Multivariable Competing Risk Models Research Articles

Sex Differences in the Development of Anthracycline-Associated Heart Failure

Female sex is frequently cited as a risk factor for anthracycline cardiotoxicity based on pediatric data, but the role of sex in the development of cardiotoxicity has not been clearly established in adults. To assess the effect of female sex on the development of incident heart failure (HF) in adult patients treated with anthracyclines. This was a retrospective cohort study of 1525 adult patients with no prior history of HF or cardiomyopathy who were treated with anthracyclines between 1992 and 2019. The primary outcome was new HF within 5 years of the first dose of anthracyclines. The effect of sex was assessed using Cox proportional hazards and competing risk models. Over a median (IQR) follow-up of 1.02 (0.30-3.01) years, 4.78% of patients developed HF (44 men and 29 women). Female sex was not associated with the primary outcome in a multivariable Cox proportional hazards model (HR 0.87; 95% CI 0.53-1.43; P = 0.58). Similar results were observed in a multivariable model accounting for the competing risk of death (HR 0.94; 95% CI 0.39-2.25; P = 0.88). Age, coronary artery disease and hematopoietic stem cell transplant were associated with the primary outcome in a multivariable Cox proportional hazards model. Age and body mass index were associated with the primary outcome in a multivariable competing risk model. In this large, single-center, retrospective cohort study, female sex was not associated with incident HF in adult patients treated with anthracyclines. Female sex is frequently cited as a risk factor for anthracycline cardiotoxicity based on pediatric data, but the role of sex in the development of cardiotoxicity has not been clearly established in adults. In this retrospective cohort study, we assessed the effect of female sex on the development of incident heart failure in adult patients treated with anthracyclines. Using Cox proportional hazards and competing risk regression models, we found that there was no association between female sex and heart failure after treatment with anthracyclines.

The association between histological subtypes and lymph node metastasis and prognosis in early esophageal cancer: a population-based study.

There is still a lack of high-level clinical evidence and uniform conclusions on whether there are differences in lymph node metastasis (LNM) and prognosis between early esophageal adenocarcinoma (EAC) and squamous cell carcinoma (ESCC). Patients with surgically resected, histologically diagnosed, pT1 EAC or ESCC in the Surveillance, Epidemiology and End Results registries database from 2004 to 2015 were included. Multivariable logistic regression, Cox regression, multivariate competing risk model, and propensity score matching were used to analyze association the histology and LNM or prognosis. A total of 570 early esophageal cancer patients were included. The LNM rates were 13.8% and 15.1% for EAC and ESCC ( P = 0.757), respectively. Multivariate logistic regression analysis showed no significant association between histological type and LNM (odds ratio [OR], 1.209; 95% CI, 0.538-2.715; P = 0.646). Moreover, the prognosis of early EAC and ESCC was shown to be comparable in both multivariate Cox regression (hazard ratio [HR], 1.483; 95% CI, 0.699-3.150; P = 0.305) and the multivariate competing risk model (subdistribution HR, 1.451; 95% CI, 0.628-3.354; P = 0.383). After propensity score matching, there were no significant differences between early EAC and ESCC in terms of LNM (10.6% vs.18.2%, P = 0.215), 5-year CSS (89.8% [95% CI, 81.0%-98.6%] vs. 79.1% [95% CI, 67.9%-90.3%], P = 0.102) and 5-year cumulative incidence of CSS (10.2% [95% CI, 1.4%-19.0%] vs. 79.1% [95% CI, 9.7%-32.1%], P = 0.124). The risk of LNM and prognosis of early ESCC and EAC are comparable, so the treatment choice for early esophageal cancer does not depend on the histologic type.

Metastasis, characteristic, and treatment of breast cancer in young women and older women: A study from the Surveillance, Epidemiology, and End Results registration database.

Younger age is an independent risk factor for breast cancer (BC) prognosis, and BC in young women is often considered more aggressive. BC patients with different age and molecular subtypes have different metastasis patterns and survival. Herein, we aim to explore the metastasis patterns, characteristics and treatment methods of young patients with BC, and to compare them with older patients. Data of young patients (aged ≤40 years old) and older patients (aged >40 years old) with BC were extracted from the Surveillance, Epidemiology, and End Results (SEER) registration database in 2010-2019 in this retrospective cohort study. Univariate and multivariate competing risk models and proportional hazard models were used to explore the association between different metastasis patterns and treatments and BC prognoses in young and older patients. Kaplan-Meier (KM) curves were drawn to reflect the survival probability of patients with BC who have different metastasis patterns. Also, we performed subgroup analysis of different metastasis patterns to explore the association between different treatments and overall survival (OS)/cancer specific survival (CSS) in patients with BC. The evaluation index was hazard ratios (HRs) with 95% confidence intervals (CIs). Totally, 5,984 patients died, and 92.56% of them died from BC. There were respectively 1,089 young patients and 9,105 older patients, and we found some differences of characteristics and metastasis patterns between them. After adjusting for covariates, young patients who had brain metastasis and multiple sites metastasis seemed to have high risk of both lower OS and CSS. Among older patients with BC, brain metastasis, liver metastasis, and multiple sites metastasis were all positively associated with both lower OS and CSS. In young and older patients, those who not receive radiotherapy or surgery, or received non-surgery combined with radiotherapy seemed to have high risk of both lower OS and CSS. Breast-conserving surgery (BCS) and surgery combined with radiotherapy were associated with higher OS and CSS in young patients, while only older patients received surgery combined with radiotherapy had higher OS and CSS. Results of subgroup analysis indicated that for patients with different metastasis patterns, developing a personalized treatment plan is necessary. Characteristics of BC between young patients and older patients were different. Clinicians should focus on different metastasis sites and choose appropriate treatments in patients with different ages, which may improve the prognoses.

