Robinson, Dennis P.

Abstract Background: Lymphoma survivors are at risk for long term and late effects which may negatively impact quality of life (QOL). Few studies have assessed the impact of change in health behaviors (physical activity (PA), alcohol and smoking) on QOL, specifically in lymphoma survivors. Methods: Patients were prospectively enrolled <9 months of diagnosis into the Lymphoma SPORE Molecular Epidemiology Resource cohort. At baseline and 3-year follow-up (FU3), patients self-reported QOL, smoking, alcohol use and PA. At FU3, survivors reported change in health behaviors since diagnosis. The Godin Leisure Score Index (LSI) was calculated as a measure of PA (Pophali et al. ASH 2017); modeled by tertile. QOL was measured via FACT-G; the total and sub-scales [emotional (EWB), functional (FWB), physical (PWB), and social/family well-being (SFWB)] scores were transformed to a 0-100 scale. Patients who completed <80% of QOL questions were excluded. The association of QOL with health behaviors was evaluated using Kruskal-Wallis test (categorical) and Spearman's rank correlation (continuous). Results: At baseline, 2805 participants were evaluable for QOL of which 2025 were evaluable at FU3. The median age at diagnosis was 62 (IQR 52-69) years. Majority were males (N=1148, 56.8%), with advanced stage III-IV disease (N=941, 62%) and ECOG performance status <2 (N=1935, 96%). At baseline, smoking status was current (N=237), former (N=802), never (N=1310) or not available (NA, N= 456). Current smokers had a significantly lower median QOL scores compared to former and never smokers for EWB (75 vs 79 vs 79, p<0.01), FWB (71 vs 79 vs 79, p<0.01), PWB (86 vs 89 vs 89, p=0.02), SFWB (86 vs 90 vs 89, p=0.02) and total FACT-G (67 vs 82 vs 82, p<0.01). The median number of pack years smoked among current &former smokers was 15.6 (range 0.5-126). There was a significant negative correlation of pack years smoked with FWB (p=0.05), and total FACT-G (p=0.04). At FU3, current smokers had significantly lower median scores compared to former and never smokers for FWB (80 vs 86 vs 86, p=0.01), PWB (89 vs 93 vs 93, p<0.01) and total FACT-G (82 vs 86 vs 87, p<0.01). EWB (83 vs 88 vs 88, p=0.12) and SFWB (83 vs 86 vs 86, p=0.12) were also lower than never and former smokers but not statistically significant. Smoking status was decreased (N=30), increased (N=0) and unchanged (N=1439) at FU3. There was no association of change in smoking with change in QOL. At baseline, alcohol use status was current (N=1189), former (N=506), never (N=518) and NA (N=592). Current alcohol users had the highest FWB (79 vs 75 vs 75, p<0.01) vs former and never users. Former users vs current and never users had the lowest PWB (86 vs 89 vs 89, p<0.01), SFWB (88 vs 89 vs 92, p<0.01) and total FACT-G (80 vs 83 vs 82, p<0.01). EWB was similar in all 3 groups. Among current & former alcohol users, the average number of drinks/week was 3.4. There was no association of the number of drinks/week with QOL except positive correlation with PWB (p=0.01). At FU3, alcohol use was reported as increased (N=40), decreased (N=312) and no change (N=875). There was no association between change in alcohol use and change in QOL. At baseline, participants with PA in the high vs low LSI tertile had significantly better median scores for FWB (79 vs 75, p=0.01), SFWB (93 vs 89, p<0.01) and total FACT-G (83 vs 82, p<0.01). At FU3, participants in the high vs low LSI tertile had significantly better scores on all subscales and total FACT-G with a significant positive correlation between the continuous LSI and QOL (FACT-G total and subscales). PA was reported as increased (N=107), decreased (N=394) and no change (N=495). There was a significant association of increase in PA with change in PWB (p<0.01), EWB (p=0.01), FWB (p<0.01) and total FACT-G (p<0.01) from baseline to FU3, but not with change in SFWB (p=0.06). In a sensitivity analysis, QOL was modeled for health behaviors in the subset of patients with QOL assessed prior to treatment; results were consistent with the QOL in all patients. Conclusions: Health behaviors in lymphoma survivors are significantly associated with QOL: better QOL is associated with not smoking, more PA, and moderate alcohol use. Current alcohol consumption, as opposed to former alcohol use, may be related to overall health status. An increase in PA after lymphoma diagnosis is associated with an increase in the QOL. Thus, modification of health behaviors has the potential to impact QOL in lymphoma survivors. Figure. Figure. Disclosures Cerhan: Celgene: Research Funding; Nanostring: Research Funding; Jannsen: Other: Scientific Advisory Board.

