Physicians' Health Study Research Articles

Physical activity before and after cancer diagnosis and mortality risk in three large prospective cohorts.

Physical activity (PA) can improve cancer survival; however, whether the timing of PA differentially affects mortality risk is unclear. We evaluated the association between PA levels pre- and post-diagnosis and mortality risk in the Women's Health Study (WHS), Physicians' Health Study (PHS)-I, and PHS-II prospective cohorts. We categorized PA pre- and post-diagnosis as active (WHS: ≥ 7.5 metabolic equivalent (MET)-h/week; PHS: vigorous PA ≥ 2-4 times/week) or inactive. We analyzed changes in pre- and post-diagnosis PA levels as four joint categories: (1) Inactive → Inactive, (2) Active → Inactive, (3) Inactive → Active, and (4) Active → Active, on mortality risk using multivariable Cox proportional hazards regression. We identified 10,541 participants with incident cancer and 3,696 deaths during follow-up. Compared to maintaining inactivity in both periods, remaining active pre- and post-diagnosis observed lower all-cause (Hazard Ratio [95% confidence interval]: WHS: 0.55 [0.47-0.64]; PHS-I: 0.77 [0.67-0.88]), cancer (WHS: 0.55 [0.45-0.67]; PHS-I: 0.75; [0.61-0.92]) and non-cancer/cardiovascular disease (CVD) mortality risks (WHS: 0.49 [0.38-0.65]). Similarly, becoming active post-diagnosis was associated with lower all-cause (WHS: 0.60 (0.48-0.75]; PHS-I: 0.72 [0.61-0.88]), cancer (WHS: 0.65 [0.49-0.86]; PHS-I: 0.64 [0.49-0.84]), and non-cancer/CVD mortality risk (WHS: 0.49 [0.33-0.75]). Being active pre- and post-diagnosis was associated with lower mortality risks in separate analyses, although significance differed by cohort and outcome. Remaining active pre- and post-diagnosis and becoming active post-diagnosis may be associated with improvements in cancer survival, however, research is needed across diverse cancer populations.

Associations Of Vitamin D-Related Biomarkers With Hypertension And The Renin-Angiotensin System In Men And Women.

Vitamin D may prevent the development of hypertension through down-regulation of renin-angiotensin system. However, epidemiologic studies assessing the interrelation of vitamin D-related biomarkers with hypertension are sparse. We examined the prospective associations between vitamin D-related biomarkers and risk of hypertension in a nested case-control study. In each of the Women's Health Study (WHS) and Physicians' Health Study (PHS) II, 500 incident hypertension cases and 500 age and race matched controls were randomly selected. Baseline plasma 25(OH)-vitamin D [25(OH)D], parathyroid hormone (PTH), and total renin concentrations were measured. Among controls, 25(OH)D and PTH were inversely correlated, but neither was correlated with total renin. In the crude model, there was a trend of association between increasing quintiles of 25(OH)D and lower risk of hypertension in women, with relative risks and 95% CIs of 1.00, 1.24 (0.84-1.83), 0.82 (0.53-1.25), 0.75 (0.48-1.16), and 0.81 (0.52-1.27) (p, trend: 0.07). Adjustment for body mass index and other hypertension risk factors eliminated this association (RR of 5th quintile: 1.03). No associations were found in men. Baseline PTH and ratio of 25(OH)D to PTH were not associated with risk of hypertension in women or men. When men and women were included in the same model, vitamin D insufficiency (defined as 25(OH)D <20ng/mL) also was not associated with increased risk of hypertension. No interactions were found across subgroups. Our study found no association of baseline plasma 25(OH)D or PTH with risk of hypertension or total renin concentration in middle-aged and older men and women.

Validation of a Blood-Based Protein Biomarker Panel for a Risk Assessment of Lethal Lung Cancer in the Physicians' Health Study.

This study aimed to assess a four-marker protein panel (4MP)'s performance, including the precursor form of surfactant protein B, cancer antigen 125, carcinoembryonic antigen, and cytokeratin-19, for predicting lung cancer in a cohort enriched with never- and ever-smokers. Blinded pre-diagnostic plasma samples collected within 2 years prior to a lung cancer diagnosis from 25 cases and 100 sex-, age-, and smoking-matched controls were obtained from the Physicians' Health Study (PHS). The 4MP yielded AUC performance estimates of 0.76 (95% CI: 0.61-0.92) and 0.69 (95% CI: 0.56-0.82) for predicting lung cancer within one year and within two years of diagnosis, respectively. When stratifying into ever-smokers and never-smokers, the 4MP had respective AUCs of 0.77 (95% CI: 0.63-0.92) and 0.72 (95% CI: 0.17-1.00) for a 1-year risk of lung cancer. The AUCs of the 4MP for predicting metastatic lung cancer within one year and two years of the blood draw were 0.95 (95% CI: 0.87-1.00) and 0.78 (95% CI: 0.62-0.94), respectively. Our findings indicate that a blood-based biomarker panel may be useful in identifying ever- and never-smokers at high risk of a diagnosis of lung cancer within one-to-two years.

