Multivariate Cox Proportional Hazards Model Research Articles

Associations Between Ultra-Processed Food Consumption and Adverse Brain Health Outcomes.

Ultra-processed foods (UPFs) are linked to cardiometabolic diseases and neurologic outcomes, such as cognitive decline and stroke. However, it is unclear whether food processing confers neurologic risk independent of dietary pattern information. We aimed to (1) investigate associations between UPFs and incident cognitive impairment and stroke and (2) compare these associations with other commonly recommended dietary patterns in the REasons for Geographic and Racial Differences in Stroke study. This prospective, observational cohort study enrolled Black and White adults in the United States from 2003 to 2007. The NOVA system was used to categorize items from a baseline food frequency questionnaire according to the level of processing. Participants with incomplete or implausible self-reported dietary data were excluded. Consumption for each category (grams) was normalized to total grams consumed. Scores quantifying adherence to a Mediterranean, Dietary Approaches to Stop Hypertension (DASH), and Mediterranean-DASH Intervention for Neurodegenerative Delay (MIND) diet were also calculated. Incident cognitive impairment was defined using performance relative to a normative sample on memory and fluency assessments. Incident stroke was identified through adjudicated review of medical records. The cognitive impairment cohort (n = 14,175) included participants without evidence of impairment at baseline who underwent follow-up testing. The stroke cohort (n = 20,243) included participants without a history of stroke. In multivariable Cox proportional hazards models, a 10% increase in relative intake of UPFs was associated with higher risk of cognitive impairment (hazard ratio [HR] = 1.16, 95% CI 1.09-1.24, p = 1.01 × 10-5) and intake of unprocessed or minimally processed foods with lower risk of cognitive impairment (HR = 0.88, 95% CI 0.83-0.94, p = 1.83 × 10-4). Greater intake of UPFs (HR = 1.08, 95% CI 1.02-1.14, p = 1.12 × 10-2) and unprocessed or minimally processed foods (HR = 0.91, 95% CI 0.86-0.95, p = 2.13 × 10-4) were also associated with risk of stroke in multivariable Cox models. The effect of UPFs on stroke risk was greater among Black than White participants (UPF-by-race interaction HR = 1.15, 95% CI 1.03-1.29, p = 1.50 × 10-2). Associations between UPFs and both cognitive impairment and stroke were independent of adherence to the Mediterranean, DASH, and MIND diets. Food processing may be important to brain health in older adults independent of known risk factors and adherence to recommended dietary patterns.

Survival after Resection of Malignant Peripheral Nerve Sheath Tumors: introducing and validating a novel type-specific prognostic model

Abstract Background This study aimed to assess the performance of currently available risk calculators in a cohort of patients with malignant peripheral nerve sheath tumors (MPNST) and to create an MPNST-specific prognostic model including type-specific predictors for overall survival (OS). Methods This is a retrospective multicenter cohort study of patients with MPNST from eleven secondary or tertiary centers in The Netherlands, Italy and the USA. All patients diagnosed with primary MPNST who underwent macroscopically complete surgical resection from 2000-2019 were included in this study. A multivariable Cox proportional hazard model for OS was estimated with pre-specified predictors (age, grade, size, NF-1 status, triton status, depth, tumor location and surgical margin). Model performance was assessed for the Sarculator and PERSARC calculators by examining discrimination (C-index) and calibration (calibration plots and observed-expected statistic; O/E-statistic). Internal-external cross-validation by different regions was performed to evaluate the generalizability of the model. Results A total of 507 patients with primary MPNSTs were included from 11 centers in 7 regions. During follow-up (median 8.7 years), 211 patients died. The C-index was 0.60 (95% CI 0.53-0.67) for both Sarculator and PERSARC. The MPNST-specific model had a pooled C-index of 0.69 (95%CI 0.65-0.73) at validation, with adequate discrimination and calibration across regions. Conclusions The MPNST-specific MONACO model can be used to predict 3-, 5-, and 10-year OS in patients with primary MPNST who underwent macroscopically complete surgical resection. Further validation may refine the model to inform patients and physicians on prognosis and support them in shared decision-making.

Thigh muscle mass evaluated by point-of-care ultrasound is associated with short-term mortality in patients with sepsis in the emergency department

Muscle mass depletion is associated with mortality and morbidity in various conditions including sepsis. However, few studies have evaluated muscle mass using point-of-care ultrasound in patients with sepsis. This study aimed to evaluate the association between thigh muscle mass, evaluated using point-of-care ultrasound with panoramic view in patients with sepsis in the emergency department, and mortality. From March 2021 to October 2022, this prospective observational study used sepsis registry. Adult patients who were diagnosed with sepsis at the emergency department and who underwent point-of-care ultrasounds for lower extremities were included. The thigh muscle mass was evaluated by the cross-sectional area of the quadriceps femoris (CSA-QF) on point-of-care ultrasound using panoramic view. The primary outcome was 28 day mortality. Multivariable Cox proportional hazard model was performed. Of 112 included patients with sepsis, mean CSA-QF was significantly lower in the non-surviving group than surviving group (49.6 [34.3–56.5] vs. 63.2 [46.9–79.6] cm2, p = 0.002). Each cm2 increase of mean CSA-QF was independently associated with decreased 28 day mortality (adjusted hazard ratio 0.961, 95% CI 0.928–0.995, p = 0.026) after adjustment for potential confounders. The result of other measurements of CSA-QF were similar. The muscle mass of the quadriceps femoris evaluated using point-of-care ultrasound with panoramic view was associated with mortality in patients with sepsis. It might be a promising tool for determining risk factors for mortality in sepsis patients in the early stages of emergency department.

Prognostic factors of patients with thyroid cancer and bone metastasis at presentation.