Competing risk analysis of cardiovascular-specific mortality in typical carcinoid neoplasms of the lung: A SEER database analysis.

Cardiovascular mortality (CVM) is a growing concern for cancer survivors. This study aimed to investigate the mortality patterns of individuals with typical carcinoid (TC) tumors, identify independent predictors of CVM, and compare these risk variables with those associated with TC deaths. The Surveillance, Epidemiology, and End Results (SEER) database from 2000 to 2019 was utilized for obtaining data on patients with TC. Standardized mortality rates were employed to evaluate the risk of CVM while multivariate competing risk models were used to determine the association between patient characteristics and the probability of CVM or TC-related deaths. Our findings show that TC patients had an increased risk of CVM, with an standardized mortality rates of 1.12 (95% CI:1.01-1.25). Furthermore, we discovered that age at diagnosis, marital status, year of diagnosis, SEER stage, site, year of diagnosis, surgery, radiotherapy, and chemotherapy all contributed independently to the risk of CVM in patients with TC, whereas age at diagnosis, sex, race, SEER stage, site, year of diagnosis, surgery, radiotherapy, and chemotherapy all contributed significantly to TC mortality. Compared to the general population in the United States, patients with TC are significantly more likely to acquire CVM. Timely introduction of cardioprotective treatments is critical for preventing CVM in patients with TC.

Statin use, survival and incidence of thrombosis among older patients with polycythemia vera and essential thrombocythemia.

Polycythemia vera (PV) and essential thrombocythemia (ET) are linked to increased risk of cardiovascular morbidity and mortality. In addition to the reduction in of arterial thrombotic events, statins may prevent venous thrombosis including among patients with cancer. As previous registry- and claims-based studies revealed that the use of statins may improve the survival of patients with various malignancies we evaluated their impact on outcomes of older adults with PV and ET. We identified 4010 older adults (aged 66-99 years at diagnosis) with PV (n = 1809) and ET (n = 2201) in a population-based cohort study using the Surveillance, Epidemiology, and End Results-Medicare database with median follow-up of 3.92 (interquartile range: 2.58-5.75) years. Propensity score matching (PSM) and inverse probability of treatment weighting (IPTW) approaches were utilized to assess potential association between statins and overall survival. Multivariable competing risk models with death as a competing risk were used to evaluate possible relationship between statins and the incidence of thrombosis. 55.8% of the patients used statins within the first year after PV/ET diagnosis, and statin use was associated with a 22% reduction in all-cause mortality (PSM: hazard ratio [HR] = 0.78, 95% confidence interval [CI]: 0.63-0.98, p = 0.03; IPTW: HR = 0.79, 95% CI: 0.64-0.97, p = 0.03). Statins also reduced the risk of thrombosis in this patient population (PSM: HR = 0.63, 95% CI: 0.51-0.78, p < 0.01; IPTW: HR = 0.57, 95% CI: 0.49-0.66, p < 0.01) as well as in PV and ET subgroups. These findings suggest that it may be important to incorporate statins into the therapeutic strategy for older adults with PV and ET.

Rate of Kidney Function Decline is Associated With Kidney and Heart Failure in Individuals With Type 1 Diabetes

Diabetes is the most common cause of chronic kidney disease (CKD). Urinary albumin excretion rate (AER) and estimated glomerular filtration rate (eGFR) are commonly used to monitor the onset and progression of diabetic kidney disease (DKD). We studied if the preceding rate of kidney function decline, that is, the eGFR slope, is independently associated with incident clinical cardiorenal events. This study included longitudinal data for 2498 Finnish individuals with type 1 diabetes (T1D). The eGFR slope was calculated from 5 years preceding the study visit. Data on kidney failure, coronary heart disease (CHD), stroke, 3-point major adverse cardiovascular events (MACE), heart failure, and death were obtained from national registries. The associations between the eGFR slope and incident events were assessed with multivariable competing risk models during the average follow-up of 9.2 years. The eGFR slopes were associated (P≤ 0.001) with all outcomes when adjusted for age, sex, and HbA1c. However, eGFR slope remained associated only with the composite outcome of kidney failure or death when the albuminuria group and eGFR at the study visit were included in the model (P= 0.041). In addition, eGFR slope was independently associated with kidney failure in individuals without CKD (eGFR > 60 ml/min per 1.73 m2; P= 0.044), and with heart failure in those with CKD (P= 0.033). However, eGFR slope did not markedly improve the model C-index. The eGFR slope was independently associated with kidney failure in those without CKD, and with heart failure in those with CKD. However, it is unlikely to have major relevance for clinical practice when the current eGFR and albuminuria status are known.