Abstract 2448: Field cancerization proteomic analysis provides a novel class of functional risk biomarkers for non small cell lung cancer

Abstract Lung cancer carcinogenesis is the outcome of a field of premalignant changes that occur in the bronchial tree leading to the overt emergence of the malignant lung cancer (field cancerization). A tissue or blood proteomic signature of these premalignant changes represent a novel concept of risk biomarkers that can predict the probability of future emergence of Lung Cancer. In this study, we analyzed two independent cohorts of non-small cell lung cancer (NSCLC) patients. One cohort consisted of 18 patients who donated blood before and after tumor resection surgery and matched tissue samples (UAMS Medical Center). The second set derived from a longitudinal case-control study of plasma sample collected during the physician health study that were donated at different times prior to the diagnosis of Non-Small Cell Lung Cancer (N=272). We used affinity-enriched mass spectrometry analysis and statistical learning techniques to identify a set of common markers that discriminated plasma cases and controls and were conserved in different bronchial tissue compartments including tumor and adjacent tissue. The tumor microenvironment adjacent to and surrounding cancer cells undergoes to progressive stages of carcinogenesis during field cancerization. We identified 22 proteins involved in cell adhesion, protein folding, immune response, angiogenesis, and cancer proliferation. This is a novel class of functional risk biomarkers to determine the probability of NSCLC onset. In conclusion, our findings shed light on a new class of functional biomarkers that underly tumor pathogenesis mechanisms and can provide hints for new strategies of tumor prevention therapy or lifestyle modification. Citation Format: Rayan Ibrahim Alhammad, Ngoc Bao Vuong, Weidong Zhou, Donald J Johann Jr, Barbara Birkaya, Lance Liotta, Alessandra Luchini. Field cancerization proteomic analysis provides a novel class of functional risk biomarkers for non small cell lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2448.

Abstract 7320: Expression of circadian rhythm genes and lethal prostate cancer

Abstract Background: The circadian rhythm regulates diverse physiological and oncologic processes. We have previously shown in epidemiology studies that disruption of the circadian rhythm may be associated with increased risk of aggressive prostate cancer, including germline polymorphisms in certain circadian rhythm genes. This study sought to investigate the relationship between tumor circadian rhythm gene expression and lethal prostate cancer. Methods: We studied patients with primary prostate cancer in the Health Professionals Follow-up Study (HPFS) and Physicians’ Health Study (PHS). We created a composite score of mRNA expression using principal components that incorporated 11 circadian rhythm genes: AANAT, ARNTL, CLOCK, CRY1, CSNK1E, MTNR1A, MTNR1B, NPAS2, PER1, PER2, PER3. An age-stratified analysis was performed, and the mean differences of the score both by Gleason score and compared to normal tissue were calculated. The relationship between the score in prostate tumor tissue and lethal prostate cancer (metastases/prostate cancer-specific death) vs. indolent disease (no metastases after &amp;gt;8 years of follow-up) was assessed using logistic regression. The contribution of each gene was evaluated by leave-one-out analysis, generating scores leaving out one gene and assessing the relationship between the modified score and lethal disease. Results: Among 404 cancer patients in HPFS/PHS, there were 119 lethal events over 30 years of follow-up (median: 17 years). The circadian rhythm gene score was lower in tumor tissue compared to tumor-adjacent, histologically normal tissue (mean difference in units of standard deviation [SD] of gene score: -0.31, 95% confidence interval [CI]: -0.48 to -0.15). In addition, the score was lower in Gleason 9-10 compared to Gleason 6 tumors (mean difference in SD units: -1.05, 95% CI: -1.55 to -0.54). The three PER genes were the most strongly correlated with each other and with the score (Spearman correlation coefficients from 0.52-0.88 for the score and 0.31-0.63 between PER genes). Higher circadian rhythm gene scores were inversely associated with lethal prostate cancer (odds ratio [OR] for highest vs. lowest quartile: 0.43, 95% CI: 0.20 to 0.90) in a dose-response fashion and after adjusting for clinical and pathologic factors. The association appeared to be stronger in patients aged ≥65 years (OR for highest vs. lowest quartile: 0.21, 95% CI: 0.08 to 0.47) compared to &amp;lt;65 years at diagnosis (OR: 0.78, 95% CI: 0.29, 2.17). Results were consistent in the leave-one-out analysis for all genes except PER2 (OR for highest vs. lowest quartile: 1.00, 95% CI: 0.55 to 1.83) and PER1 (OR: 0.55, 95% CI: 0.30 to 1). Conclusion: This study examining intratumoral circadian rhythm gene expression contributes further evidence in support of the relationship between circadian rhythm disruption and aggressive prostate cancer. Tumor expression of PER1 and PER2 may be a predominant factor in the association. Citation Format: Jane Bailey Vaselkiv, Sarah C. Markt, Hannah E. Guard, Anqi Wang, Kathryn L. Penney, Konrad H. Stopsack, Lorelei A. Mucci. Expression of circadian rhythm genes and lethal prostate cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 7320.