Abstract Background MBL is a precursor state to CLL, with a prevalence of ~5-8% in the general population and 15-18% in CLL families. To date, 41 single nucleotide polymorphisms (SNPs) have been found to be associated with CLL. We previously showed that a PRS of the weighted average of the number of risk alleles of these SNPs is associated with CLL risk using cases and controls from the International Lymphoma Epidemiology (InterLymph) Consortium. We validated this score in an independent sample of CLL cases and controls from the Genetic Epidemiology of CLL (GEC) Consortium, a cohort of families with ≥2 members with CLL. In the CLL families we also reported an association between the PRS and MBL risk, where 93% of MBLs were low-count MBL. Here we evaluate the PRS in an independent sample ascertained from the Mayo Clinic Biobank. Methods The Mayo Clinic Biobank is a large-scale bio-repository of adult patients assembled to provide a wide array of health-related research studies. Biobank participants (N=2530) were screened for MBL using a highly-sensitive, 8-color (CD38, CD45, Kappa, Lambda, CD19, CD23, CD5 and CD20) flow-cytometry assay on stored cryopreserved peripheral blood mononucleotide cells. Low-count and high-count MBL were defined as those individuals who had <85% and ≥85 clonal B-cells out of total B-cells, respectively. Individuals without MBL ("controls") were frequency matched to MBLs based on age and sex, after excluding non-Caucasian and non-interpretable flow results. Participants were genotyped using the Illumina OmniExpress array. Standard genotyping quality-control metrics were applied. We computed the PRS as previously published and categorized it by quintiles based on values previously used with InterLymph controls. The middle quintile served as the reference category. We used logistic regression to estimate odds ratios (ORs) and 95% confidence intervals (CI), adjusting for age at consent, sex, and socioeconomic status. In addition, we compared the median PRS by low-count MBL and high count MBL using the Mann-Whitney U test. Results We identified 276 MBLs overall and 2109 controls; 97% (N=267) of MBLs had low-count MBL and only 3% (N=9) had high-count MBL. We then selected 844 controls frequency matched to the MBL cases on age and sex. Family history of leukemia/lymphoma was 11.3% among MBLs overall and 8.8% among controls. The median PRS was 7.88 and 7.46 (Figure) among MBLs overall and controls, respectively. The continuous PRS had a 1.75-fold increased risk for MBL overall (CI:1.5-2.1, P=2.65x10-12). Compared to the middle quintile, the highest quintile had 2.2-fold increased risk for MBL overall (CI:1.5-3.3, P=1.46x10-4) and the lowest quintile had 0.6-fold decreased risk (CI:0.4-0.96, P=0.03). The median PRS for low-count and high count MBL were 7.86 and 8.04, respectively. The continuous PRS had a 1.75-fold increased risk for low-count MBL (CI:1.5-2.1, P=4.81x10-12) and a 1.89-fold increased risk for high-count MBL (CI:0.8-4.2, P=0.13), however, the effect is not statistically significant between high-count and low-count MBL (P=0.89). Conclusion In this independent sample, we validated our previous findings that the CLL PRS is associated with MBL risk. Although the sample size of high-count MBL was limited (N=9), there is evidence of a weak trend towards a higher PRS compared to low-count MBL and controls Larger sample size of high-count MBL is warranted in order to stratify MBL risk by low-count and high-count MBL. These results may help identify individuals at higher risk of MBL including low-count and high-count MBL, beyond the risk associated with age and family history of CLL. Figure: Polygenic risk score distribution by control and MBL status Figure. Figure. Disclosures Parikh: Janssen: Research Funding; Pharmacyclics: Honoraria, Research Funding; Gilead: Honoraria; Abbvie: Honoraria, Research Funding; AstraZeneca: Honoraria, Research Funding; MorphoSys: Research Funding. Cerhan:Nanostring: Research Funding; Celgene: Research Funding; Jannsen: Other: Scientific Advisory Board. Kay:Janssen: Membership on an entity's Board of Directors or advisory committees; Acerta: Research Funding; Infinity Pharm: Membership on an entity's Board of Directors or advisory