Identification of risk factors and development of a predictive model for bloodstream infection in intensive care unit COVID-19 patients

To identify risk factors for nosocomial bloodstream infections (BSIs) in intensive care unit (ICU) patients with COVID-19 and to build a predictive model for BSIs. The retrospective case-control study included 236 ICU COVID-19 patients with BSIs group and 234 patients in the control group. Demographic and laboratory data, comorbidities, drug use, invasive procedures and identified pathogens were recorded separately for patients directly admitted and transferred to ICU. Fine and Gray's multi-variate competing risk model was used to build a predictive model for patients transferred to ICU. The risk factors were: interleukin inhibitors (HR= 6.1 (95% CI: 2.0-18.5)) and dexamethasone (HR= 3.0 (95% CI: 1.3-7.1)) use in previous hospitalization, glomerular filtration rate <60mL/min per 1.73m2 (HR= 4.0 (95% CI: 2.1-7.6)) and blood glucose >9mmol/L (HR= 2.5 (95% CI: 1.4-4.6)) in patients directly admitted to ICU; and dexamethasone use in previous hospitalization (HR= 4.5 (95% CI: 1.8-11)), the total dexamethasone dose before transfer to ICU (HR= 1.2 (95% CI: 1.06-1.37)), diabetes mellitus (HR= 1.4 (95% CI: 1.1-1.9)), alanine transaminase (ALT) ≥35.5 U/L on hospital admission (HR= 1.5 (95% CI: 1.1-2.1)), and the use of low-flow oxygen versus high-flow oxygen therapy or non-invasive mechanical ventilation on admission to ICU ((HR= 2.7 (95% CI: 5.6-11.1)) in patients transferred to ICU. A predictive model had sensitivity of 63-73% and specificity of 71-83% at different times of ICU stay. Our findings may help clinicians detect patients at high risk of developing BSIs.

High risk of non-cancer mortality in bladder cancer patients: evidence from SEER-Medicaid

PurposeThe objective of this study was to investigate non-cancer causes of death and associated risk factors after bladder cancer (BC) diagnosis.MethodsEligible BC patients were obtained from the SEER database. SEER*Stat software 8.3.9.2 was used to calculate the standardized mortality ratios (SMRs). The proportions of different non-cancer cause of death were calculated and analyzed in different follow-up periods. Multivariate competing risk model was used to analyze the risk factors for death of BC and non-cancer diseases.ResultsIn total, 240,954 BC patients were included and 106,092 patients experienced death, with 37,205 (35.07%), 13,208 (12.45%) and 55,679 (52.48%) patients experienced BC, other cancer and non-cancer disease-related deaths, respectively. Overall SMR for BC patients who died from non-cancer diseases was 2.42 (95% CI [2.40–2.44]). Cardiovascular diseases were the most common non-cancer cause of death, followed by respiratory diseases, diabetes mellitus, and infectious diseases. Multivariate competing risk analysis identified the following high-risk factors for non-cancer mortality: age > 60 years, male, whites, in situ stage, pathological type of transitional cell carcinoma, not receiving treatment (including surgery, chemotherapy, or radiation), and widowed.ConclusionsCardiovascular diseases are the leading non-cancer cause of death in BC patients, followed by respiratory disease, diabetes mellitus and infectious diseases. Physicians should pay attention to the risk of death from these non-cancer diseases. Also, physicians should encourage patients to engage in more proactive self-surveillance and follow up.

Investigating the association (assoc) between mental health illness (MHI) and development of prostate cancer (PC) in a nationwide matched cohort of &gt;5,000,000 US Veterans.