While bone metastases (BMs) are present in a minority of thyroid cancer (TC) patients at the time of initial diagnosis, there has been growing concern regarding their impact on life expectancy and quality of life. The aim of this study was to identify prognostic factors associated with overall survival (OS) and cancer-specific survival (CSS) in these patients and provide therapeutic recommendations based on the findings. In this retrospective cohort study, we included 82 patients diagnosed as TC with BM received treatment in our department from 2011.03 to 2023.03 (average follow-up duration was 3.02 years). The retrospective study was performed according to the inclusion and exclusion criteria. Kaplan-Meier analysis was used to estimate the OS and CSS, while the univariate and multivariate Cox proportional hazard models were employed to determine prognostic factors associated with OS and CSS. Also, 287 patients' data were collected from the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) database between 2010 and 2015 to confirm the prognostic factors identified in the retrospective study. The average survival time of the 82 patients was estimated to be 5.818 years (with a 95% confidence interval (CI) of 4.767 to 6.868 years). The cox regression analysis showed that older age (hazard ratio (HR) = 1.045, 95% CI: 1.001-1.092, P = 0.047), larger tumor size (>5cm, HR = 11.087, 95% CI: 3.728 - 32.976, P = 0.000), and the presence of extraosseous metastasis (HR = 3.247, 95% CI: 1.376 - 7.665, P = 0.007) were statistically significant factors associated with worse CSS. The results were furtherly confirmed in 287 SEER-sourced patients (age (HR = 1.020, 95% CI: 1.006 - 1.034, P = 0.006), tumor size (HR = 2.917, 95% CI: 2.044 - 4.161, P = 0.000), and extraosseous metastasis (HR = 3.726, 95% CI: 2.571 - 5.398, P = 0.000)). These results offer a population-based assessment of prognostic factors for patients with TC and BMs, revealing that age, primary tumor size (>5cm), and presence of extraosseous metastases are independent prognostic factors that correlate with worse survival. Accordingly, treatment for such patients ought to concentrate on systemic integrative therapy instead of surgical intervention.

Gustave Roussy Immune score as a prognostic biomarker in patients with platinum-refractory metastatic urothelial carcinoma treated with pembrolizumab: YUSHIMA study.

This study aimed to investigate the prognostic value of the Gustave Roussy Immune score (GRIm-score) in platinum-refractory metastatic urothelial carcinoma (UC) treated with pembrolizumab. This multicenter retrospective study (YUSHIMA study) evaluated 331 patients with metastatic UC treated with pembrolizumab after platinum-based chemotherapy between January 2018 and June 2023 at 13 institutions. We collected pretreatment variables, including the GRIm-score based on serum albumin, lactate dehydrogenase, and neutrophil-to-lymphocyte ratio. The patients were divided into low and high GRIm-score groups. Prognostic factors for overall survival (OS) and progression-free survival (PFS) were determined using the multivariate Cox proportional hazard model. During the median follow-up period of 7.3months, 278 (84%) patients showed disease progression, and 223 (67%) died from any cause. Multivariate analysis revealed that the high GRIm-score group was an independent and significant adverse prognostic factor of both OS and PFS (hazard ratio, 1.65 and 1.82, respectively; both p < 0.001) along with Eastern Cooperative Oncology Group Performance Status of ≥ 2 (both p < 0.001), presence of visceral metastasis (both p < 0.001), and hemoglobin of < 9.2g/dL (p = 0.030 and p = 0.038). C-reactive protein of > 42mg/L was a significant prognostic factor for OS (p = 0.001). The GRIm-score is an independent prognostic marker for survival outcomes in patients with platinum-refractory metastatic UC treated with pembrolizumab.

Proposal for an optimised definition of adverse pathology (unfavourable histology) that predicts metastatic risk in prostatic adenocarcinoma independent of grade group and pathological stage.

Histological grading of prostate cancer is a powerful prognostic tool, but current criteria for grade assignment are not fully optimised. Our goal was to develop and test a simplified histological grading model, based heavily on large cribriform/intraductal carcinoma, with optimised sensitivity for predicting metastatic potential. Two separate non-overlapping cohorts were identified: a 419-patient post-radical prostatectomy cohort with long term clinical follow-up and a 209-patient post-radical prostatectomy cohort in which all patients had pathologically confirmed metastatic disease. All prostatectomies were re-reviewed for high-risk histological patterns of carcinoma termed 'unfavourable histology'. Unfavourable histology is defined by any classic Gleason pattern 5 component, any large cribriform morphology (> 0.25 mm) or intraductal carcinoma, complex intraluminal papillary architecture, grade 3 stromogenic carcinoma and complex anastomosing cord-like growth. For the outcome cohort, Kaplan-Meier analysis compared biochemical recurrence, metastasis and death between subjects with favourable and unfavourable histology, stratified by pathological stage and grade group. Multivariable Cox proportional hazards models evaluated adding unfavourable histology to the Memorial Sloan Kettering Cancer Center (MSKCC) post-prostatectomy nomogram and stratification by percentage of unfavourable histology. At 15 years unfavourable histology predicted biochemical recurrence, with sensitivity of 93% and specificity of 88%, metastatic disease at 100 and 48% and death at 100 and 46%. Grade group 2 prostate cancers with unfavourable histology were associated with metastasis independent of pathological stage, while those without had no risk. Histological models for prediction of metastasis based on only large cribriform/intraductal carcinoma or increasing diameter of cribriform size improved specificity, but with lower sensitivity. Multivariable Cox proportional hazards models demonstrated that unfavourable histology significantly improved discriminatory power of the MSKCC post-prostatectomy nomogram for biochemical failure (likelihood ratio test P < 0.001). In the retrospective review of a separate RP cohort in which all patients had confirmed metastatic disease, none had unequivocal favourable histology. Unfavourable histology at radical prostatectomy is associated with metastatic risk, predicted adverse outcomes better than current grading and staging systems and improved the MSKCC post-prostatectomy nomogram. Most importantly, unfavourable histology stratified grade group 2 prostate cancers into those with and without metastatic potential, independent of stage. While unfavourable histology is driven predominantly by large cribriform/intraductal carcinoma, the recognition and inclusion of other specific architectural patterns add to the sensitivity for predicting metastatic disease. Moreover, a simplified dichotomous model improves communication and could increase implementation.

Hepatic steatosis, metabolic dysfunction and risk of mortality: findings from a multinational prospective cohort study.