committees; Celgene: Membership on an entity's Board of Directors or advisory committees; Agios Pharm: Membership on an entity's Board of Directors or advisory committees; Pharmacyclics: Membership on an entity's Board of Directors or advisory committees, Research Funding; Cytomx Therapeutics: Membership on an entity's Board of Directors or advisory committees; Gilead: Membership on an entity's Board of Directors or advisory committees; Tolero Pharmaceuticals: Membership on an entity's Board of Directors or advisory committees, Research Funding; Morpho-sys: Membership on an entity's Board of Directors or advisory committees. Shanafelt:Genentech: Research Funding; Pharmacyclics: Research Funding; GlaxoSmithKline: Research Funding; Jansen: Research Funding.

Abstract Genome-wide association studies (GWAS) conducted among populations of European ancestry (EA) have identified 23 common single nucleotide polymorphisms (SNPs) associated with multiple myeloma (MM) risk. We hypothesize that the combination of these SNPs in a polygenic risk score (PRS) is likely to be a strong risk factor for MM. However, it is unclear whether the genetic variation associated with MM susceptibility also predisposes to monoclonal gammopathy of undetermined significance (MGUS). Thus, we calculated a PRS and evaluated the association with risk of MM and its precursor, MGUS. We pooled genotype data for 2434 MM and 3446 controls from ten MM GWAS of individuals of EA within the Interlymph Consortium, for 23 MM risk SNPs identified by prior GWAS. An additional 754 MGUS cases were ascertained from Mayo Clinic and MD Anderson clinical practices. To calculate the PRS, we used the risk estimates corresponding to the 23 SNP associations from the largest published MM GWAS. The log of the odds ratio (OR) for each SNP was multiplied by the respective number of risk alleles and summed to generate a PRS for each individual. The PRS was examined continuously, per one standard deviation (SD), and as quintiles, based on the PRS distribution in the controls. Associations of PRS with MM and MGUS risk were examined separately, using multivariable logistic regression assuming an additive model to assess ORs and 95% confidence intervals adjusted for age, sex, and site. We also evaluated age and sex stratified models. The distribution of sex within MM cases, MGUS cases and controls were each ~60% male and ~40% female. The median age was 61, 66, and 66 years for MM cases, MGUS cases and controls, respectively. PRS ranged from 1.52-4.91, with a median PRS of 3.21 for MM cases, 3.19 for MGUS cases, and 3.05 for controls. PRS was significantly associated with MM risk when assessed continuously (OR=1.19 per SD, p=2.2x10-16) and categorically; compared with the middle quintile (Q3), individuals in the highest quintile (Q5) had a 66% increased MM risk (OR=1.66, p=2.3x10-9) and those in the lowest quintile (Q1) had a 38% decreased MM risk (OR=0.62, p=1.3x10-6). PRS was also significantly associated with MGUS risk (OR=1.19 per SD, p=1.7x10-11); individuals with the highest PRS (Q5) had a 77% increased risk (OR=1.77, p=4.0x10-4) and those with lowest PRS (Q1) had 30% decreased risk (OR=0.70, p=0.04), compared with Q3. When stratified by age and sex, similar associations and trends were found. Using an independent sample of MM / MGUS cases and controls, we showed that a PRS constructed from 23 common genetic variants for MM risk is associated with risk of both MM and MGUS, regardless of age or sex. A future direction of this work is testing associations with PRS and clinical characteristics of the MM cases, as well as differences between MGUS cases that progress and those that do not. Our results suggest that common genetic variation may predispose to MGUS as the precursor to MM. Citation Format: Alyssa I. Clay-Gilmour, Michelle A. Hildebrandt, Nicola J. Camp, Elad Ziv, Elizabeth E. Brown, Jonathan N. Hofmann, John J. Spinelli, Graham G. Giles, Parveen Bhatti, Wendy Cozen, Xifeng Wu, Dennis P. Robinson, Aaron D. Norman, Jason P. Sinnwell, Shaji K. Kumar, S Vincent Rajkumar, Susan L. Slager, Celine M. Vachon. Associations between a polygenic risk score and risk of multiple myeloma and its precursor [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2686.