e17113 Background: Previous reports suggest that men with MHI who subsequently develop PC have worse outcomes. However, it has not been established whether a history of MHI is associated with a PC diagnosis (dx) or more aggressive PC. The objective of this study was to (i) investigate the assoc between MHI and development of PC, and (ii) in a subset of men with available Gleason Score (GS) data, assess the assoc between MHI and aggressive PC as measured by GS ≥7 PC at dx. Methods: This was a retrospective matched-cohort analysis to assess the assoc between MHI and time to PC dx in the Veterans Affairs (VA) Health Care System. Men with an ICD code for MHI dx between 2000-2020 were matched 1:1 to men with no MHI dx in the same time frame. The MHI dx date of the exposed (exp) male was assigned as the index date of the unexp male. Variables matched on included race, age at MHI exp (+/- 3 years), census region, and median household income. Uni- and multivariable competing risks models were used to test the assoc between MHI and time to PC using death from other causes as competing risk. In a subset of men with GS data available, logistic regression was used to test the assoc between MHI and GS ≥7 at PC dx. All multivariable models were adjusted for age, race, census region, income, Charlson Comorbidity Index, year of MHI index, and year of VA entry. Results: There were 2,597,810 MHI-exp men matched 1:1 to unexp men. During a median (Q1, Q3) follow up of 128 (65, 192) months, 390,977 PC diagnoses were observed (172,442 MHI-exp vs. 218,535 unexp). MHI-exp men were significantly less likely to be diagnosed with PC than unexp men in both uni- (HR: 0.798, 95% CI: 0.793-0.803) and multivariable (HR: 0.763, 95% CI: 0.758-0.768) analysis. Cumulative incidence estimates at 3, 5, 10, and 20 years were 2.1%, 3.1%, 5.2%, and 8.5%, respectively, for MHI-exp men vs 2.9%, 4.2%, 6.9%, and 11.7%, respectively, for unexp men. Among men with PC and available GS data (n = 51,404), 32,645 had GS ≥7 PC, of which 20,749 were MHI-exp vs 11,896 unexp. MHI-exp men had significantly higher odds of GS ≥7 PC in both uni- (OR: 2.38, 95% CI: 2.32-2.43) and multivariable (OR: 1.97, 95% CI: 1.92-2.02) analysis. Conclusions: Men with MHI are 20% less likely to be diagnosed with PC, however when diagnosed, are nearly 2 times more likely to have aggressive PC compared to non-MHI men. Although speculative based on the nature of the study, this may be due to poorer access to preventative health services resulting in delayed diagnoses.

Competing risk nomogram for predicting cancer-specific mortality in patients with non-melanoma skin cancer.

This study aimed to assess the cumulative incidences of Non-melanoma skin cancer (NMSC)-specific mortality (NMSC-SM) and develop a competing risk nomogram for NMSC-SM. Data on patients diagnosed with NMSC between 2010 and 2015 were extracted from the Surveillance, Epidemiology, and End Results (SEER) database. To identify the independent prognostic factors, univariate and multivariate competing risk models were used, and a competing risk model was constructed. Based on the model, we developed a competing risk nomogram to predict the 1-, 3-, 5-, and 8-year cumulative probabilities of NMSC-SM. The precision and ability to discriminate of the nomogram were evaluated through the utilization of metrics, such as receiver-operating characteristic (ROC) area under the curve (AUC), concordance index (C-index), and a calibration curve. Decision curve analysis (DCA) was employed to assess the clinical usefulness of the nomogram. Race, age, the primary site of the tumor, tumor grade, size, histological type, summary stage, stage group, order of radiation and surgery, and bone metastases were identified as independent risk factors. The prediction nomogram was constructed using the variables mentioned above. The ROC curves implied the good discrimination ability of the predictive model. The nomogram's C-index was 0.840 and 0.843 in the training and validation sets, respectively, and the calibration plots were well fitted. In addition, the competing risk nomogram demonstrated good clinical usefulness. The competing risk nomogram displayed excellent discrimination and calibration for predicting NMSC-SM, which can be used in clinical contexts to help guide treatment decisions.

The prognostic value of postoperative radiotherapy in right tumor for lung related death: based on SEER database and real-world data.

Postoperative radiotherapy (PORT) is a therapeutic strategy for patients with non-small cell lung cancer (NSCLC). Nevertheless, some studies suggesting PORT does not improve overall survival (OS) including Lung ART phase III trial. The role of PORT and high-risk groups need to be confirmed. Patients from the Surveillance, Epidemiology, and End Results program (SEER) from 2004 to 2015 were eligible. Aged ≥18 years with stage IIIA-N2 NSCLC, accepted PORT or not were considered for the study. Cox regression analyses and multivariate competing risk model were performed. Propensity score matching (PSM) was conducted. Data from a single-center study in China were used for validation. In all patients with IIIA-N2 NSCLC, death from respiratory illness increased year by year, with right lung-related deaths accounting for the main proportion. In SEER database, PORT was detrimental for OS after PSM (hazard ratio [HR], 1.088; 95% CI, 1.088-1.174; P = 0.031), with a same trend for death from the lungs (HR, 1.13; 95% CI, 1.04-1.22; P = 0.005). Right tumor receiving PORT were prone to death from lung disease(HR, 1.14; 95% CI, 1.02-1.27; P = 0.018). In China single-center cohort, PORT was significantly correlated with deteriorated OS (HR 1.356; 95% CI 1.127-1.632; P <0.01), especially in the right laterality (HR 1.365; 95% CI 1.062-1.755; P = 0.015). PORT was a risk factor for stage IIIA-N2 NSCLC patients, particularly with characters of right laterality, male sex, age ≥65 years, and advanced tumor stage. These patients are more likely to death from lung disease after PORT.