Non-alcoholic fatty liver disease (NAFLD) and metabolic syndrome (MetS) are implicated in the aetiology of non-communicable diseases. Our study aimed to evaluate associations between NAFLD and MetS with overall and cause-specific mortality. We used dietary, lifestyle, anthropometric and metabolic biomarker data from a random subsample of 15,784 EPIC cohort participants. NAFLD was assessed using the fatty liver index (FLI) and MetS using the revised definition. Indices for metabolic dysfunction-associated fatty liver disease (MAFLD) were calculated. The individual associations of these indices with overall and cause-specific mortality were assessed using multivariable Cox proportional hazards models to estimate hazard ratios (HRs) and 95% confidence intervals (95%CIs). As a subobjective, risk associations with adaptations of new classifications of metabolic dysfunction-associated steatotic liver disease (MASLD) and metabolic and alcohol-related liver disease (MetALD) were also assessed. Among the 15,784 sub-cohort participants, a total of 1997 deaths occurred (835 due to cancer, 520 to CVD, 642 to other causes) over a median 15.6 (IQR, 12.3-17.1) years of follow-up. Compared to an FLI < 30, FLI ≥ 60 was associated with increased risks of overall mortality (HR = 1.44, 95%CI = 1.27-1.63), and deaths from cancer (HR = 1.32, 95%CI = 1.09-1.60), CVD (HR = 2.06, 95% CI = 1.61-2.63) or other causes (HR = 1.21, 95%CI = 0.97-1.51). Mortality risk associations were also elevated for individuals with MAFLD compared to those without. Individuals with MetS were at increased risk of all mortality endpoints, except cancer-specific mortality. MASLD and MetALD were associated with higher risk of overall mortality. Our findings based on a prospective cohort suggest that individuals with hepatic steatosis or metabolic dysfunction have a higher overall and cause-specific mortality risk.

The relationship between cardiometabolic abnormalities and mortality among women with and without cancer in the Women’s Health Initiative (WHI).

e24007 Background: Prior studies of women with obesity-associated cancers in the Women’s Health Initiative (WHI) showed higher overall and mortality due to cardiovascular disease (CVD) for women with a higher number of cardiometabolic risk factors at baseline. It is unclear if this risk differs from women without cancer. In order to determine if this relationship differs among both groups, we compared the association between cardiometabolic risk factors and mortality among women in the WHI with and without cancer. Methods: Women with one of five early-stage obesity-associated cancers (breast, colorectal, endometrial, ovarian, non-Hodgkin lymphoma), and matched controls without history of cancer, were selected from the WHI-Life and Longevity After Cancer (LILAC) study. Cardiometabolic risk factors included high waist circumference (WC) (≥88 cm), systemic hypertension ( &gt; 130/85 mm Hg) measured at study visits, and self-reported history of diabetes and elevated cholesterol. Multivariable Cox proportional hazards models (all-cause mortality) and Fine-Gray models (CVD and non-CVD mortality) were used to calculate hazard ratios (HR) and 95% confidence intervals (CI) to evaluate the association between baseline cardiometabolic risk factors and survival outcomes for the cancer and non-cancer cohorts in separate models. Sensitivity analyses included women in the cancer-cohort who received cardiotoxic chemotherapy. Results: A total of 7,491 and 35,508 women were analyzed in the cancer and non-cancer cohorts respectively. CVD was the most common cause of death (excluding cancer) in both non-cancer (41.4%) and cancer (41.5%) cohorts. Adjusted Cox models showed that the proportion of deaths due to any cause, by number of cardiometabolic risk factors, was significantly higher in the non-cancer compared to the cancer cohort (p-value = 0.02), with the HR values being higher for those with 3-4 risk factors compared to 1-2. Similar results were seen for each individual risk factor except for WC. Regarding CVD and non-CVD mortality, in competing risks models, there were no significant differences in the risk of death by number of risk factors, between the cancer and non-cancer cohorts (p = 0.21 and 0.22) respectively. Sensitivity analyses for women who received a cardiotoxic regimen, (81.5% anthracyclines alone, 13.4% anti-HER2 alone, and 5.1% both) showed no differences in the results for all-cause mortality, and CVD mortality. However, for non-CVD mortality, the proportion of deaths was significantly higher for women in the non-cancer cohort (p = 0.03). Conclusions: Cardiometabolic risk factors had a greater impact on mortality among women without cancer compared to those with cancer, except for death due to CVD which was similar between groups, suggesting that CVD risk reducing interventions are of equal importance for women with and without cancer.

Clinical and molecular features and survival in thyroid cancer with brain metastases.

2023 Background: Brain metastases (BM) in advanced thyroid cancer are rare; however, they may be underreported as staging brain imaging is not routine. Prognostic features and molecular alterations for this group are poorly described. We characterized the clinical and molecular features of thyroid cancer with BM and evaluated differences in survival. Methods: In this single-center retrospective analysis, patients with metastatic radioactive iodine refractory well-differentiated, poorly-differentiated, or medullary thyroid carcinoma seen at the Princess Margaret Cancer Center from 2007-2022 were included. Electronic medical records were reviewed to collect clinicopathologic and treatment data. All patients had tumor next-generation sequencing (NGS) by a targeted gene panel. Overall survival (OS) from initial medical oncology consultation was analyzed using the Kaplan-Meier method and univariate and multivariate Cox proportional hazards models. Results: Of 248 patients with advanced thyroid cancer, 41(17%) were diagnosed with BM. The median interval from thyroid cancer diagnosis to BM development was 6.49 years (y) (IQR 3.51- 14.35 y). The median age at the time of BM diagnosis was 63.6 y and 23 (56%) were male. Most (76%) were asymptomatic at the diagnosis of BM and 80% had an ECOG of 0-1. Among patients with well-differentiated histology, 27 (18%) developed BM, while 7 (12%) with poorly-differentiated and 7 (19%) with medullary thyroid cancer developed BM. The most common molecular alterations included BRAFV600E (n=19), TERT (n=11), RAS (n=9), RET (n=8), and FANCA (n=3). Patients were treated with surgery (n=5), stereotactic radiation (n=21), whole brain radiation therapy (n=12), or a combination of both (n=2). OS was similar in patients with BM compared to those without (6.05 vs 6.45 y). There were no differences in histology or molecular alterations between patients that developed BM compared to those who did not. The median survival from the time of BM was 3.23 y. Patients with BM with ≥4 lesions had worse survival (0.52 vs 5.09 y, p=0.01). There were no differences in survival based on histology, symptoms, treatment or molecular alterations among those with BM. Conclusions: This is one of the largest reported thyroid BM cohorts and the first with complete NGS. The prevalence of BM in our cohort was higher than previously described, often diagnosed incidentally. Screening for BM in advanced thyroid cancer may detect higher rates, offering prognostic insights. While the number of BM was predictive of survival, molecular alterations did not predict BM development or survival and larger studies are needed.