Abstract Background: Chronic lymphocytic leukemia (CLL) is a lymphoid malignancy with a strong genetic component. There are over 30 common single nucleotide polymorphisms (SNPs) associated with the risk of CLL. Moreover, in the InterLymph Subtypes Project a number of non-genetic exposures have been found to be associated with CLL, including family history (FH), height, history of atopic conditions, UV radiation, and farming exposures. However there has not been a study evaluating the joint effects among these genetic and non-genetic factors with CLL risk. Methods: Using the Mayo Clinic CLL case-control study of 587 newly diagnosed CLL cases and 790 controls, we performed analyses evaluating joint effects of genetic and non-genetic factors. For genetic effects, we computed a polygenetic risk score (PRS), a weighted averaged of the number of risk alleles across 34 SNPs, with the weights being the log of the odds ratio for each SNP. Exposure data was available for 65% of the cases and 79% of the controls. We evaluated individual and joint associations of FH of any hematological malignancy, total sun exposure categorized by quartiles based on the controls, ever living or working on a farm, any atopy, any allergies, asthma, height, and PRSs categorized by quintiles based on the controls. Multivariate logistic regression analyses were used to estimate odds ratios (ORs) and 95% confidence intervals (CIs). Results: In our preliminary results, the frequency of CLL cases in the upper PRS quintile was 49% while in the lowest quintile only 5%. The PRS has a strong evidence of association with CLL (OR= 3.04, CI =2.20-4.20 for highest versus the middle quintile). When adjusting for PRS, we still found a positive association with FH for any hematological malignancy (OR = 1.90, CI = 1.25-2.88), FH of non-Hodgkin lymphoma (OR = 2.02, CI = 1.08-3.79), and FH of leukemia (OR = 1.73, CI = 0.98-3.07). When stratifying by FH, the upper quintile of the PRS had a 5.85-fold (CI = 1.69-20.3) increased risk of CLL relative to those in the middle quintile in the FH strata, and a 2.65-fold (CI = 1.75-4.01) increased risk in the non-FH strata. After adjusting for PRS, FH and age, there remained an inverse association with sun exposure in the highest quartile (hours per week) (OR = 0.49, CI = 0.31-0.79) and a positive association with height (per 10 cm change) (OR = 1.37, CI = 1.17-1.62), but there were no associations with atopy, any allergies, history of asthma, or farming. No statistical evidence of an interaction among the variables was observed. Conclusions: We found evidence of independent effects among the genetic and non-genetic factors with risk of CLL. Among these factors, the PRS had the largest effect size. Although we did not observe any statistical interactions, larger sample sizes are warranted to fully evaluate these effects on risk of CLL. We are currently increasing our sample size through collaboration with other research groups. Final results will be presented in the meeting. This abstract is also being presented as Poster A06. Citation Format: Geffen Kleinstern, Dennis Robinson, Tim G. Call, Mark Liebow, Silvia de Sanjosé, Yolanda Benavente, James R. Cerhan, Susan L. Slager. Association of environmental risk factors, family history, and polygenic risk scores with chronic lymphocytic leukemia. [abstract]. In: Proceedings of the AACR Special Conference: Improving Cancer Risk Prediction for Prevention and Early Detection; Nov 16-19, 2016; Orlando, FL. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2017;26(5 Suppl):Abstract nr PR03.

Key Points PRS, based on the known CLL loci, predicts CLL risk with high discrimination. This PRS predicts risk of monoclonal B-cell lymphocytosis, a precursor to CLL and a condition that has clinical impact beyond risk for CLL.