Abstract 734: Comprehensive validation of high-risk clinicopathologic features in early-stage, node-negative non-small cell lung cancer

Abstract Objective: Current guidelines recommend adjuvant therapy for patients with completely resected non-small cell lung cancer (NSCLC) with high-risk clinical or pathologic features. Despite this, the relationship between these features and cancer recurrence is poorly elucidated. Methods: We conducted a retrospective cohort study using a uniquely compiled dataset from the US Veterans Health Administration (VHA) including all Veterans with pathologic early-stage (≤5cm, N0) NSCLC receiving definitive surgical treatment (2010-2016). Based on National Comprehensive Cancer Network guidelines, we evaluated 6 high-risk features: tumor size, tumor grade, visceral-pleural invasion, lymphovascular invasion, non-anatomic wedge resection, and adequacy of nodal sampling. We developed a score reflecting the relationship between these high-risk features and recurrence, using a multivariable competing risk model (death as competing event). The score performance was then tested in an external cohort from the National Cancer Database (NCDB). Results: The study included 3,799 Veterans. The median follow-up was 7.1 years. Recurrence was detected in 800 (21.1%) patients. The association between high-risk features and cancer recurrence were as follows: tumor size (e.g., 31-40mm vs. 0-10mm, multivariable-adjusted hazard ratio, aHR 1.676, 95% CI 1.229-2.285, p=0.001), tumor grade (e.g., III vs. I, aHR 1.884, 95% CI 1.448-2.449, p&amp;lt;0.001), visceral-pleural invasion (aHR 1.096, 95% CI 0.905-1.329, p=0.35), lymphovascular invasion (aHR 1.747, 95% CI 1.441-2.117, p&amp;lt;0.001), non-anatomic wedge resection (aHR 1.335, 95% CI 1.101-1.619, p=0.003), and adequacy of nodal sampling (e.g., 1-4 lymph nodes vs. ≥10 lymph nodes, aHR 1.392, 95% CI 1.149-1.687, p&amp;lt;0.001). Using these parameters, a score was created reflecting the association between high-risk features and recurrence. The score ranged from 0-36, with higher scores reflecting higher cumulative incidence of recurrence. The score was further divided into low- (0-11, n=1,263, 33.3%; 5-yr recurrence risk 13.0%), moderate- (12-15, n=1,134, 29.9%; 5-yr recurrence risk 19.0%), and high-risk (16-36, n=1,402, 36.9%; 5-yr recurrence risk 27.1%) categories. Higher scores were also associated with diminished overall survival (median OS, low-risk: 9.0 yrs; moderate-risk: 7.3 yrs; high-risk: 5.4 yrs). The score was further tested in a cohort of 63,232 patients from the NCDB and higher scores remained associated with worse overall survival (median OS, low-risk: 9.4 yrs; moderate-risk: 8.0 yrs; high-risk: 6.3 yrs). Conclusions: High-risk clinicopathologic features are associated with dramatically higher risk of recurrence and worse overall survival. Multivariable assessment of these features using a comprehensive yet pragmatic score may help to standardize adjuvant treatment eligibility following curative intent resection. Citation Format: Brendan T. Heiden, Daniel B. Eaton, Su-Hsin Chang, Yan Yan, Martin W. Schoen, Bindiya G. Patel, Theodore S. Thomas, Bryan F. Meyers, Benjamin D. Kozower, Varun Puri. Comprehensive validation of high-risk clinicopathologic features in early-stage, node-negative non-small cell lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 734.

Competing risk nomogram for predicting prognosis of patients with spinal and pelvic chordoma: A SEER-based retrospective study

PurposeRecently, competing risk nomograms were widely applied to predict prognosis in numerous tumors other than chordoma. Here, we aimed to construct and validate a competing-risk-based prognostic nomogram to predict 3- and 5-year cancer-specific death (CSD) in patients with spinal and pelvic chordoma.MethodsAll chordoma patient data were abstracted from the Surveillance, Epidemiology, and End Results (SEER) resource, and a total of 485 chordoma patients were eventually included in this study. Multivariate competing risk model and multivariate Cox model were used to determine independent prognostic factors, respectively, and the results of the two models were compared. Nomogram was employed to visualize the competing risk model. The discrimination, calibration, and clinical utility of this model were evaluated by Harrell concordance index (C-index), time-dependent receiver operating characteristic (ROC) curves, calibration plots, and decision curve analysis (DCA). Ten-fold cross-validation was further utilized to validate the prognostic nomogram.ResultsSignificant prognostic factors affecting CSD were age (P = 0.016), localized involvement (P < 0.0001), and radical resection (P < 0.001) in the multivariate competing risk model. C-indexes were 0.799 and 0.76, and AUC were 0.812 and 0.778 for 3- and 5-year CSD. Calibration plots demonstrated the nomogram was well-fitted, and DCA indicated good clinical utility. The nomogram showed good performance in the 10-fold cross-validation.ConclusionWe successfully built the first competing-risk-based nomogram to predict clinical outcomes in patients with spinal and pelvic chordoma. This well-established nomogram hopes to help clinicians with precise prognostic assessment and thus improve clinical outcomes.

Cardiovascular-Specific Mortality among Gastrointestinal Stromal Tumor Patients: A Population-Based Analysis.