The survival benefits and the risk of severe renal injury of immune checkpoint inhibitors in patients with cancer: The real-world evidence from the National Clinical Cohort Collaborative (N3C).

e24142 Background: The survival benefits and the risks of severe immune-related adverse events (sirAEs) of Immune checkpoint inhibitors (ICI) vary dramatically among cancer patients. Among sirAEs, acute kidney injury (AKI) incidences are rising, often cause sustained renal damage, and are responsible for the deterioration of patients’ overall health and quality of life. There is an urgent need to support oncologists and nephrologists in choosing the right ICI drugs for the right patients. We aim to comprehensively scrutinize the tradeoff between the survival benefits and the severe AKI risk for ICI patients by leveraging the national electronic medical records (EMR) from the National Clinical Cohort Collaborative (N3C) registry. Our work provides real-world evidence for clinical decisions on ICI use. Methods: Cancer patients treated with ICI are followed up for at least 6 months using N3C data (01/01/2018 - 01/19/2024). Primary outcomes included severe AKI within 6 months of initiation or mortality up to the last follow-up. We fit multivariable Cox proportional hazards models to evaluate the effect of various ICI types, adjusting for age, gender, race, cancer types, smoking status, and comorbidities. Results: Our study cohort included 55,767 patients who received an ICI, with 25,197 deaths and 6,458 severe AKI events. The most common ICI used were PD-1 inhibitors (74.3%), PD-L1 inhibitors (16.6%), CTLA-4 inhibitors (0.3%), and combination treatments of PD-1 and CTLA-4 inhibitors (8.8%). Observed long-term severe AKI incidences exceeded prior reports, with 96%, 92%, 88%, and 71% at 1, 3, 6, and 36 months. Survival benefits of PD-1, CTLA-4 and their combinations versus PD-L1 inhibitors were comparable (HR: 0.94 [95%CI: 0.91–0.97]; HR: 0.81 [95%CI: 0.63–1.02]; HR: 1.09 [95%CI: 1.04–1.15]). Severe AKI risks were higher with CTLA-4 and combination therapy (HR: 1.48 [95%CI: 0.96–2.28]; HR: 1.96 [95%CI: 1.78–2.16]) compared to PD-L1 inhibitors, with PD-1 inhibitors also showing increased risk (HR: 1.15 [95%CI: 1.07–1.23]). We also scrutinize risk factors associated with significant AKI risk, including a history of AKI, congestive heart failure, dementia, complicated diabetes, hypertension, liver disease, other immunocompromised diseases, and cancers of the gynecologic type, prostate cancer, and gastrointestinal cancers, and being of Black race. Conclusions: Our work provides up-to-date and comprehensive real-world evidence of the survival benefits and severe renal injury risks for clinical decisions of ICI therapy. PD-L1 inhibitors are associated with a lower risk of severe AKI compared to PD-1, CTLA-4 inhibitors, and their combination. Future research should focus on identifying patient subgroups that may benefit from specific ICI therapies while minimizing adverse outcomes.

Impact of DNA damage repair alterations on real-world response to therapy in metastatic colorectal cancer.

e15557 Background: Alterations in the DNA damage repair (DDR) pathway genes contribute to tumor development. These alterations can be exploited by cytotoxic chemotherapy regimens that induce DNA damage. Platinum chemotherapy agents have been shown to be particularly effective in breast and ovarian cancer patients with alterations in the DDR pathway. However, the clinical impact of DDR alterations in metastatic colorectal cancer (mCRC) is still poorly understood. In this study, we utilize a large national database to evaluate clinical outcomes in patients with mCRC and alterations in DDR pathway genes. Methods: We reviewed 5602 patients that received standard fluorouracil based first-line chemotherapy regimens for mCRC using the nationwide Flatiron Health EHR-derived de-identified database. Chemotherapy regimens were categorized as a platinum-based doublet (FOLFOX, CAPOX), irinotecan-based doublets (FOLFIRI), or triplet (FOLFOXIRI). Overall survival (OS) was summarized within groups via Kaplan-Meier survival estimates. OS was compared between groups in the context of univariable and multivariable Cox proportional hazards models controlling for age, gender, DDR status, stage at diagnosis, ECOG, RAS/RAF status, MMR status, albumin and CEA at diagnosis. OS was defined as the time in months between date of first line therapy initiation to death. Results: DDR gene alterations were indentified in 35.7% (2001) patients.The most frequent mutations were in ATM (13.3%) and BRCA2 (8.3%). There was no significant difference in OS based on DDR mutation status. Within DDR altered patients, there was no difference in OS between those who received a platinum-based doublet, irinotecan-based doublet or triplet first-line therapy. Time to next treatment or death (TTNTD) was significantly improved with an irinotecan-containing doublet regimen and triplet regimen compared to a platinum-based doublet in a multivariable model (HR 0.79, p = 0.001 and HR 0.68, p = 0.002, respectively). There was no difference in OS in DDR-mutated patients that received an irinotecan-based regimen in the first line followed by an oxaliplatin-based regimen in the second line or the reverse sequence. When evaluating the impact of commonly co-mutated genes on survival we observed a worse OS in patients with a KRAS mutation in addition to a DDR mutation (HR 1.19, p = 0.045). An APC mutation was associated with an improved OS in the entire cohort, regardless of DDR status (HR 0.76, p &lt; 0.001). Conclusions: In this real-world population, approximately one third of patients with mCRC harbored an alteration in a DDR pathway gene. In these patients, TTNTD was greater with irinotecan-containing first line therapies (doublet or triplet). However, OS was similar regardless of which first-line chemotherapy was used.