Abstract Background Thirty-four single nucleotide polymorphisms (SNPs) are associated with CLL risk to-date. Moreover, family history (FH) of hematological malignancy has been consistently found to be associated with CLL, with an 8.5-fold increased risk of CLL among first-degree relatives. However, there has not been an evaluation of the interactive effects among genetic factors and FH with CLL risk. Methods We pooled data from 8 CLL case-control studies within the InterLymph Consortium (1499 CLL cases and 2601 controls). We computed a polygenic risk score (PRS), a weighted average of the number of risk alleles across the 34 SNPs, with the weights being the log of the previously reported odds ratio (OR) for each SNP. We categorized the PRS by quintiles using the cutoff points based on the distribution of all InterLymph controls (N=8228). Self-reported FH data was available for 60% of cases and 73% of controls. FH was defined as any hematological malignancy in one or more first-degree relative. Logistic regression was used to estimate ORs and 95% confidence intervals (CIs) adjusted for age, sex, socioeconomic status and study. Results The median age at diagnosis of CLL was 63 years and median age of consent was 60 years in the controls. 67% were male in CLL cases and 57% in controls. As expected, FH was associated with CLL risk (OR= 2.14, CI= 1.60-2.86). The median PRS in the cases was 0.40 and in the controls was -0.36 with the frequency of CLL cases in the upper PRS quintile as 48% while in the lowest quintile only 6%. The PRS was strongly associated with CLL risk (OR= 2.90, CI= 2.35-3.56 for upper versus middle quintile). When jointly modeling FH with PRS, a significant interaction was observed (P=0.03). When stratifying by FH, the upper quintile of the PRS had an 11.8-fold (CI= 3.97-34.8) increased risk relative to those in the middle quintile in the FH+ strata, while a 3.11-fold (CI= 2.35-4.10) increased risk was observed in the FH- strata. Conclusions Our data suggest that the PRS has a strong association with CLL risk and this association varies with FH status. Among those with FH-, the risk of CLL was 3-fold for those with many inherited variants while the CLL risk was much higher than that in those with a FH+. Studies are needed to see whether this PRS stratifies risk among those with monoclonal B-cell lymphocytosis, the CLL precursor condition that affects 5-7% of the general population. Citation Format: Geffen Kleinstern, Silvia de Sanjosé, Nicola Camp, Claire M. Vajdic, Timothy G. Call, Mark Liebow, Dennis Robinson, Neil E. Kay, Julie Cunningham, Yolanda Benavente, Alain Monnereau, John Spinelli, James R. Cerhan, Susan L. Slager. Association of polygenic risk scores and family history with the risk of chronic lymphocytic leukemia (CLL) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 4267. doi:10.1158/1538-7445.AM2017-4267

Abstract Background: Pancreatic cancer is a devastating disease for which the role of dietary factors remains inconclusive. The study objective was to evaluate risk of pancreatic cancer associated with meat preparation methods and meat-related mutagen consumption using a clinic-based case–control design. Methods: There were 384 cases and 983 controls; subjects provided demographic information and completed a 144-item food frequency questionnaire, which was used to estimate meat mutagen intake using the National Cancer Institute's CHARRED database (Bethesda, MD). Logistic regression was used to calculate ORs and 95% confidence intervals (CI), adjusted for factors including age, sex, cigarette smoking, body mass index, and diabetes mellitus. Results: Overall, the findings were null with respect to meat mutagen intake and pancreatic cancer. Conclusions: The results do not support an association between well-done meat or meat-related mutagen intake and pancreatic cancer and contrast with generally increased risks reported in previous studies. Impact: These data contribute to evidence about pancreatic cancer and potentially carcinogenic compounds in meat. Cancer Epidemiol Biomarkers Prev; 22(7); 1336–9. ©2013 AACR.