Abstract 6143: Association of polygenic risk score and prostate tumor biomarkers with lethal prostate cancer

Abstract Prostate cancer (PCa) is a highly heritable cancer. While polygenic risk scores (PRS) stratify overall PCa risk among cancer-free men, they have not been reported to predict lethal outcomes in PCa patients. We aimed to examine to what extent PRS modifies associations between tumor biomarkers with cancer prognosis after diagnosis. We included men with primary, non-metastatic PCa in the Health Professionals Follow-up Study (HPFS) and Physicians’ Health Study (PHS). The PRS was constructed with 451 risk variants identified from prior PCa GWAS, measured on DNA from blood or buccal, standardized by adjusting for mean and standard deviation (SD) based on men without PCa with array-based genotyping data in HPFS/PHS. Prostate tumor biomarkers of molecular subtypes (TMPRSS2:ERG fusion, PTEN loss, and TP53 missense mutations from validated immunohistochemistry assays), insulin/lipid signaling pathways (FASN, IR, and IGF1-R), and cellular proliferation and tumor microenvironment (Ki-67, apoptosis rates, and angiogenesis) were measured on tumor tissue from radical prostatectomy or TURMP. Sample sizes varied by tumor biomarkers, ranging from 214 for angiogenesis markers to 598 for ERG protein expression. We first used either logistic or negative binomial regression to evaluate the association between the PRS and the protein expression (levels) of each individual biomarker, and then evaluated the addition of multiplicative interaction terms between PRS and biomarkers to logistic regression models for lethal PCa, defined as metastases or death from PCa. Lastly, we evaluated the predictive performance for lethal PCa by area under the curve (AUC). All models adjusted for age at diagnosis and Gleason score. Among all patients included in the analysis, the median age at diagnosis was 66 (interquartile range: 62-70) years, with 20% having a Gleason score≥8 and 10% having lethal PCa. PRS showed no association with lethal PCa (OR per SD increase PRS=0.95, 95%CI 0.85-1.06) or most tumor biomarkers, except for the apoptosis index (per SD in PRS corresponding to 0.22 units increase, 95%CI 0.02-0.42). Statistically significant interactions were found between PRS and p53 overexpression (p interaction = 1.6x10-5), PTEN loss (p=3.7x10-3), FASN (p=0.03), and Ki-67 (p=1.6x10-3), suggesting that compared to men with a PRS higher than the median, TP53 missense mutation, PTEN loss, higher FASN intensity, and higher Ki-67 indices were more strongly associated with lethal PCa in patients with lower PRS. Incorporating PRS with tumor biomarkers into predictive models enhanced the AUC beyond a model with tumor biomarker and age. The largest improvements were observed when combining the PRS with Ki-67 and apoptosis, showing absolute increases of 5% and 7% respectively. In conclusion, PRS integration with tumor biomarkers at diagnosis potentially improves the prediction of lethal PCa, thereby informing more targeted management strategies. Citation Format: Anqi Wang, Anna Plym, Konrad H. Stopsack, Lorelei A. Mucci, Kathryn L. Penney. Association of polygenic risk score and prostate tumor biomarkers with lethal prostate cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6143.

Germline Genetic Variants Associated with Somatic TMPRSS2:ERG Fusion Status in Prostate Cancer: A Genome-Wide Association Study.

The prostate cancer subtype defined by the presence of TMPRSS2:ERG has been shown to be molecularly and epidemiologically distinct. However, few studies have investigated germline genetic variants associating with TMPRSS2:ERG fusion status. We performed a genome-wide association study with 396 TMPRSS2:ERG(+) cases, 390 TMPRSS2:ERG(-) cases, and 2,386 cancer-free controls from the Physicians' Health Study (PHS), the Health Professionals Follow-up Study (HPFS), and a Seattle-based Fred Hutchinson (FH) Cancer Center Prostate Cancer Study. We applied logistic regression models to test the associations between ∼5 million SNPs with TMPRSS2:ERG fusion status accounting for population stratification. We did not identify genome-wide significant variants comparing the TMPRSS2:ERG(+) to the TMPRSS2:ERG(-) prostate cancer cases in the meta-analysis. When comparing TMPRSS2:ERG(+) prostate cancer cases with controls without prostate cancer, 10 genome-wide significant SNPs on chromosome 17q24.3 were observed in the meta-analysis. When comparing TMPRSS2:ERG(-) prostate cancer cases with controls without prostate cancer, two SNPs on chromosome 8q24.21 in the meta-analysis reached genome-wide significance. We observed SNPs at several known prostate cancer risk loci (17q24.3, 1q32.1, and 8q24.21) that were differentially and exclusively associated with the risk of developing prostate tumors either with or without the gene fusion. Our findings suggest that tumors with the TMPRSS2:ERG fusion exhibit a different germline genetic etiology compared with fusion negative cases.

Hyperinsulinemia and risk of lethal prostate cancer.

e17008 Background: Epidemiologic studies have found that insulin-like growth factor (IGF)-1 causally increases prostate cancer incidence and mortality, but the effect of insulin is less clear. IGF-1 and insulin act as mitogens by binding non-selectively to their receptors. To this end, this study tests the hypothesis that high circulating insulin increases risk for lethal (fatal or metastatic) prostate cancer among men with non-metastatic prostate cancer. Methods: The study population was drawn from two U.S.-based longitudinal cohorts: the Health Professionals Follow-Up Study (HPFS) of 51,529 health professionals and the Physicians’ Health Study (PHS) of 29,067 physicians. Pre-diagnostic plasma concentration of C-peptide was used as a proxy for insulin and measured on blood collected in 1993-1995 (HPFS) or 1982 (PHS) and stratified into quartiles for analysis. The empiric lifestyle index (based on body mass index (BMI), physical activity, and diet) for hyperinsulinemia (ELIH) was calculated from questionnaire data. Covariates of interest (age and fasting status at blood draw, BMI, height, clinical stage, and tumor grade) were attained from questionnaires and medical records. We followed patients for development of metastasis or death (lethal) through 2019 (HPFS) and 2015 (PHS). SAS 9.4 and R Studio were used for analyses. Chi-squared tests and Mann Whitney U tests were used to compare differences between groups, and Cox proportional hazards models were used to estimate hazard ratios (HR) and 95% CI. Results: The sample included 2,812 men with 353 lethal prostate cancer events and a mean follow-up of 164 months from diagnosis to death. The mean time from blood draw to cancer diagnosis was 8.1 years. Compared to men with non-lethal prostate cancer, those with lethal disease had significantly more severe clinical grade and stage at diagnosis, higher BMI (25.1 vs 24.6), older age at blood draw (64.5 vs 61.9 years), and higher ELIH index. There were no differences by other covariates. There was no association between pre-diagnostic c-peptide levels and cancer-specific survival in a multivariable model that included clinical stage and grade (HR, 95% CI for high vs. low quartiles: 1.17, 0.83, 1.67). High ELIH was positively associated with risk of lethal prostate cancer (HR, 95% CI: 2.41, 1.36, 4.28), even when controlling for C-peptide (HR, 95% CI: 2.18, 1.20, 3.96). Conclusions: C-peptide levels measured prior to cancer diagnosis were not associated with increased mortality from prostate cancer. Higher ELIH was associated with increased risk of lethal prostate cancer among those diagnosed with prostate cancer, which could be partly mediated by hyperinsulinemia, but other dietary, physical activity, or BMI mechanisms could contribute.