A plasma-based proteomic platform for predicting clinical benefit from immune checkpoint inhibitors in multiple cancers.

2568 Background: Immune checkpoint inhibitor (ICI)-based therapies are preferred treatment options for multiple cancer types. However, there is an unmet need for predictive tests that can identify patients likely to benefit. PROphet-NSCLC is a commercially available pretreatment plasma proteomic test that predicts clinical benefit from first-line PD-(L)1 inhibitor-based therapy in patients with metastatic non-small cell lung cancer (NSCLC), guiding the choice between ICI monotherapy versus ICI-chemotherapy. Here, we evaluate the broader clinical utility of PROphet-NSCLC by testing its predictive capabilities across diverse cancer types. Methods: Pre-treatment plasma samples and clinical data were collected for an observational study from patients with metastatic melanoma (n=68) and HPV-related cancers (n=43; NCT03427411) treated with PD-(L)1 inhibitor-based therapies. HPV-related cancers included anogenital squamous cell carcinoma (n=16), cervical carcinoma (n=18), and head and neck squamous cell carcinoma (n=9). Patients were assigned a PROphet-positive or -negative result based on computational analysis of SomaScan-derived proteomic profiles in pre-treatment plasma samples. Overall survival (OS) and progression-free survival (PFS) in PROphet-positive- and -negative groups was analyzed with the Kaplan-Meier method. Hazard ratios (HR) were calculated from univariate and multivariate Cox proportional hazard models. Results: In the melanoma cohort, PROphet-positive patients (n=51) displayed a significant OS benefit in comparison to PROphet-negative patients (n=17; median OS 92.8 months vs. 9.5 months, HR=0.14, 95% CI: 0.06-0.34, p&lt;0.0001), consistent with the clinically validated predictive performance of the test in patients with NSCLC. OS results remained significant after correcting for sex, age, histology, ECOG performance status, and treatment type (HR=0.05, 95% CI: 0.01-0.25, p&lt;0.001). In PROphet-positive patients, PD-1+CTLA-4 inhibitor combination therapy (n=27) was superior to PD-1 inhibitor monotherapy (n=24; OS: median 118.4 vs. 48.3 months, HR=0.58, p=0.24; PFS: median not reached vs. 10.8 months, HR=0.48, p=0.04). PROphet-negative patients displayed similarly poor outcomes with either treatment, providing a rationale to consider alternative therapies for such patients. In the HPV-related cohort, the median OS in PROphet-positive (n=10) vs PROphet-negative (n=33) groups was 43.6 vs 4.4 months (HR=0.22, 95% CI: 0.08-0.59, p=0.001), with similar trends per sub-cohort. Conclusions: Our findings show that the PROphet-NSCLC test can be applied to PD-(L)1 inhibitor-treated melanoma and HPV-related cancers, suggesting applicability to cancers beyond NSCLC. Analysis of additional patient samples is needed to explore the potential utility of PROphet-NSCLC for informing treatment decisions for a broad range of cancer types.

Historic redlining, modern neighborhood structures, and mortality in children, adolescents, and young adults with cancer.

11037 Background: Historic redlining, referring to the Home Owners Loan Corporation (HOLC) program’s racially biased risk monitoring maps in the 1930’s, has indelible impacts on modern day neighborhood characteristics and health outcomes. The impact of redlining, a form of structural racism, on children, adolescents, and young adults with cancer (CAYACs) is unknown. This retrospective cohort study evaluates the association between redlining and mortality in CAYACs. Methods: Using the California Cancer Registry, we identified cases &lt;25 years old diagnosed with first primary malignant cancer between 2000-2019. Using merged HOLC maps and U.S. Census data we determined case redlining status (A “Best”, B “Still Desirable”, C “Declining”, D “Hazardous”, and N “Not graded”) by residence at diagnosis. We defined residents of D neighborhoods redlined-exposed and residents of A or B neighborhoods unexposed or referent. Kaplan Meier methods and multivariable Cox proportional hazard models were applied to calculate overall survival. Additional models accounted for census tract-level neighborhood SES (nSES, a composite measure incorporating seven SES domains), neighborhood racial and ethnic composition (or typology) and racialized economic segregation (via index of concentration at the extremes). Results: In total 8,210 cases were HOLC-graded, the median age was 16 years, 46% were female, and 48% had solid tumors. Overall survival at 5 years was inferior for D cases vs. A or B cases (OS 80.2%, 95% CI: 78.5-81.7 vs. OS 88.2, 95% CI: 84.0-91.3). Adjusting for individual clinical characteristics (age, sex, year of diagnosis, and cancer stage), D cases experienced greater mortality (HR 1.32, 95% CI: 1.12-1.55) compared to A or B cases. Additional adjustment for insurance and nSES attenuated the relationship (HR 1.17, 95%CI: 1.00-1.36 and HR 1.07, 95% CI: 0.92-1.24 respectively). Similar patterns were observed in models adjusting for racialized economic segregation (HR 1.07, 95% CI: 0.92-1.24) and typology (HR 1.10, 95% CI: 0.94-1.28) in place of nSES. Conclusions: In this population-based study, redlined-exposed CAYACs experienced greater mortality compared to unexposed CAYACs. Adjusting for insurance and contemporary nSES attenuated the association between redlining and mortality. Findings suggest current neighborhood attributes may be downstream manifestations of structural racism with survival implications for CAYACs.

The impact of body mass index (BMI) on overall survival (OS) among patients receiving immune checkpoint inhibitors (ICIs): A population-based study.

12041 Background: While obesity is a risk factor for cancer, BMI has been previously identified as a potential prognostic marker in different solid tumors. Prior studies have identified that among cancer survivors receiving ICIs, higher BMI may be associated with better OS, but there have been heterogeneous results among studies. Here we use population-level administrative data to evaluate the association between BMI and OS among patients receiving ICIs. Methods: We used administrative data deterministically linked across databases to identify a cohort of solid tumors patients initiating ICI therapy in Ontario, Canada from June 2012 to October 2018 and obtained information on socio-demographics including BMI at start of ICI, clinical covariates, and OS. We applied multivariable Cox proportional hazards models to evaluate the impact of BMI on OS, adjusting for sex, age, cancer center, autoimmune history, recent hospitalization and comorbidity score. Subgroup analyses were performed based on disease site and sex. Results: Among 4863 patients, median age was 67, 57% male; 46% had lung cancer, 35% melanoma, 9% renal cancers; 40% received nivolumab, 36% pembrolizumab, 17% ipilimumab. Median BMI was 26.1, with 3% low, 29% normal, 27% overweight, 19% obese. Median OS 317 days. Overall, greater BMI was associated with better OS (aHR=0.98 per unit, 95% CI [0.97-0.99] p&lt;0.001). When compared to normal BMI, obese (aHR=0.77 [0.70-0.85] P&lt;0.001) and overweight patients (aHR=0.85 [0.78-0.93] p&lt;0.001) had better OS while those with low BMI had poorer OS (aHR=1.39 [1.16-1.66] P&lt;0.001). Among melanoma patients, those who were obese had better OS (aHR=0.84 [0.71-0.98] p=0.03) and low BMI patients had poorer OS (aHR=1.80 [1.20-2.69] p=0.004) when compared to normal BMI. For lung and renal patients, increased BMI was associated with better OS when BMI was evaluated continuously (aHRlung=0.99 per unit [0.98-1.00] p=0.05; aHRrenal=0.98 per unit [0.95-0.99] p=0.04), but no significant associations were observed when BMI was evaluated categorically (p&gt;0.05). Among males, patients who were obese (aHR=0.70 [0.62-0.80] p&lt;0.001) and overweight (aHR=0.80 [0.72-0.90] p&lt;0.001) had better OS compared to those with normal BMI, while those with low BMI had poorer OS (aHR=1.58 [1.22-2.06] p&lt;0.001). However, among females, low BMI were associated with poorer OS (aHR=1.29 [1.01-1.65] p=0.04) compared to normal BMI, while no significant associations with OS were observed for overweight (aHR=0.93,p=0.30) or obese (aHR=0.89, p=0.12) patients. Conclusions: BMI was identified as a potential prognostic factor among cancer survivors receiving ICIs where greater BMI was associated with better OS. This association varied by cancer type and sex and is particularly notable in melanoma and among males. Further studies understanding these prognostic associations are warranted.

Prevalence of obesity and its impact on outcome in children diagnosed with cancer in Canada: A population-based study.

10517 Background: Childhood obesity can result in a variety of adverse health outcomes. Objectives were to describe the prevalence of obesity and determine the association between baseline obesity with event-free survival (EFS) and overall survival (OS) in children diagnosed with cancer in Canada. Methods: We conducted a retrospective cohort study using the Cancer in Young People in Canada database, a population-based surveillance program collecting data from all cancer patients &lt; 15 years old before 2015 and &lt; 19 years old after 2015. This study included all children with newly diagnosed cancer aged 2 to 18 years across Canada from 2001 to 2020. Obesity was defined as age and sex-adjusted body mass index ≥ 95th percentile. EFS was defined as the time from diagnosis to first event (relapse, progression, secondary malignancies, or death). Univariate and multivariable Cox proportional hazards models compared EFS and OS between patients with and without obesity. The prevalence of obesity was compared annually from diagnosis. All analyses were stratified by cancer categories. Results: A total of 11,291 patients were included (37.1% leukemias, 14.5% lymphomas, 21.8% central nervous system (CNS) tumors, 26.6% non-CNS solid tumors; median age: 7.6 years [interquartile range (IQR): 4.2-12.5] and median follow-up time: 4.7 years [IQR: 2.6-5.1]). At diagnosis, 10.5% were obese and the prevalence of obesity was significantly higher at 1 year (15.7%, p &lt; 0.001) and 2 years (20.1%, p &lt; 0.001) after diagnosis. The prevalence of obesity was higher in the leukemia and lymphoma cohort compared to other cancer categories (11.4% vs. 9.5%, p = 0.001). In multivariable models controlling for age at diagnosis, sex, ethnicity, income quintile, treatment era and cancer categories, obesity at diagnosis remained significantly associated with inferior EFS [adjusted HR (aHR) 1.16, 95% CI 1.02–1.32] and OS [aHR 1.29, 95% CI 1.11–1.49]. In children with acute lymphoblastic leukemia (ALL) (n = 3458, 82.5% of all leukemias), after adjusting for high-risk features such as age, white blood cell count and CNS status, obesity at diagnosis remained associated with inferior EFS [aHR 1.55, 95% CI 1.17–2.04] and OS [aHR 1.75, 95% CI 1.23–2.49]. In patients with CNS tumors, obesity at diagnosis was also associated with inferior EFS [aHR 1.38, 95% CI 1.09–1.76] and OS [aHR 1.47, 95% CI 1.13–1.91]. Obesity at diagnosis did not confer adverse outcomes in other cancer categories. Conclusions: In this population-based study of children with cancer, the prevalence of obesity was 11% at diagnosis and continued to increase in the following 2 years. Obesity at cancer diagnosis was independently associated with inferior survival across the entire cohort, especially in children with ALL and CNS tumors. Further studies are needed to better characterize the relationship between obesity and survival in children with cancer to inform future trial design.