Abstract Background: Aspirin and nonsteroidal antiinflammatory drugs (NSAIDs) show indisputable promise as cancer chemoprevention agents. While aspirin and NSAID use have been suggested to lower risk of developing pancreatic cancer, the results of several epidemiological studies are not in consensus. Methods: Using a case-control design, we examined the association between aspirin, NSAID and acetaminophen use and the risk of pancreatic cancer. We enrolled 904 histological or clinically documented pancreatic ductal adenocarcinoma cases, and 1,223 age and gender matched healthy clinic-based controls evaluated at Mayo Clinic from April 2004 and September 2010. All subjects were ≥ 55 years old and self-reported by questionnaire their use of aspirin, NSAIDs, and acetaminophen at ages 41-60, along with extensive epidemiologic and demographic information. Multivariate logistic regression analysis was used to estimate risk of pancreatic cancer after adjusting for potential confounders. Results: The use of aspirin was associated with a decreased risk of pancreatic cancer. The unadjusted odds ratio (OR) associated with aspirin use ≥ 1 day/month, compared with never or &lt; 1 day/month, was 0.79 [95% confidence interval (CI) 0.65. 0.96, p = 0.015]. This effect remained significant [OR 0.71, 95% CI 0.58,0.87, p = 0.0008], after adjusting for the potential confounders of body mass index and smoking status (never, former, current). The inverse relationship was stronger among former smokers (OR 0.61, 95% CI 0.46, 0.82, p =0.0008) than never and current smokers, although there was not a statistically significant interaction (Pintereaction = 0.333). NSAID and acetaminophen use was not associated with the risk of pancreatic cancer. Conclusions: Results from this large case-control study suggest that aspirin use, but not non-aspirin NSAID use, was associated with lower risk of developing pancreatic cancer and was more pronounced among former smokers than never and current smokers. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 1902. doi:10.1158/1538-7445.AM2011-1902

Abstract Background: The role of dietary fruit, vegetable, fiber, and whole grain intake is implicated as a potential protective factor against pancreatic cancer, but studies are inconsistent. A possible explanation for the variability is that individuals who carry constitutional metabolism gene variants (minor alleles) may differentially benefit compared to those who are homozygous for the major allele, which over a life-time could lead to altered pancreatic cancer risk. Methods: We genotyped 70 SNPs that tag ten candidate metabolism genes (CAT, GAA, GCK, MT1E, UGT1A6, UGT1A7, UGT1A8, UGT1A9, UGT2B4, and UGT2B7) involved in the metabolism of components of fruits, vegetables, fibers, and whole grains to test if differential associations exist with pancreatic adenocarcinoma. We used a clinic-based case-control design, excluding participants who reported changing their diet within 5 years prior to entering the study. Three hundred eighty-four rapidly ascertained cases and nine hundred eighty-three non-cases from a primary care clinic (frequency matched on age at recruitment (+ 5 years), race, sex, and region of residence) provided blood samples for DNA and completed epidemiologic surveys plus a 144-item food frequency questionnaire developed by the National Cancer Institute. We used a dominant model to code SNPs and split dietary intake categories based on the sex-specific median intake level among non-cases. Logistic regression was used to calculate odds ratios (OR) and 95% confidence intervals, adjusted for age, sex, family-history of pancreatic cancer, smoking (never/ former/ current), body mass index, energy intake, and alcohol consumption. The reference group was defined as carriers of no minor alleles in the lowest intake category. Results: The greatest benefit was observed for carriers of no minor alleles, for rs11032702 (CAT) with higher insoluble dietary fiber intake (OR [95%CI]: 0.515 [0.37, 0.71]); for rs735670 (GCK) with higher soluble dietary intake (0.473 [0.34, 0.67]); for rs2070971 (GCK) and rs2268569 (GCK) with higher fruit intake not from citrus, melon, and berries (0.513 [0.36, 0.73] and 0.554 [0.40, 0.78]); and for rs17832252 (GCK) with higher whole grain intake (0.530 [0.37, 0.76]); all comparisons had a reference group of no minor alleles with low dietary intake and a p-value for interaction &lt;0.001. Conclusions: Inter-individual variation in metabolism of fruits, vegetables, fiber, and whole grain via metabolism gene variants may influence pancreatic cancer risk. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 3749. doi:10.1158/1538-7445.AM2011-3749