The overall risk of cardiovascular mortality (CVM) in cancer survivors has increased with time. The trend of CVM in patients with gastrointestinal stromal tumors (GISTs) remains unclear. This study is aimed at assessing the risks and independent predictors of CVM in GIST patients. Data of the GIST patients were extracted from the Surveillance, Epidemiology, and End Results (SEER) database (2000-2019). The standardized mortality ratio (SMR) was used to evaluate the risk of CVM, and a multivariate competing risk model was utilized to identify the predictors for CVM. A total of 12,058 patients with GIST were included in this study, of whom 477 (4.0%) patients died of cardiovascular disease (CVD). The SMR for CVM among GIST patients was significantly higher than in the general population (SMR, 3.23, 95% CI: 2.97-3.52), and all categories of CVD were associated with a significantly elevated SMR. The cumulative mortality of CVD was the lowest among all causes of death, while the CVM was the second most common cause of death in patients ≥ 80 years when stratified by age at diagnosis. Furthermore, male sex, older age at diagnosis, White race, unmarried, earlier year of diagnosis, and not receiving chemotherapy were the poor prognostic factors for CVM. The CVM risk in GIST patients was significantly higher relative to the general US population. Timely screening and cardioprotective interventions should be implemented to prevent the occurrence of CVM in patients with GIST.

Higher risk of cardiovascular mortality than cancer mortality among long-term cancer survivors.

Previous studies focused more on the short-term risk of cardiovascular (CV) death due to traumatic psychological stress after a cancer diagnosis and the acute cardiotoxicity of anticancer treatments than on the long-term risk of CV death. Time trends in the proportions of CV death (PCV), cancer death (PCA), and other causes in deaths from all causes were used to show preliminary relationships among the three causes of death in 4,806,064 patients with cancer from the Surveillance, Epidemiology, and End Results (SEER) program. Competing mortality risk curves were used to investigate when the cumulative CV mortality rate (CMRCV) began to outweigh the cumulative cancer mortality rate (CMRCA) for patients with cancer who survived for more than 10 years. Multivariable competing risk models were further used to investigate the potential factors associated with CV death. For patients with cancer at all sites, the PCV increased from 22.8% in the 5th year after diagnosis to 31.0% in the 10th year and 35.7% in the 20th year, while the PCA decreased from 57.7% in the 5th year after diagnosis to 41.2 and 29.9% in the 10th year and 20th year, respectively. The PCV outweighed the PCA (34.6% vs. 34.1%) since the 15th year for patients with cancer at all sites, as early as the 9th year for patients with colorectal cancer (37.5% vs. 33.2%) and as late as the 22nd year for patients with breast cancer (33.5% vs. 30.6%). The CMRCV outweighed the CMRCA since the 25th year from diagnosis. Multivariate competing risk models showed that an increased risk of CV death was independently associated with older age at diagnosis [hazard ratio and 95% confidence intervals [HR (95%CI)] of 43.39 (21.33, 88.28) for ≥ 80 vs. ≤ 30 years] and local metastasis [1.07 (1.04, 1.10)] and a decreased risk among women [0.82 (0.76, 0.88)], surgery [0.90 (0.87, 0.94)], and chemotherapy [0.85 (0.81, 0.90)] among patients with cancer who survived for more than 10 years. Further analyses of patients with cancer who survived for more than 20 years and sensitivity analyses by cancer at all sites showed similar results. CV death gradually outweighs cancer death as survival time increases for most patients with cancer. Both the cardio-oncologist and cardio-oncology care should be involved to reduce CV deaths in long-term cancer survivors.

Comparison of LNM and survival in T1 stage esophageal cancer patients based on histological classification: A large population-based study

Limited evidence and contradictory results have been reported regarding the impact of squamous cell carcinoma (SCC) and adenocarcinoma (AC) classification on lymph node metastasis (LNM) and prognosis in esophageal cancer (EC). We aimed to compare 2 histology types in terms of LNM and prognosis using a comprehensive statistical analysis of a large population. The Surveillance, Epidemiology, and End Results (SEER) database was used to extract patient information. Univariate and multivariate logistic or Cox regression, a multivariate competing risk model and propensity score matching (PSM) were used to explore the association between LNM or survival and the 2 histology types. Information for 4764 patients, including 1712 SCC and 3052 AC patients, was extracted from the Surveillance, Epidemiology, and End Results (SEER) database. Multivariate logistic regression analysis revealed a correlation between LNM and histology (odds ratio [OR] = 0.654, P = .037). We found that patients with AC had a better prognosis than SCC patients through both the multivariate Cox regression (hazard ratio [HR] = 0.866) and the multivariate competing risk model (subdistribution hazard ratio [SHR] = 0.704). However, no positive relation was found between LNM and histology type (P = .844) based on propensity score matching (PSM), and the prognosis remained poorer for the patients with SCC (P < .001). T1-stage EC with a histology of SCC may have a comparable risk of LNM as the AC type, while SCC has a poorer prognosis than the AC type.

Establishment and verification of a nomogram to predict tumor-specific mortality risk in triple-negative breast cancer: a competing risk model based on the SEER cohort study.