Prevalence of mismatch repair deficiency in a large prospective prostate cancer cohort.

5031 Background: Mismatch repair (MMR) deficiency and microsatellite instability (MSI) are predictive biomarkers for immunotherapy response. The best approach to identify patients with such tumors is unclear in prostate cancer. Methods: This study included men diagnosed with primary prostate cancer during prospective follow-up of the Health Professionals Follow-up Study (HPFS) and Physicians’ Health Study (PHS). The highest-grade/index lesions of tumor tissue from radical prostatectomy (95%) or transurethral resections of the prostate were mounted in triplicate on tissue microarrays. Immunohistochemistry for the MMR proteins MLH1, MSH2, MSH6, and PMS2 was performed, with scoring as MMR-deficient requiring a visible staining in a non-tumor internal positive control of the same case. For validation, a polymerase chain reaction-based MSI assay (Idylla MSI Test, Biocartis) was performed on tumor DNA of MMR-deficient cases and a selection of MMR-intact cases. Results: The study included 1016 men with prostate cancer. Tumor stage was predominantly pathologically localized (71% stage pT1/2, 18% T3a/b) with a full distribution of Gleason scores, including 20% Gleason score 6 (grade group 1), 36% 3+4 (grade group 2), 23% 4+3 (grade group 3), 8% 8 (grade group 4) and 13% 9-10 (grade group 5). MMR tumor scoring could be performed for MLH1 in 747 cases (75% of those with tumor tissue), for MSH2 in 903 cases (90%), for MSH6 in 708 cases (74%), and for PMS2 in 703 cases (72%). The remaining tumors were unevaluable due to lack of non-tumor tissue necessary as an internal positive control. Of the 903 tumors evaluable for MSH2 protein loss, 4 tumors had loss of MSH2 (prevalence 0.4%, 95% confidence interval [CI] 0.2–1.1%), and 3 of 708 evaluable tumors had concomitant loss of MSH6 (prevalence 0.4%, 95% CI 0.1–1.2%). No tumor had loss of MLH1 or PMS2. The MMR-deficient tumors had Gleason scores of 3+4, 8 (two cases), and 9–10. Tumor DNA of 1 of the 4 MMR-deficient tumors met the manufacturer’s cut-off for MSI-high; two additional MMR-deficient cases had non-zero repeats, with good reproducibility between tumor cores and technical replicates. One MSH2-deficient tumor and 6 DNA samples from 3 MMR-intact cases had repeat scores of zero. Conclusions: In this nationwide prospective study, MMR deficiency was rare in primary, surgically treated prostate cancer. This low prevalence contrasts with hospital-based studies of MMR deficiency and MSI that may have represented tumors with certain clinical features. The low prevalence and the need for an internal positive control for reliable scoring calls into question to what extent immunohistochemistry-based screening for MMR deficiency on limited tissue specimens, such as prostate biopsies, is routinely feasible.

Abstract 1449: Intratumoral vitamin D signaling and lethal prostate cancer

Abstract Background: High plasma 25(OH)-vitamin D and vitamin D supplementation may be associated with lower rates of advanced and lethal prostate cancer. We previously found that intratumoral expression of one protein, the vitamin D receptor, was associated with better prognosis among men with prostate cancer. Little data exists about long-term outcomes according to a more comprehensive profile of intratumoral vitamin D signaling. Methods: We undertook a comprehensive assessment of intratumoral vitamin D-related gene expression among men diagnosed with prostate cancer in the Health Professionals Follow-up Study (HPFS) and the Physicians’ Health Study (PHS) by creating a score based on gene expression. Of 11 literature-derived candidate genes, 6 genes had sufficient mRNA expression in tumor tissue to include in this study. The equally-weighted vitamin D pathway-related score captured genes that are upregulated (VDR, RXRA, CYP2R1, CYP27A1) and downregulated (DHCR7, CYP24A1) with activation of vitamin D signaling. Higher pathway-related scores reflected expression associated with increased vitamin D signaling. The association between the vitamin D score in prostate tumors and lethal disease (metastases/prostate cancer-specific death) vs. indolent disease (no metastases after &amp;gt;8 years of follow-up) was assessed using logistic regression. Results: Among 404 men with prostate cancer, there were 119 lethal events over up to 30 years of follow-up (median: 17 years). Correlations in expression between the six vitamin D-related genes were weak in both tumor tissue (Spearman |r|: 0.03–0.18) and tumor-adjacent histologically normal-appearing tissue (Spearman |r|: 0.00–0.24), consistent with their involvement in multiple signaling pathways. Vitamin D-related genes and the vitamin D pathway score were downregulated in tumor tissue compared to tumor-adjacent histologically normal-appearing tissue (mean difference for the score: -0.35, 95% confidence interval: -0.52 to -0.19). Higher intratumoral vitamin D signaling based on the pathway-related score was inversely associated with lethal prostate cancer even after adjusting for age at diagnosis and Gleason score (odds ratio for the highest vs. lowest quartile: 0.41, 95% CI: 0.19 to 0.86). This inverse association was similar in tumors with and without the TMPRSS2:ERG fusion as well as those with high or low vitamin D receptor protein expression. Conclusion: This study of prostate tissue vitamin D-related gene expression provides novel evidence in support of the hypothesis that more active intratumoral vitamin D signaling may lead to better prostate cancer outcomes. Citation Format: Jane Bailey Vaselkiv, Sydney T. Grob, Caroline Ericsson, Isabel Giovannucci, Irene Shui, Kathryn L. Penney, Lorelei A. Mucci, Konrad H. Stopsack. Intratumoral vitamin D signaling and lethal prostate cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 1449.

Abstract 3055: Field cancerization and microbiome in lung cancer

Abstract Lung Cancer carcinogenesis is the outcome of a field of premalignant changes that occur in the bronchial tree leading to the overt emergence of the malignant lung cancer. We sought to study the proteome of the microbiome associated with the premalignant bronchial mucosa epithelia. The overall goal is the identify functional relationships between the bronchial mucosa microbiome and the risk of developing lung cancer. We evaluated the proteome of the microbiome using mass spectrometry and specialized microbiome authentication software. This method was applied to a case-control study of plasma samples donors of physician health study who donated plasma at different times prior to the onset of Non-Small Cell Lung Cancer. We also examined bronchial mucosa tissue samples near to, and distant from, the tumor harvested from surgical lobectomy specimens. Highly significant protein biomarker differences were found in the plasma and in the tissue samples that were blindly associated with cancer. These included specific proteins from the mucosal microbiome that may have functional role in the carcinogenic process. These key diagnostic risk markers were shared between the premalignant bronchial mucosa tissue and the plasma, supporting the concept that risk markers of bronchial mucosa lung cancer carcinogenesis can be detected in the peripheral circulation. Citation Format: Rayan I. Alhammad, Ngoc B. Vuong, Weidong Zhou, Claudius Mueller, Donald J Johann Jr, Amith R. Katta, Monique V. Hoek, Fahad M. Alsaab, Alessandra Luchini, Lance Liotta. Field cancerization and microbiome in lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 3055.

The association between heart failure and incident cancer in women: an analysis of the Women's Health Initiative.

There is conflicting evidence whether heart failure (HF) is a risk factor for incident cancer. Despite population-based cohorts demonstrating this association, an analysis of the Physician's Health Study found no association in a cohort of mostly healthy males. We investigated the association of HF with incident cancer among a large cohort of post-menopausal women. A prospective cohort study of 146 817 post-menopausal women age 50 to 79 years enrolled in the Women's Health Initiative from 1993-1998, and followed through 2015. The primary exposure was adjudicated incident HF diagnosis, including preserved and reduced ejection fraction in a sub-cohort. The primary outcome was adjudicated incident total and site-specific cancers. Hazard ratios were calculated using multivariable-adjusted Cox proportional hazard regression models. Over a median follow-up of 8.4 years, 3272 and 17 474 women developed HF and cancer, respectively. HF developed in 235 women prior to cancer. HF was associated with subsequent incident cancer [hazard ratio (HR) 1.28, 95% confidence interval (CI) 1.11-1.48]. Associations were observed for obesity-related cancers (HR 1.24, 95% CI 1.02-1.51), as well as lung and colorectal cancers (HR 1.58, 95% CI 1.09-2.30 and HR 1.52, 95% CI 1.02-2.27, respectively). HF with preserved ejection fraction (HR 1.34, 95% CI 1.06-1.67), but not HF reduced ejection fraction (HR 0.99, 95% CI 0.74-1.34), was associated with total cancer. Heart failure was associated with an increase in cancer diagnoses in post-menopausal women. This association was strongest for lung cancer. Further research is needed to appreciate the underlying mechanisms responsible for this association.

Association of Prediagnostic Blood Metabolomics with Prostate Cancer Defined by ERG or PTEN Molecular Subtypes.

The TMPRSS2:ERG gene fusion and PTEN loss are two of the most common somatic molecular alterations in prostate cancer. Here, we investigated the association of prediagnostic-circulating metabolomics and prostate cancer defined by ERG or PTEN status to improve understanding of these etiologically distinct molecular prostate cancer subtypes. The study was performed among 277 prostate cancer cases with ERG status, 211 with PTEN status, and 294 controls nested in the Health Professionals Follow-up Study (HPFS) and the Physicians' Health Study (PHS). We profiled 223 polar and non-polar metabolites using LC-MS in prediagnostic plasma specimens. We applied enrichment analysis and multinomial logistic regression models to identify biological metabolite classes and individual metabolites associated with prostate cancer defined by ERG or PTEN status. Compared with noncancer controls, sphingomyelin (P: 0.01), ceramide (P: 0.04), and phosphatidylethanolamine (P: 0.03) circulating levels were enriched among ERG-positive prostate cancer cases. Sphingomyelins (P: 0.02), ceramides (P: 0.005), and amino acids (P: 0.02) were enriched among tumors exhibiting PTEN-loss; unsaturated diacylglycerols (P: 0.003) were enriched among PTEN-intact cases; and unsaturated triacylglycerols were enriched among both PTEN-loss (P: 0.001) and PTEN-intact (P: 0.0001) cases. Although several individual metabolites identified in the above categories were nominally associated with ERG or PTEN-defined prostate cancer, none remained significant after accounting for multiple testing. The molecular process of prostate carcinogenesis may be distinct for men with different metabolomic profiles. These novel findings provide insights into the metabolic environment for the development of prostate cancer.

Coffee consumption and risk of heart failure in the Physicians' Health Study

Previous studies have reported the benefits of coffee consumption on diabetes, stroke, hyperlipidemia, and coronary artery disease (CAD). However, no large-scale long-term prospective study has evaluated the relation between coffee consumption and heart failure (HF) among US population. To test the hypothesis that coffee consumption is associated with risk of HF among male physicians. We prospectively studied 20,433 middle-aged and older men from the Physicians' Health Study (PHS). Coffee consumption was assessed using a semi-quantitative food frequency questionnaire. The incidence of HF was assessed based on self-reports on annual questionnaires which were validated in a subsample using by review of medical records. We used Cox proportional hazard models to compute the hazard ratios (HR) and corresponding 95% confidence intervals (95% CI). The mean (SD) age of men was 66.4 (9.2) years. During a mean follow-up of 9.3 years, 901 new cases of HF were reported. In a multivariable Cox model adjusting for age, alcohol, smoking, and exercise, the HR (95% CI) of HF were 1.00 (reference), 1.04 (0.84-1.28), 0.90 (0.73-1.11), and 1.09 (0.91-1.30) for coffee consumption of almost never, <1 cup/day, 1 cup/day, and ≥2 cups/day, respectively (P for linear trend - 0.47). In a secondary analysis, dietary caffeine intake was not associated with HF risk: multivariable adjusted HR (95% CI) were 1.00 (reference), 1.07 (0.87-1.31), 0.95 (0.77-1.18), 1.06 (0.86-1.31), and 1.15 (0.92-1.44) across consecutive quintiles of dietary caffeine (P for linear trend - 0.34). We found no association between either coffee consumption or dietary caffeine intake with HF risk among US male physicians.

Abstract B31: Single-cell multiplex immunofluorescence of formalin-fixed, paraffin-embedded prostate cancer tissue identifies PI3K pathway activation in two prospective cohort studies

Abstract Background: The PI3K pathway is activated in a large proportion of solid cancers, including prostate cancer. Loss of its inhibitor PTEN is associated with higher risk of lethal prostate cancer. However, clinical trials of PI3K and Akt inhibitors have had mixed results in prostate cancer, likely because patients were not selected for PI3K pathway activation. Here, we established a novel assessment of PI3K pathway activity on a single-cell level. Methods: Formalin-fixed, paraffin-embedded tumor tissue from patients diagnosed with primary prostate cancer during prospective follow-up of the Health Professionals Follow-up Study (HPFS) and the Physicians' Health Study (PHS) underwent centralized re-review by genitourinary pathologists. Tissue microarrays (TMAs) were sequentially stained for PTEN, stathmin, and phospho-S6 (pS6) as indicators of PI3K activation, and for AMACR. Slides were scanned on Vectra 3 (PerkinElmer), followed by multispectral imaging analysis. Tumor cells were identified using histology-based machine learning and AMACR staining intensity. Software for visualization of histomorphology and marker colocalization was developed. Cell-level PI3K activation scores were defined using a predefined analytical approach and compared to a data-driven approach. An independent, genomically validated core-level immunohistochemistry for PTEN loss and an mRNA signature of 105 PI3K/Akt/mTOR target genes served for validation. Patients were followed prospectively for lethal disease (metastases and cancer-specific death). Odds ratios (OR) with 95% confidence intervals (CIs) were estimated using logistic regression. Results: In 1,103 patients (719 from HPFS, 384 from PHS), 1.05 million tumor cells were assessed on 14 TMAs (median, 669 tumor cells per patient; interquartile range, 208 to 1362). Visualizations revealed intratumoral heterogeneity in PI3K activation in some tumors. Predefined and data-driven approaches to quantifying PI3K activity had similar results, with cell-level PI3K scores being driven by higher stathmin and pS6 and lower cell-level PTEN intensity. PI3K scores were higher in tumors with higher Gleason grade (ptrend &amp;lt; 0.001), core-level PTEN loss (p &amp;lt; 0.001; n = 866), and a higher transcriptional signature (ptrend = 0.025; n = 295). Over a median follow-up of 12.5 years with 95 lethal events, tumors in the highest quartile of PI3K activation were more than twice as likely to become lethal (OR, 2.56; 95% CI, 1.40–4.66; ptrend &amp;lt; 0.001) than those in the lowest quartile, even after adjusting for core-level PTEN status and Gleason grade (OR, 2.27; 95% CI, 1.01–5.08). Conclusion: Single-cell multiplex assessment of immunofluorescence PI3K markers can validly identify tumors with high activity of the PI3K pathway and may harbor more information than PTEN status alone. It provides a scalable measure that can be applied to routine specimens, including archival tissue from observational studies, and that can be used for patient selection into clinical trials of PI3K inhibitors. Citation Format: Konrad H. Stopsack, Ying Huang, Svitlana Tyekucheva, Travis A. Gerke, Clyde Bango, Habiba Elfandy, Xueliang Gao, Zhenying Cai, Thomas M. Roberts, Philip W. Kantoff, Lorelei A. Mucci, Massimo Loda. Single-cell multiplex immunofluorescence of formalin-fixed, paraffin-embedded prostate cancer tissue identifies PI3K pathway activation in two prospective cohort studies [abstract]. In: Proceedings of the AACR Special Conference on Targeting PI3K/mTOR Signaling; 2018 Nov 30-Dec 8; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Res 2020;18(10_Suppl):Abstract nr B31.

THE ASSOCIATION BETWEEN HEART FAILURE AND INCIDENT CANCER IN WOMEN: AN ANALYSIS OF THE WOMEN'S HEALTH INITIATIVE

There is conflicting evidence whether heart failure (HF) is associated with incident cancer. Despite population-based cohorts demonstrating this association, no association was found in the Physician's Health Study. We aimed to study this in women. In this Women's Health Initiative cohort, 146,817

Abstract 4470: Aneuploidy drives lethal progression in prostate cancer

Abstract Purpose: Aneuploidy, defined as chromosome gains and losses, is a hallmark of cancer. However, extensive aneuploidy is relatively rare in prostate cancer. We sought to investigate whether extent of aneuploidy is associated with lethal progression in a large cohort of prostate cancer patients, an untested hypothesis. Methods: To assess chromosome arm gains and losses in archival tumor specimens, we developed a computational method based on normalized gene expression sums of whole-transcriptome profiling in primary prostate tumors from The Cancer Genome Atlas (TCGA) and validated it against DNA copy numbers. We applied this method to tumor tissue from cancer diagnosis from patients diagnosed during prospective follow-up of two cohort studies, the Health Professionals Follow-up Study (HPFS) and the Physicians’ Health Study (PHS). These tumors underwent centralized histopathologic review including Gleason grading and immunohistochemistry to characterize proliferation (Ki-67), apoptosis (TUNEL), PTEN loss, and MYC amplification. Patients were followed for lethal disease (metastases and prostate cancer-specific death) on median for 15 years. Results: Extensive aneuploidy (&amp;gt;=5 gained or lost chromosome arms) was present in 23% of the 333 primary prostate cancers from TCGA. Even 68% of low-risk cancers (Gleason grade &amp;lt;=3+4) had &amp;gt;= 1 altered chromosome arm. Our algorithm detected aneuploidy based on transcriptome profiling with high discriminative accuracy (area under the curve [AUC] for any aneuploidy, 0.83; AUC for &amp;gt;=5 chromosome arms, 0.87). Tumors with greater extent of aneuploidy had higher Gleason scores, but little to no increase in proliferative and apoptotic indices. Among the 404 patients from HPFS and PHS, 113 lethal events occurred. Greater extent of aneuploidy was strongly associated with a higher risk of lethal disease. Patients with &amp;gt;=5 predicted altered chromosome arms had 5.3 times higher odds of lethal disease (95% confidence interval, 2.2-13.1) compared to those without aneuploidy after adjusting for Gleason grade. Aneuploidy was strongly associated with lethal disease even among men with high-risk Gleason grade 8-10 tumors. Chromosome arms 8q and 10q were among the commonly altered arms. The elevated risk associated with 10q deletion and 8q gain was not considerably attenuated when accounting for PTEN loss and MYC overexpression on the respective arms. Conclusions: The extent of aneuploidy can be estimated from transcriptome profiling, which allows assessment of archival, formalin-fixed paraffin-embedded tumors. Given the complex interactions between genes colocalized on a chromosome arm, the loss or gain of an entire chromosome arm confers aggressiveness beyond affecting well-described tumor suppressors or oncogenes on that arm and beyond altering histologic tumor grade. Our results have clinical implications and point to a key role of aneuploidy in driving aggressive disease in primary prostate cancer. Citation Format: Konrad H. Stopsack, Charles A. Whittaker, Travis A. Gerke, Massimo Loda, Philip W. Kantoff, Lorelei A. Mucci, Angelika Amon. Aneuploidy drives lethal progression in prostate cancer [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 4470.

Abstract 1574: Circadian clock gene expression and lethal prostate cancer outcomes

