 2014; 2014(48):130-144. PubMedhttps://doi.org/10.1093/jncimonographs/lgu013Google Scholar
  32. Ekstrom Smedby K, Vajdic CM, Falster M. Autoimmune disorders and risk of non-Hodgkin lymphoma subtypes: a pooled analysis within the InterLymph Consortium. Blood. 2008; 111(8):4029-4038. PubMedhttps://doi.org/10.1182/blood-2007-10-119974Google Scholar
  33. Herishanu Y, Perez-Galan P, Liu D. The lymph node microenvironment promotes B-cell receptor signaling, NF-kappaB activation, and tumor proliferation in chronic lymphocytic leukemia. Blood. 2011; 117(2):563-574. PubMedhttps://doi.org/10.1182/blood-2010-05-284984Google Scholar
  34. Endo T, Nishio M, Enzler T. BAFF and APRIL support chronic lymphocytic leukemia B-cell survival through activation of the canonical NF-kappaB pathway. Blood. 2007; 109(2):703-710. PubMedhttps://doi.org/10.1182/blood-2006-06-027755Google Scholar
  35. Planelles L, Castillo-Gutierrez S, Medema JP, Morales-Luque A, Merle-Beral H, Hahne M. APRIL but not BLyS serum levels are increased in chronic lymphocytic leukemia: prognostic relevance of APRIL for survival. Haematologica. 2007; 92(9):1284-1285. PubMedhttps://doi.org/10.3324/haematol.10317Google Scholar
  36. Bojarska-Junak A, Hus I, Chocholska S. BAFF and APRIL expression in B-cell chronic lymphocytic leukemia: correlation with biological and clinical features. Leuk Res. 2009; 33(10):1319-1327. PubMedhttps://doi.org/10.1016/j.leukres.2009.03.030Google Scholar
  37. Haiat S, Billard C, Quiney C, Ajchenbaum-Cymbalista F, Kolb JP. Role of BAFF and APRIL in human B-cell chronic lymphocytic leukaemia. Immunology. 2006; 118(3):281-292. PubMedhttps://doi.org/10.1111/j.1365-2567.2006.02377.xGoogle Scholar
  38. Strati P, Shanafelt TD. Monoclonal B-cell lymphocytosis and early-stage chronic lymphocytic leukemia: diagnosis, natural history, and risk stratification. Blood. 2015; 126(4):454-462. PubMedhttps://doi.org/10.1182/blood-2015-02-585059Google Scholar
  39. Goto H, Tsurumi H, Takemura M. Serum-soluble interleukin-2 receptor (sIL-2R) level determines clinical outcome in patients with aggressive non-Hodgkin’s lymphoma: in combination with the International Prognostic Index. J Cancer Res Clin Oncol. 2005; 131(2):73-79. PubMedhttps://doi.org/10.1007/s00432-004-0600-9Google Scholar
  40. Mir MA, Maurer MJ, Ziesmer SC. Elevated serum levels of IL-2R, IL-1RA, and CXCL9 are associated with a poor prognosis in follicular lymphoma. Blood. 2015; 125(6):992-998. PubMedhttps://doi.org/10.1182/blood-2014-06-583369Google Scholar
  41. Goto N, Tsurumi H, Goto H. Serum soluble interleukin-2 receptor (sIL-2R) level is associated with the outcome of patients with diffuse large B cell lymphoma treated with R-CHOP regimens. Ann Hematol. 2012; 91(5):705-714. PubMedhttps://doi.org/10.1007/s00277-011-1363-4Google Scholar
  42. Birmann BM, Breen EC, Stuver S. Population differences in immune marker profiles associated with human T-lym-photropic virus type I infection in Japan and Jamaica. Int J Cancer. 2009; 124(3):614-621. PubMedGoogle Scholar
  43. Skibola CF, Berndt SI, Vijai J. Genome-wide association study identifies five susceptibility loci for follicular lymphoma outside the HLA region. Am J Hum Genet. 2014; 95(4):462-471. PubMedhttps://doi.org/10.1016/j.ajhg.2014.09.004Google Scholar

Data Supplements

Figures & Tables

Article Information

Vol. 103 No. 10 (2018): October, 2018 : Articles
 
DOI
https://doi.org/10.3324/haematol.2017.183236
Pubmed
29930163
Pubmed Central
PMC6165815
 
Published
2018-10-01
Published By
Ferrata Storti Foundation, Pavia, Italy
Print ISSN
0390-6078
Online ISSN
1592-8721
 

Article Usage

Online Views
1998
PDF Downloads
450

Statistics from Altmetric.com

No Data
Navigate
For Authors
For Reviewers
For Advertisers
Education
Privacy
More
Copyright © 2024 by the Ferrata Storti Foundation | Web design ￫ | Development ￫
ISSN 0390-6078 print | ISSN 1592-8721 online