Triple-negative breast cancer (TNBC) is the subtype of breast cancer with the worst prognosis, and traditional survival analysis methods are biased when predicting mortality. To predict the risk of death in patients with triple-negative breast cancer more precisely, a competing risk model was developed. The clinicopathological data of the TNBC patients from 2010 to 2015 were collected from the Surveillance, Epidemiology, and End Results (SEER) database. The data were assigned into a training set and testing set at a ratio of 7:3 in a randomized pattern. Univariate and multivariate competing risk models were applied to find the independent prognostic factors. A prediction nomogram for cancer-specific mortality (CSM) risk was constructed. The accuracy and discrimination of the nomogram were assessed using receiver operating characteristic (ROC) area under the curve (AUC), concordance index (C-index), and a calibration curve using the training and testing sets, respectively. A total of 28,430 TNBC patients were randomly grouped into the training set (n=19,900) and the testing set (n=8,530). The median time for follow-up was 59 [1-107] months. A total of 7,014 (24.67%) patients died, among whom 4,801 (68.45%) died from breast cancer and 2,213 (31.55%) due to non-breast cancer events. The independent risk factors were primary site of tumor, grade, tumor-node-metastasis (TNM) stage, T stage, approach of surgery, chemotherapy, axillary lymph node metastases, brain metastases, and liver metastases. The prediction nomogram was constructed by using the aforementioned variables. The 1-, 3-, and 5-year AUC of CSM were predicted to be 0.856, 0.81, and 0.782, respectively, in the training set, and 0.856, 0.81, and 0.782 in the testing set, respectively. The C-index of the nomogram was 0.801 and 0.799 in the training and testing sets, respectively. As confirmed by the validation and training calibration curves, the nomogram was consistent with the results. We used competing risk models to identify risk factors for CSM and constructed a CSM risk prediction nomogram for TNBC patients, that may be utilized to predict CSM risk in TNBC patients clinically and assist in the creation of individualised clinical treatment options.

Establishment and validation of a nomogram for predicting immune-related prognostic features in trunk melanoma-specific death.

Trunk melanoma is one of the most common and deadly types of melanomas. Multiple factors are associated with the prognosis of patients with trunk melanoma. Currently, direct, and reliable clinical tools for early assessment of individual specific risk of death are limited, and most of them are prediction models for all-cause death. Their accuracy in predicting competitiveness events, which make up a relatively large portion, may be substantially compromised. Hence, we conducted this study to investigate the risk factors of trunk melanoma-specific death to establish a comprehensive prediction model suitable for clinical application. Patients with trunk melanoma analyzed in this study were from the SEER program [2010-2015]. The random sampling method was used to split the included cases into the training and validation cohorts at a ratio of 7:3. Univariate and multivariate competing risk models were used to screen the independent influencing factors of specific death, and then a nomogram covering these independent predictors was constructed. The concordance index (C-index) and a calibration curve were used to evaluate the calibration degree and accuracy of the nomogram. We identified 21,198 patients with trunk melanoma from the SEER database, and 3,814 of them died (17.99%). Among the death cases, deaths from other causes accounted for 66.50%The prognostic nomogram included 8 variables and 16 independent influencing factors. The overall C-index in the training set was 0.89, and the receiver operating characteristic (ROC) curve for predicting 1-, 3-, and 5-year survival was 0.928 [95% confidence interval (CI): 0.911-0.945], 0.907 (95% CI: 0.895-0.918), and 0.891 (95% CI: 0.879-0.902), respectively. The C-index of the model in the validation set was 0.89, and the area under the ROC curve (AUC) for predicting 1-, 3-, and 5-year cancer-specific death (CSD) was 0.927 (95% CI: 0.899-0.955), 0.916 (95% CI: 0.901-0.930), and 0.905 (95% CI: 0.899-0.921). Both the training set and the validation set showed the ideal calibration degree. This model can be used as a potential tool for prognostic risk management of trunk melanoma in the presence of many competing events.

Impact of adequate lymph nodes dissection on survival in patients with stage I rectal cancer.

The NCCN guidelines recommended an assessment of ≥ 12 lymph nodes (LN) as an adequate LN dissection (LND) for rectal cancer (RC). However, the impact of adequate LND on survival in stage I RC patients remained unclear. Thus, we aimed to compare the survival between stage I RC patients with adequate and inadequate LND. A total of 1,778 stage I RC patients in the SEER database from 2010 to 2017 treated with radical proctectomy were identified. The association between ≥ 12 LND and survival was examined using the multivariate Cox regression and the multivariate competing risk model referenced to < 12 LND. Stage I RC patients with ≥ 12 LND experienced a significantly lower hazard of cancer-specific death compared with those with < 12 LND in both multivariate Cox regression model (adjusted HR [hazard ratio], 0.44, 95% CI, 0.29-0.66; P < 0.001) and the multivariate competing risk model (adjusted subdistribution HR [SHR], 0.45, 95% CI, 0.30-0.69; P < 0.001). Further, subgroup analyses performed by pT stage. No positive association between ≥ 12 LND and survival was found in pT1N0 RC patients (adjusted HR: 0.62, 95%CI, 0.32-1.19; P = 0.149; adjusted SHR: 0.63, 95%CI, 0.33-1.20; P = 0.158), whereas a positive association between ≥ 12 LND and survival was found in pT2N0 RC patients (adjusted HR: 0.35, 95%CI, 0.21-0.58; P < 0.001; adjusted SHR: 0.36, 95%CI, 0.21-0.62; P < 0.001). The long-term survival benefit of adequate LND was not found in pT1N0 but in pT2N0 RC patients, which suggested that pT2N0 RC patients should be treated with adequate LND and those with inadequate LND might need additional therapy.