Abstract Background: Converging evidence points to a potential role of disruption of the circadian rhythm in prostate cancer progression. There is limited human data in prostate tissue examining the biological consequences of alterations in genes involved in circadian rhythm on outcomes among prostate cancer patients. Methods: To test the hypothesis that tissue expression of circadian-related genes was associated with prostate cancer outcomes, we leveraged data from the Physicians' Health Study (PHS) and the Health Professionals Follow-Up Study (HPFS) prostate tumor cohorts. Whole transcriptome expression profiling was performed on 404 prostate cancer cases, including 113 lethal cases (metastatic disease or prostate cancer death) and 291 indolent cases (&amp;gt;8 years from diagnosis without evidence of metastasis). Using logistic regression models, we assessed whether mRNA expression levels in tumor (N=404) or paired normal prostate tissue (N=202) of eleven circadian-related genes (AANAT, CLOCK, CRY1, CRY2, CSNK1E, MTNR1B, NPAS2, OPN4, PER1, PER2, PER3) were associated with lethal prostate cancer, stage (T2 vs. T3+), Gleason grade at diagnosis (&amp;lt;8 vs. ≥8), and biomarkers of angiogenesis and apoptosis. We calculated odds ratios and 95% confidence intervals (95% CIs) per standard deviation increase in gene expression. Results: Pathway analyses showed a statistically significant association between the eleven circadian genes and lethal disease (global test p-value = 5.1e-05). On an individual gene level, men with higher tumor expression of Period Circadian Regulator 1 (PER1) gene had a reduced risk of lethal disease, independent of grade and stage (OR: 0.75, 95% CI: 0.59-0.97). High tumor expression of PER1 (OR: 0.72, 95% CI: 0.57-0.91) and Cryptochrome Circadian Regulator 2 (CRY2) (OR: 0.79, 95% CI: 0.63-1.00) were associated with lower Gleason grade tumors. High tumor expression of PER1 was associated with lower levels of cell proliferation and lower levels of angiogenesis markers; Neuronal PAS Domain Protein 2 (NPAS2) was associated with markers of apoptosis. For none of the genes was expression in normal prostate tissue associated with lethal prostate cancer (global test p-value = 0.06). Discussion: We found higher expression of several of the circadian related genes were associated with less aggressive prostate cancer features. In line with these findings, PER1 has been suggested to be a tumor suppressor in previous studies. This supports the idea that maintenance of circadian clock function may protect tumor progression. Citation Format: Sarah C. Markt, Ericka Ebot, Iona Cheng, Lynne Wilkens, Ayesha Shafi, Karen Knudsen, Kathryn Penney, Lorelei Mucci, Travis Gerke. Circadian clock gene expression and lethal prostate cancer outcomes [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 1574.

Fish and omega-3 fatty acid consumption and risk of hypertension.

Experimental studies suggest that omega-3 fatty acid have favorable effects on blood pressure (BP). However, data on the association of long-term dietary intake of omega-3 fatty acid or fish with risk of hypertension in healthy subjects are sparse. We examined whether fish or omega-3 fatty acid consumption was associated with incident hypertension in the Physicians' Health Study (PHS). In a prospective cohort study, we analyzed data on 12 279 PHS participants (mean age: 53.0 ± 8.7 years) free of hypertension at baseline. Fish and omega-3 fatty acid consumption were assessed from a baseline semiquantitative food-frequency questionnaire. Incident hypertension was ascertained via self-reports on annual follow-up questionnaires. During a mean follow-up of 15.8 years, 6299 men (51.3%) developed hypertension. In a multivariable model controlling for established risk factors for hypertension, fish and omega-3 fatty acid consumption was not significantly associated with incident hypertension. The hazard ratio (95% CI) of hypertension was 1.10 (0.93-1.30) for men who consumed at least five servings per week of fish compared with those who did not consume any fish (P for trend = 0.29). For the highest versus lowest quintile of omega-3 fatty acid intake, the hazard ratio of hypertension was 1.02 (0.94-1.11) (P for trend = 0.34). The associations did not vary by type of fish. There was also no evidence of effect modification by baseline BP, BMI, or history of hypercholesterolemia. Overall, long-term dietary intake of fish and omega-3 fatty acid was not associated with incident hypertension in a cohort of middle-aged and older US men.

Intratumoral Sterol-27-Hydroxylase (CYP27A1) Expression in Relation to Cholesterol Synthesis and Vitamin D Signaling and Its Association with Lethal Prostate Cancer.

Higher intratumoral cholesterol synthesis is associated with a worse prognosis in prostate cancer. The vitamin D-regulated enzyme sterol-27-hydroxylase (CYP27A1) converts cholesterol to 27-hydroxycholesterol, potentially lowering intracellular cholesterol levels. We hypothesized that low CYP27A1 expression is associated with high cholesterol synthesis, low vitamin D signaling, and higher risk of lethal prostate cancer. In 404 patients from the prospective prostate cancer cohorts within the Health Professionals Follow-up Study (HPFS) and the Physicians' Health Study (PHS), we assessed intratumoral CYP27A1 expression and proxies of cholesterol synthesis using transcriptome profiling, prediagnostic plasma 25-hydroxyvitamin D [25(OH)D; n = 132], and intratumoral vitamin D receptor protein expression (VDR; n = 300). Patients were followed for metastases and prostate cancer mortality (lethal cancer; median follow-up, 15.3 years). CYP27A1 expression was lower in tumors with higher Gleason grade and higher expression of cholesterol synthesis enzymes, including the second rate-limiting enzyme, SQLE. We did not detect consistent associations between CYP27A1 and 25(OH)D, VDR, or CYP24A1 mRNA expression. Lower CYP27A1 was associated with higher risk of lethal cancer in both cohorts, independent of SQLE [adjusted OR for lowest vs. highest quartile of CYP27A1, 2.64; 95% confidence interval (CI), 1.24-5.62]. This association was attenuated when additionally adjusting for Gleason grade (OR, 1.76; 95% CI, 0.75-4.17). Low CYP27A1 expression was associated with higher cholesterol synthesis and a higher risk of lethal disease. These observations further support the hypothesis that intratumoral cholesterol accumulation through higher synthesis and decreased catabolism is a feature of lethal prostate cancer.