Association between antibiotic use and mortality among older persons receiving first-line immune checkpoint inhibitors for non-small cell lung cancer (NSCLC).

e14607 Background: There is biologic plausibility that antibiotics (ABX) may reduce the effectiveness of cancer therapy, most notably immune checkpoint inhibitors (ICI). As ABX are commonly prescribed to older persons and more than a third of US persons with cancer are eligible for ICIs, use of ABX has important population health implications. We assess the association between ABX use and mortality among persons with NSCLC who receive first-line (1L) treatment. Methods: Using SEER-Medicare, we assembled a cohort of persons ≥66 years with NSCLC who received 1L treatment with platinum-based chemotherapy (PLT) or ICI (PLT without ICI, PLT with ICI, and ICI monotherapy) within one year of cancer diagnosis between 2017-2019. We assessed overall ABX exposure from 1 month to 1 year prior to treatment initiation and by ABX class. Primary outcome was all cause deaths within three years of treatment initiation or until 12/31/2020. Multivariable Cox proportional hazard models adjusting for individual-level characteristics (demographics, time-to-treatment, comorbidities, healthcare use, SEER registry, and neighborhood socioeconomic status) and landmark analyses of 1-6 months after cancer treatment were used to assess the association between ABX exposure and mortality. Results: Among 4,182 persons (mean age 75 years, 85% white, 68% chemo-immunotherapy within 2 months of NSCLC diagnosis), 63%, 18%, and 18% received PLT, PLT+ICI, and ICI, respectively. ICI+/-PLT increased in prevalence from 20% of treatments (2017) to 55% (2019). The mean age of persons who received ICI alone (78 years old) was higher than PLT or PLT+ICI. Prior to 1L treatment, 22% and 72% of persons received ABX within 30 days and 1 year prior to cancer treatment, respectively. The three most common ABX classes were fluoroquinolones (48%), penicillins (21%), and macrolides (18%). The cohort was followed for 3,274 person-years and included 2,958 deaths within 1 year and 4,144 deaths within 3 years of follow-up. Compared with persons who did not receive ABX within 3 months of cancer treatment, ABX use was associated with higher risk of death within 3 years for ICI alone [HR 1.24 (1.05-1.46) but not for PLT alone or PLT+ICI. Fluoroquinolones were associated with an elevated risk of mortality among persons who received ICI alone HR 1.41(95%CI: 1.1-1.8) and PLT alone HR 1.18 (95%CI: 1.05-1.32) but not PLT+ICI. Our findings were similar across definitions of ABX exposure (30 days to 1 year prior to treatment) and analytic landmarks after treatment initiation. Conclusions: Among patients with NSCLC who received 1L treatment, we identified an association between ABX exposure prior to treatment and increased mortality among persons who received ICI monotherapy but not PLT+ICI. As ICIs expand to more cancer types and earlier stages, ABX use may have a significant impact on outcomes and population health.

Gestational diabetes mellitus is associated with greater incidence of dementia during long-term post-partum follow-up.

The impact of gestational diabetes mellitus (GDM) on incident dementia is unknown. Our aim was to evaluate the relationship between GDM and all-cause dementia and the mediating effects of chronic diseases on this relationship. This prospective cohort study included women from the UK Biobank who were grouped based on GDM history. Multivariate Cox proportional hazard models were used to explore the associations between GDM and dementia. We further analysed the mediating effects of chronic diseases on this relationship and the interactions of covariates. A total of 1292 women with and 204,171 women without a history of GDM were included. During a median follow-up period of 45 years after first birth, 2921 women were diagnosed with dementia. Women with a GDM history had a 67% increased risk of incident dementia (hazard ratio 1.67, 95% confidence interval: 1.03-2.69) compared with those without a GDM history. According to mediation analyses, type 2 diabetes, coronary heart disease, chronic kidney disease and comorbidities (diagnosed with any two of the three diseases) explained 34.5%, 8.4%, 5.2% and 18.8% of the mediating effect on the relationship. Subgroup analyses revealed that physical activity modified the association between GDM history and dementia (p for interaction=0.030). Among physically inactive women, GDM was significantly associated with incident dementia; however, this association was not observed among physically active women. A history of GDM was associated with a greater risk of incident dementia. Type 2 diabetes partially mediated this relationship. Strategies for dementia prevention might be considered for women with a history of GDM.

Estimating risk for pancreatic cancer among 9.4 million veterans in care.

10544 Background: Screening for pancreatic ductal adenocarcinoma (PDAC) has focused on individuals with a hereditary component, accounting for only 10% of PDAC events. Using the electronic health record, we developed a PDAC risk prediction model in a general population in the Veterans Health Administration (VA), the largest integrated health system in the US. Methods: We included all individuals from 2002- 2021 with ≥1 inpatient or outpatient VA encounter and recorded race, starting follow-up 1.5 years after first VA encounter. We excluded those who developed PDAC before start of follow-up. We used the multivariable Cox proportional hazards model in the derivation dataset (random 70% of cohort) to estimate the coefficients associated with each predictor for our outcome, time to PDAC (3 and 10 years), censoring at loss to follow-up or 12/31/2021. Candidate predictors included age, gender, race/ethnicity, sex, BMI, smoking status, alcohol use (alcohol use/related disorders and AUDIT-C), diabetes, pancreatitis, Hepatitis C, HIV, liver disease, cholecystectomy, pancreatic cysts, Charlson comorbidity index and common laboratory values. Selection of predictors was based on clinical reasoning, extent of missing values and model fit (Akaike Information Criterion and Harrell’s c–statistic). We evaluated model discrimination using c – statistic and model calibration comparing observed Kaplan – Meier estimated incidence vs. predicted risk score. Results: Among 9.4 million individuals, we observed 19,839 cases of PDAC within 10 years (6,022 cases within 3 years). Median baseline age was 58.5 years, 75.0% were White, and 16.3% were Black. Final predictors included age, gender, smoking, alcohol use (alcohol use/related disorders and AUDIT-C), BMI, diabetes, pancreatitis, pancreatic cyst, and history of cancer (Table, c-statistic in the derivation cohort 3 years: 0.79; 10 years: 0.77). Ten-year risk ranged maximally from 0-5.5%. Preliminary studies suggest labs do not clinically improve model performance. Conclusions: We developed a novel prediction model using available clinical data in a general population that identifies PDAC with high accuracy. Future work will try to identify additional predictors to improve the range of risk predicted. We will validate our findings in the testing set and in external cohorts. [Table: see text]

Impact of young age at diagnosis on survival outcomes in patients with early breast cancer according to tumor subtype: Pooled analysis of 5 randomized clinical trials from the Mammella InterGruppo (MIG) and Gruppo Italiano Mammella (GIM).

523 Background: The incidence of breast cancer (BC) among young patients is increasing. Historically, young age itself has been recognised as a poor prognostic factor for BC. However, there is limited evidence regarding the impact of age at diagnosis and on survival outcomesaccording to tumor subtype. Methods: We collected individual patient (pt)-level data from newly diagnosed stage I to III BC pts participating in 5 randomized clinical trials conducted by Mammella InterGruppo (MIG) and Gruppo Italiano Mammella (GIM) study groups. Patients were categorized into the following age groups based on their age at BC diagnosis: ≤40, 41-50, 51-60, 61-70, &gt;70 years (yr). Multivariable Cox proportional hazards models were used to assess the relationship between age and breast cancer–free interval (BCFI) and breast cancer specific survival (BCSS). Results: 8,338 pts with stage I to III BC were included in the study: 778 (9%) aged ≤40 years, 1629 (20%) 41-50 years, 2,595 (31%) 51-60 years, 2,435 (29%) 61-70 years, and 901 (11%) &gt;70 years. Compared to older, young pts were more likely to have ductal tumors as histology, larger than 2 cm, with nodal involvement, G3 and triple negative or HER2 positive subtypes (p&lt;0.001). With a median follow-up of 9.0 years (IQR 5.5-14.4), 2,161 BCFI and 927 BCSS events were observed in the overall population. Pts ≤40 years at diagnosis had greater risk of reporting BC event (adjusted hazard ratio (adjHR) for BCFI 1.27, 95% CI 0.93-1.74) compared to pts aged 51-60 years (reference group), while pts 41-50 years had a lower risk (adjHR 0.74, 95%CI 0.55-0.99). Pts aged ≤40 years presented some evidence of a lower risk of BCSS events (adjHR 0.80, 95%CI 0.50-1.26) compared to pts aged 51-60 years, while pts aged 41-50 years had the lowest risk (adjHR 0.50, 95%CI 0.32-0.78). There was evidence of interaction between age and tumor subtype, nodal status both for BCFI and BCSS. A significant increase in risk of BC events was observed among pts ≤40 years with hormone-receptor positive/HER2 negative tumors (HR 1.70 for BCFI) and among ≤40 years with node negative tumors (HR for BCFI 2.43). Conclusions: The prognostic significance of young age at diagnosis may vary depending on tumor subtypes and nodal status, indicating poorer outcomes for young pts with hormone-receptor positive/HER2 negative and node negative tumors.

The prognostic impact of transforming growth factor beta 2 (TGFB2) mRNA levels, in conjunction with interferon-gamma receptor activation of interferon regulatory factor 5 (IRF5) and expression of CD276/B7-H3 in low-grade gliomas.

e14564 Background: Low-grade glioma (LGG) tumors are characterized by a low infiltration of immune cells, requiring therapeutic interventions to boost the immune response. Methods: We analyzed mRNA expression datasets from the UCSC Xena web platform to compare normal brain tissue with LGG tumor samples to screen for upregulated genes. We also used cBioportal to determine the relationship between mRNA expression levels of 513 LGG patients and their overall survival (OS) outcomes. Results: Tumor-associated macrophage (TAM) markers, MSR1/CD204, CD86, and CD68, exhibited a 6-fold (P&lt;0.0001), 8.9-fold (P&lt;0.0001), and 15.6-fold increases in mRNA expression levels, respectively in LGG tumors. Both TGFB1 (4.1-fold increase, P&lt;0.0001) and TGFB2 (2.2-fold increase, P&lt;0.0001) were upregulated in brain tumors, suggesting that TGFB ligands are pivotal in establishing an immunosuppressive TME. In addition, mRNA upregulation of interferon-gamma receptors, IFNGR1 and IFNGR2, and signaling molecules STAT1, IRF1, and IRF5, pointed to an essential role for IFN-γ mediated remodeling of the TME. A tumor-associated antigen, CD276/B7-H3, mRNA expression showed a significant (P&lt;0.0001) 4.03-fold increase in tumor tissue. Multivariate Cox proportional hazards models investigating the interaction of TGFB2 and activation of IFNGR2, STAT1, IRF1, or IRF5 showed that the prognostic impact of high mRNA levels (25 percentile cut-off) of TGFB2high was independent of IFNGR2, STAT1, IRF1, or IRF5 mRNA levels (TGFB2high HR (95% CI) = 4.07 (2.35-7.06), 6 (3.62-10.11), 4.38 (2.67-7.17), and 4.48 (2.82-7.12) for models with IFNGR2, STAT1, IRF1, or IRF5 respectively). Patients with high levels of TGFB2 were overrepresented by LGG patients with IDH wild-type mutation status. The prognostic impact of TGFB2high and IDH wild-type observed by the increases in hazard ratios for TGFB2 (HR (95% CI range) = 2.02 (1.05-3.89)) and IDH wild-type (HR (95% CI range) = 4.44 (1.9-10.4)) were independent predictors of survival requiring risk stratification of patients identify LGG patients with IDH wild-type and TGFB2high in the design of clinical trials. The ratio of TGFB1:TGFB2 expression levels was also a prognostic indicator for OS, whereby patients who exhibited a 2.5-fold greater level of TGFB1 compared to TGFB2 mRNA levels experienced improved survival outcomes. A 1.8-fold higher TGFB2 levels, the mRNA expression corresponded to worse OS. Conclusions: The specific abrogation of TGFB2 and not TGFB1 by OT101, a TGFB2-specific synthetic phosphorothioate antisense oligodeoxynucleotide, is a prerequisite to employing a strategy to target TGFB pathways. IRF5 and CD276/B7-H3 as prognostic markers are potential targets for combination therapies with TGFB2 inhibitors.