Second Malignancies Among Older Patients with Classical Myeloproliferative Neoplasms Treated with Ruxolitinib

Introduction: Myeloproliferative neoplasms (MPN) are associated with an increased risk of clonal evolution to more aggressive myeloid disorders and the development of second malignancies (SM) including lymphoproliferative and solid tumors. The reasons for SM are unclear, but a therapy-related effect has been invoked. Ruxolitinib (RUX) is a guideline-recommended cytoreductive agent used in the US for myelofibrosis (MF) and polycythemia vera (PV) since 2011 and 2014, respectively. RUX has been linked to an increased risk of non-melanoma skin cancers (NMSC), but its influence on the development of other SM is debated. We aimed to evaluate the effect of RUX on the development of solid and lymphoid SM in a large, population-based cohort of patients (pts) with MPN. Methods: We conducted a retrospective cohort study of older adults with MPN including PV, essential thrombocythemia (ET) and MF in the Surveillance, Epidemiology, and End Results (SEER)-Medicare linked database. Pts were required to 1) be diagnosed in 2012-2017, 2) be 66-99 years of age at diagnosis, 3) have continuous enrollment in Medicare Parts A and B from one year before MPN diagnosis through end of follow-up, and 4) have continuous Part D coverage from diagnosis to the end of follow-up. We excluded pts who were health maintenance organizations members, or treated with chlorambucil, busulfan or allogeneic stem cell transplant. We followed pts from MPN diagnosis to 12/31/2019, SM diagnosis, death or change of insurance status, whichever came first. RUX proportion of days covered (PDC) was calculated as the ratio of the number of days the patient was covered by RUX to the number of days from RUX initiation to 6 months before end of follow-up. Our outcome of interest was first SM, i.e., any new malignancy other than new myeloid malignancy and NMSC, diagnosed at least 6 months after cohort entry. We identified SM using SEER data for the years 2012-2017 and Medicare claims for the years 2018-2019. We categorized SM as solid or lymphoid. The cumulative incidence function (CIF) of SM was computed via a competing risk model. Death and development of a SM other than malignancy of interest (solid or lymphoid) were considered as competing events. Gray's test was utilized to assess difference across strata. The impact of RUX on the development of SM was assessed by multivariable competing risks models. The models adjusted for age at MPN diagnosis, sex, race, marital status, SEER region, Elixhauser comorbidity score, history of prior malignancy, disability status, state buy-in, census tract Yost index and hydroxyurea exposure. All statistical tests were two-sided and performed with SAS Version 9.4. Results: The study included 4019 pts (1556 PV, 1896 ET, 567 MF) with a median age at diagnosis of 77 (interquartile range [IQR]: 71-83) years. The median follow-up was 3.5 (IQR: 2.23- 4.98) years and 3.18 (IQR: 2.17- 4.75) years for 408 RUX users and 3611 non-users, respectively. Among RUX users, RUX was initiated after a median of 0.66 years since diagnosis. Pts received RUX for a median of 1.15 (IQR: 0.42-2.12) years with a median PDC of 94.6% (IQR: 71.3- 99.7%). Compared with those who did not receive RUX, RUX users were more likely to be younger, have MF and a history of prior malignancy. A total of 396 pts developed a SM including 306 solid SMs and 90 lymphoid SM. About 9% RUX users and 10% non-users developed a SM, resulting in a CIF of 17.8% (95% confidence interval [CI]: 10.1-27.4%) and 16.3% (95% CI: 13.9-18.9%), respectively (Gray's test, p=0.17; (Figure 1). The CIF of solid SM among RUX users was 11.8% (95% CI: 5.4-21.1%), and 13.0% among non-users (95% CI: 10.7-15.5%, p= 0.06); in lymphoid SM, CIF was 6.0 % (95% CI: 2.64-11.3%) among RUX users, and 3.3% (95% CI: 2.6- 4.2%) among non-users (p=0.52). In the multivariable model, the risk of any SM (hazard ratio [HR]=1.07, 95% CI: 0.73-1.58, p=0.72), solid (HR=0.97, 95% CI: 0.62-1.51, p=0.89) and lymphoid SM (HR=1.45, 95% CI: 0.66-3.20, p=0.35) did not differ by RUX use status. No difference in the risk of any SM was observed when evaluating the influence of RUX PDC (Figure 2) Similar findings were seen in MPN subgroups. Conclusions: In this large population-based study of older pts with MPN, the risk of solid and lymphoid SM was similar among RUX users and non-users. Our results suggest that the use of RUX does not increase the risk of SM in older pts with MPN. Acknowledgments: The study was supported by Frederick A. Deluca Foundation. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal