Proportional Subdistribution Hazards Model Research Articles

Impact of bariatric surgery on the development of diabetic microvascular and macrovascular complications.

While bariatric surgery has been shown to improve type 2 diabetes (DM) control in the obese population, the effect on long-term DM complications has been less thoroughly investigated. The purpose of this study was to assess the development of microvascular and macrovascular complications in obese DM patients undergoing bariatric surgery. New York patients' records from the SPARCS database in years 2006-2012 were used to identify obese patients with DM. Patients undergoing bariatric surgery were compared with patients managed medically, matched for age and gender. Patients were grouped based on baseline presence of controlled or uncontrolled DM and followed over time for the development of micro- and macrovascular complications. Cumulative incidence of complications was estimated with death treated as a competing risk event. Multivariable proportional sub-distribution hazards models were used to compare the risk of complications among different patient groups after adjusting for possible confounding factors. A total of 88,981 patients were reviewed, including 15,585 (18%) that were treated with bariatric surgery. Surgery patients had significantly lower risk of microvascular complications compared to non-surgery patients (controlled diabetes: HR = 0.40, 95% CI 0.37-0.42; uncontrolled diabetes: HR = 0.51, 95% CI 0.37-0.71). Similarly, the surgical patients were noted to have a significantly lower risk for macrovascular complications compared to non-surgery patients (controlled diabetes: HR = 0.43, 95% CI 0.40-0.46; uncontrolled diabetes: HR = 0.44, 95% CI 0.28-0.69). Cumulative incidence of microvascular complications was lower at 1, 5 and 9years for the surgical groups for controlled and uncontrolled DM. Similar trends were observed for the macrovascular complications. Bariatric surgery appears to prevent complications of DM. Bariatric surgery patients with DM experienced significantly lower rates of microvascular and macrovascular complications, compared to non-surgically treated comparison group. Bariatric surgery was noted to offer protective benefits for both complicated and non-complicated DM patients. This reduced rate of complications was sustained in the long term.

The associations of interstitial lung abnormalities with cancer diagnoses and mortality.

An increased incidence of lung cancer is well known among patients with idiopathic pulmonary fibrosis. It is not known whether interstitial lung abnormalities, i.e. early fibrotic changes of the lung, are a risk factor for lung cancer in the general population.The study's objective was to assess whether interstitial lung abnormalities were associated with diagnoses of, and mortality from, lung cancer and other cancers. Data from the AGES-Reykjavik study, a cohort of 5764 older Icelandic adults, were used. Outcome data were ascertained from electronic medical records. Gray's tests, Cox proportional hazards models and proportional subdistribution hazards models were used to analyse associations of interstitial lung abnormalities with lung cancer diagnoses and lung cancer mortality as well as diagnoses and mortality from all cancers.There was a greater cumulative incidence of lung cancer diagnoses (p<0.001) and lung cancer mortality (p<0.001) in participants with interstitial lung abnormalities than in others. Interstitial lung abnormalities were associated with an increased hazard of lung cancer diagnosis (hazard ratio 2.77) and lung cancer mortality (hazard ratio 2.89) in adjusted Cox models. Associations of interstitial lung abnormalities with all cancers were found in models including lung cancers but not in models excluding lung cancers.People with interstitial lung abnormalities are at increased risk of lung cancer and lung cancer mortality, but not of other cancers. This implies that an association between fibrotic and neoplastic diseases of the lung exists from the early stages of lung fibrosis and suggests that interstitial lung abnormalities could be considered as a risk factor in lung cancer screening efforts.

Stratified proportional subdistribution hazards model with covariate‐adjusted censoring weight for case‐cohort studies

AbstractThe case‐cohort study design is widely used to reduce cost when collecting expensive covariates in large cohort studies with survival or competing risks outcomes. A case‐cohort study dataset consists of two parts: (a) a random sample and (b) all cases or failures from a specific cause of interest. Clinicians often assess covariate effects on competing risks outcomes. The proportional subdistribution hazards model directly evaluates the effect of a covariate on the cumulative incidence function under the non‐covariate‐dependent censoring assumption for the full cohort study. However, the non‐covariate‐dependent censoring assumption is often violated in many biomedical studies. In this article, we propose a proportional subdistribution hazards model for case‐cohort studies with stratified data with covariate‐adjusted censoring weight. We further propose an efficient estimator when extra information from the other causes is available under case‐cohort studies. The proposed estimators are shown to be consistent and asymptotically normal. Simulation studies show (a) the proposed estimator is unbiased when the censoring distribution depends on covariates and (b) the proposed efficient estimator gains estimation efficiency when using extra information from the other causes. We analyze a bone marrow transplant dataset and a coronary heart disease dataset using the proposed method.

Spatial access to primary care providers and colorectal cancer-specific survival in Cook County, Illinois.

BackgroundSpatial access to primary care has been associated with late‐stage and fatal breast cancer, but less is known about its relation to outcomes of other screening‐preventable cancers such as colorectal cancer. This population‐based retrospective cohort study examined whether spatial access to primary care providers associates with colorectal cancer‐specific survival.MethodsApproximately 26 600 incident colorectal cancers diagnosed between 2000 and 2008 in adults residing in Cook County, Illinois were identified through the state cancer registry and georeferenced to the census tract of residence at diagnosis. An enhanced two‐step floating catchment area method measured tract‐level access to primary care physicians (PCPs) in the year of diagnosis using practice locations obtained from the American Medical Association. Vital status and underlying cause of death were determined using the National Death Index. Fine‐Gray proportional subdistribution hazard models analyzed the association between tract‐level PCP access scores and colorectal cancer‐specific survival after accounting for tract‐level socioeconomic status, case demographics, tumor characteristics, and other factors.ResultsIncreased tract‐level access to PCPs was associated with a lower risk of death from colorectal cancer (hazard ratio [HR], 95% confidence interval [CI]) = 0.87 [0.79, 0.96], P = .008, highest vs lowest quintile), especially among persons diagnosed with regional‐stage tumors (HR, 95% CI = 0.80 [0.69, 0.93], P = .004, highest vs lowest quintile).ConclusionsSpatial access to primary care providers is a predictor of colorectal cancer‐specific survival in Cook County, Illinois. Future research is needed to determine which areas within the cancer care continuum are most affected by spatial accessibility to primary care such as referral for screening, accessibility of screening and diagnostic testing, referral for treatment, and access to appropriate survivorship‐related care.

On the relation between the cause-specific hazard and the subdistribution rate for competing risks data: The Fine-Gray model revisited.

The Fine–Gray proportional subdistribution hazards model has been puzzling many people since its introduction. The main reason for the uneasy feeling is that the approach considers individuals still at risk for an event of cause 1 after they fell victim to the competing risk of cause 2. The subdistribution hazard and the extended risk sets, where subjects who failed of the competing risk remain in the risk set, are generally perceived as unnatural . One could say it is somewhat of a riddle why the Fine–Gray approach yields valid inference. To take away these uneasy feelings, we explore the link between the Fine–Gray and cause‐specific approaches in more detail. We introduce the reduction factor as representing the proportion of subjects in the Fine–Gray risk set that has not yet experienced a competing event. In the presence of covariates, the dependence of the reduction factor on a covariate gives information on how the effect of the covariate on the cause‐specific hazard and the subdistribution hazard relate. We discuss estimation and modeling of the reduction factor, and show how they can be used in various ways to estimate cumulative incidences, given the covariates. Methods are illustrated on data of the European Society for Blood and Marrow Transplantation.

Nomogram predicting cancer-specific mortality in early-onset rectal cancer: a competing risk analysis.

The incidence of rectal cancer has meaningfully increased in young patients. However, quantitative evaluation for the competing data of early-onset rectal cancer is lacking. So, we performed a competing risk analysis to calculate the cumulative incidence of death for patients with early-onset rectal cancer and developed a nomogram to predict the probability of cancer-specific mortality for these patients. We abstracted data of patients with early-onset rectal cancer between 2004 and 2016 by using the Surveillance, Epidemiology, and End Results program database. The cumulative incidence function was used to calculate the crude cancer-specific mortality of early-onset rectal cancer. Fine and Gray's proportional sub-distribution hazard model was adopted to explore the risk factors of cancer-specific death. Then, we establish a nomogram to predict their 3-, 5-, and 10-year probabilities. We identified 9917 patients with early-onset rectal cancer, and they were randomly divided into training (n = 6941) and validation (n = 2976) cohorts. In the training cohort, the 3-, 5-, and 10-year cumulative incidences of cancer-specific death after diagnosis for early-onset rectal cancer were 11.4%, 19.9%, and 28.8%, respectively. Fine and Gray's model showed that sex, race, marital status, histology, T stage, N stage, M stage, examined lymph nodes, and pretreatment carcinoembryonic antigen were independently associated with cancer-specific mortality. Such factors were selected to develop a prognostic nomogram. The competing risk nomogram has an ideal performance for predictive cancer-specific mortality in early-onset rectal cancer.

Regression Analysis of Masked Competing Risks Data under Cumulative Incidence Function Framework

In the studies that involve competing risks, somehow, masking issues might arise. That is, the cause of failure for some subjects is only known as a subset of possible causes. In this study, a Bayesian analysis is developed to assess the effect of risks factor on the Cumulative Incidence Function (CIF) by adopting the proportional subdistribution hazard model. Simulation is conducted to evaluate the performance of the proposed model and it shows that the model is feasible for the possible applications.

A competing-risk nomogram to predict cause-specific death in elderly patients with colorectal cancer after surgery (especially for colon cancer)

BackgroundClinically, when the diagnosis of colorectal cancer is clear, patients are more concerned about their own prognosis survival. Special population with high risk of accidental death, such as elderly patients, is more likely to die due to causes other than tumors. The main purpose of this study is to construct a prediction model of cause-specific death (CSD) in elderly patients using competing-risk approach, so as to help clinicians to predict the probability of CSD in elderly patients with colorectal cancer.MethodsThe data were extracted from Surveillance, Epidemiology, and End Results (SEER) database to include ≥ 65-year-old patients with colorectal cancer who had undergone surgical treatment from 2010 to 2016. Using competing-risk methodology, the cumulative incidence function (CIF) of CSD was calculated to select the predictors among 13 variables, and the selected variables were subsequently refined and used for the construction of the proportional subdistribution hazard model. The model was presented in the form of nomogram, and the performance of nomogram was bootstrap validated internally and externally using the concordance index (C-index).ResultsDataset of 19,789 patients who met the inclusion criteria were eventually selected for analysis. The five-year cumulative incidence of CSD was 31.405% (95% confidence interval [CI] 31.402–31.408%). The identified clinically relevant variables in nomogram included marital status, pathological grade, AJCC TNM stage, CEA, perineural invasion, and chemotherapy. The nomogram was shown to have good discrimination after internal validation with a C-index of 0.801 (95% CI 0.795–0.807) as well as external validation with a C-index of 0.759 (95% CI 0.716–0.802). Both the internal and external validation calibration curve indicated good concordance between the predicted and actual outcomes.ConclusionUsing the large sample database and competing-risk analysis, a postoperative prediction model for elderly patients with colorectal cancer was established with satisfactory accuracy. The individualized estimates of CSD outcome for the elderly patients were realized.

Predictors of Relapse after Haploidentical Hematopoietic Progenitor Cell Transplantation (Haplo-HCT); A Single-Institution Experience

<h3>Background</h3> Haplo-HCT emerged as a reliable option for patients with high risk hematological malignancies with no HLA-matched related or unrelated donors available. Retrospective registry data suggest comparable outcomes to HLA-matched donors; however, relatively little is known regarding predictors of disease relapse following haplo-HCT. We hypothesize that attainment of full-donor CD3 and CD33 chimerism at day 30 (D30) and 100 (D100) is associated with lower incidence of relapse after haplo-HCT. <h3>Methods</h3> We undertook a retrospective analysis of 78 patients who underwent haplo-HCT at the Ohio State University James Cancer Hospital between January 2013 and December 2017. The associations between patient characteristics including CD3/CD33 donor chimerisms and the cumulative incidence rate (CIR) of relapse were evaluated using proportional sub-distribution hazards model, treating death without relapse as the competing risk. The models were built adopting a landmark analysis approach at D30 and D100. Progression-free survival (PFS) was estimated using the method of Kaplan-Meier. <h3>Results</h3> Median age at haplo-HCT was 56 (20-74) years and 68% were males. 50% had acute myeloid leukemia/myelodysplastic syndrome, and 54% had intermediate disease risk index (DRI). Most patients (85%) received reduced-intensity conditioning and 50% received bone marrow (BM) grafts. Complete (100%) CD3 donor chimerism was observed in 63 (84%) patients at D30, and subsequently 59 (95%) patients at D100. Complete CD33 donor chimerism was observed in 67 (89%) patients at D30, and subsequently 56 (88%) patients at D100. In univariable D30 landmark analysis, the CIR of relapse was significantly lower in patients who achieved complete CD3 (and CD33) chimerism compared with patients with less than complete chimerism (Figure 1A; p<0.01 and p<0.01, respectively). Attainment of complete CD3 chimerism at D30 was also associated with PFS (Figure 1B; p<0.01). However, in multivariable analysis, complete CD3 chimerism at D30 was no longer independently associated with the CIR of relapse due to its high correlation with DRI (p=0.01). Receipt of a BM graft was associated with significantly higher CIR of relapse both in univariable and multivariable analyses (HR 2.43, 3.34; p=0.05, 0.01, respectively). D100 complete CD33 chimerism conferred a significant CIR advantage in both univariable (p<0.01) and multivariable (p<0.01) models, although CD3 chimerism did not retain significance (p=0.35). <h3>Conclusions</h3> Complete CD3 and CD33 donor chimerisms at D30 and D100 predict a patient's risk for relapse following haplo-HCT. We observed a strong association between CD3 chimeric status at D30 with DRI, as well suggestive that high risk disease may be associated with impaired immune reconstitution after transplant and perhaps a less robust graft vs disease effect.

Neighborhood context and non-small cell lung cancer outcomes in Florida non-elderly patients by race/ethnicity

ObjectiveThe purpose of this study was to investigate the relationship between neighborhood environment and lung cancer outcomes among Florida residents younger than 65 years of age. Methods and materialsThis was a retrospective cohort study that included patients diagnosed with non-small cell lung cancer (NSCLC) in Florida from January 2005 to December 2014 (n = 22,750). Multi-level, mixed-effect logistic regression models were used for two outcomes: receipt of treatment and receipt of surgery. Survival analyses, using proportional subdistribution hazard models, were conducted to examine the impact of neighborhood characteristics on risk of death due to lung cancer with adjustment for individual-level variables. Neighborhood exposures of interest were census tract level black and Hispanic segregation combined with economic deprivation. ResultsWhite patients who lived in low black segregation/high deprivation areas had 15 % lower odds of receiving surgery (95 % CI: 0.76-0.93). However, the likelihood of receiving surgery for black patients who lived in high black segregation/low deprivation and high black segregation/high deprivation was lower than for black patients who lived in low black segregation/low deprivation neighborhoods (level 3 AOR = 0.56 [0.38-0.85]; level 4 AOR = 0.69 [0.54-0.88]). Living in suburban and rural areas increased the risk of lung cancer death for white patients by 14 % (95 % CI: 1.05–1.24) and 26 % (95 % CI: 1.08–1.46), respectively. Living in rural areas increased the risk of death for black patients by 54 % r (SHR = 1.54 [1.19–2.0]). Black patients who live in high Hispanic segregation/high deprivation had 36 % increased risk of death compared to black patients who lived in low Hispanic segregation/low deprivation areas. ConclusionThis study suggests that when investigating cancer disparities, merely adjusting for race/ethnicity does not provide sufficient explanation to understand survival and treatment variations. Lung cancer outcomes are impacted by neighborhood environments that are formed based on the distribution of race, ethnicity and class.

Resistance to Acalabrutinib in CLL Is Mediated Primarily By BTK Mutations

Resistance to Acalabrutinib in CLL Is Mediated Primarily By BTK Mutations

Population-based study evaluating and predicting the probability of death resulting from thyroid cancer among patients with papillary thyroid microcarcinoma.

BackgroundPapillary thyroid microcarcinoma (PTMC) is an indolent carcinoma. The cumulative incidence of death from patients with PTMC and the nomogram built based on the competing risks model have not been described.MethodsPatients diagnosed with PTMC were selected from the Surveillance, Epidemiology, and End Results (SEER) program (1983‐2015). Cumulative incidence function was utilized to calculate the likelihood of death resulting from thyroid cancer. Gray's test was conducted to examine the difference in the cumulative incidence of death between groups. A proportional subdistribution hazard model was constructed and we further built a nomogram to quantify the likelihood of death using the model. A 10‐fold cross‐validation procedure was adopted to validate the model.ResultsA total of 46 662 patients diagnosed with PTMC were included. The median follow‐up time was 81 months (range, 1 to 407 months). The 5‐year, 10‐year, and 20‐year probabilities of death from thyroid carcinoma were 0.3%, 0.6%, and 1.4%, respectively. The age at diagnosis, sex, tumor extension, and lymph node involvement were related to the cumulative incidence of death. A proportional subdistribution hazard model was developed. The performance of the model was good. A nomogram was built based on the model to predict the likelihood of death in patients with PTMC.ConclusionThe survival rate of patients with PTMC is excellent. The nomogram constructed based on the well‐performed competing risks model is helpful for both patients and clinicians.

The health and cost burden of antibiotic resistant and susceptible Escherichia coli bacteraemia in the English hospital setting: A national retrospective cohort study.

IntroductionAntibiotic resistance poses a threat to public health and healthcare systems. Escherichia coli causes more bacteraemia episodes in England than any other bacterial species. This study aimed to estimate the burden of E. coli bacteraemia and associated antibiotic resistance in the secondary care setting.Materials and methodsThis was a retrospective cohort study, with E. coli bacteraemia as the main exposure of interest. Adult hospital in-patients, admitted to acute NHS hospitals between July 2011 and June 2012 were included. English national surveillance and administrative datasets were utilised. Cox proportional hazard, subdistribution hazard and multistate models were constructed to estimate rate of discharge, rate of in-hospital death and excess length of stay, with a unit bed day cost applied to the latter to estimate cost burden from the healthcare system perspective.Results14,042 E. coli bacteraemia and 8,919,284 non-infected inpatient observations were included. E. coli bacteraemia was associated with an increased rate of in-hospital death across all models, with an adjusted subdistribution hazard ratio of 5.88 (95% CI: 5.62–6.15). Resistance was not found to be associated with in-hospital mortality once adjusting for patient and hospital covariates. However, resistance was found to be associated with an increased excess length of stay. This was especially true for third generation cephalosporin (1.58 days excess length of stay, 95% CI: 0.84–2.31) and piperacillin/tazobactam resistance (1.23 days (95% CI: 0.50–1.95)). The annual cost of E. coli bacteraemia was estimated to be £14,346,400 (2012 £), with third-generation cephalosporin resistance associated with excess costs per infection of £420 (95% CI: 220–630).ConclusionsE. coli bacteraemia places a statistically significant burden on patient health and the hospital sector in England. Resistance to front-line antibiotics increases length of stay; increasing the cost burden of such infections in the secondary care setting.

Let us not underestimate the long-term risk of SPLC after surgical resection of NSCLC

ObjectivesSeveral studies have reported that patients operated on for non-small cell lung cancer (NSCLC) are at high risk of second primary lung cancer (SPLC). However, widely varying estimates of this risk have been reported, with very few studies taking into account that these patients are at particularly high competing risk of death, due to recurrence of the initial disease and to comorbidities. Risk factor evaluation over time has significant repercussions on the post-surgery surveillance strategy offered for NSCLC. This study primarily sought to measure the risk of SPLC in a long-term follow-up series, using statistical methods considering competing risks of death. Materials and methodsThe cumulative SPLC risk was estimated using the cumulative incidence of patients with completely resected Stage I-III NSCLC diagnosed between 2002 and 2015 based on the Doubs and Belfort cancer registry (France). A proportional sub-distribution hazard model (sdRH) was used to investigate factors associated with SPLC risk in the presence of competing risks. ResultsAmong the 522 patients, adenocarcinoma and Stage I or II disease accounted for 52.3% and 75.7% of patients, respectively. Overall, 84 patients developed SPLC (16.1%). The cumulative risk of SPLC was 20.2% at 10 years post-surgery (95% confidence interval [CI]: 15.3–23.2), and 25.2% (CI: 19.4–31.3) at 14 years post-surgery. On multivariate analysis, the SPLC risk was significantly higher in patients with postoperative thoracic radiotherapy (sdRH 2.79; 95% CI: 1.41–5.52; p = 0.003). ConclusionThis study using appropriate statistical methods to consider competing risks showed that after complete NSCLC resection, the cumulative incidence function of SPLC was high, with patients receiving postoperative thoracic radiotherapy at higher risk. These data support the need for life-long follow-up of patients who undergo NSCLC surgery, with the objective of screening for SPLC.

Second primary malignancy in patients with esophageal adenocarcinoma and squamous cell carcinoma

There have been no studies on implementing effective screening models for esophageal adenocarcinoma and squamous cell carcinoma survivors. We used a proportional subdistribution hazards model to estimate second primary malignancy risks among patients with esophageal adenocarcinoma and squamous cell carcinoma. We validated models using a bootstrap cross-validation method and performed decision curve analysis to evaluate their clinical utility. Age group and SEER historic stage were significantly associated with second primary malignancy risk after diagnosis of esophageal adenocarcinoma and squamous cell carcinoma. Saving positive lymph nodes and distant metastasis were significant factors in the adenocarcinoma group, and marital status, tumor location, and chemotherapy were significant factors in the squamous cell carcinoma group. Calibration plots show good concordance between predicted and actual outcomes except in high-probability areas for the risk of a second primary malignancy in patients with esophageal squamous cell carcinoma. Discrimination performances of the Fine-Gray models were evaluated using c-indices, which were 0.691 and 0.662 for second primary malignancies in patients with esophageal adenocarcinoma and squamous cell carcinoma, respectively. Decision curve analysis yielded a range of threshold probabilities (0.020-0.177 and 0.021-0.133 for patients with esophageal adenocarcinoma and squamous cell carcinoma, respectively) at which the clinical net benefit of the risk model was larger than those of hypothetical all-screening and no-screening scenarios. Our nomograms enable selection of patient populations at high risk for a second primary malignancy and thus will facilitate the design of prevention trials for affected populations.

Time of Day and its Association with Risk of Death and Chance of Discharge in Critically Ill Patients: A Retrospective Study

Outcomes following admission to intensive care units (ICU) may vary with time and day. This study investigated associations between time of day and risk of ICU mortality and chance of ICU discharge in acute ICU admissions. Adult patients (age ≥ 18 years) who were admitted to ICUs participating in the Austrian intensive care database due to medical or surgical urgencies and emergencies between January 2012 and December 2016 were included in this retrospective study. Readmissions were excluded. Statistical analysis was conducted using the Fine-and-Gray proportional subdistribution hazards model concerning ICU mortality and ICU discharge within 30 days adjusted for SAPS 3 score. 110,628 admissions were analysed. ICU admission during late night and early morning was associated with increased hazards for ICU mortality; HR: 1.17; 95% CI: 1.08–1.28 for 00:00–03:59, HR: 1.16; 95% CI: 1.05–1.29 for 04:00–07:59. Risk of death in the ICU decreased over the day; lowest HR: 0.475, 95% CI: 0.432–0.522 for 00:00–03:59. Hazards for discharge from the ICU dropped sharply after 16:00; lowest HR: 0.024; 95% CI: 0.019–0.029 for 00:00–03:59. We conclude that there are “time effects” in ICUs. These findings may spark further quality improvement efforts.

Late mortality after bone marrow transplant for chronic myelogenous leukemia in the context of prior tyrosine kinase inhibitor exposure: A Blood or Marrow Transplant Survivor Study (BMTSS) report.

Late mortality was investigated in patients with chronic myelogenous leukemia (CML) who underwent blood or bone marrow transplant (BMT) with or without prior tyrosine kinase inhibitor (TKI) therapy. By using data from the Blood or Marrow Transplant Survivor Study, the authors examined late mortality in 447 patients with CML who underwent BMT between 1974 and 2010, conditional on surviving ≥2years post-BMT. For vital status information, the medical records, the National Death Index, and the Accurint database were used. Standardized mortality ratios (SMRs) were calculated using general population age-specific, sex-specific, and calendar-specific mortality rates. Kaplan-Meier techniques and Cox regression were used for all-cause mortality analyses. Cumulative incidence and proportional subdistribution hazards models for competing risks were used for cause-specific mortality analyses. The 10-year overall survival rate was 65.7% and 73% for those who underwent transplant with and without pre-BMT exposure to TKI therapy, respectively. Patients who underwent transplant with and without pre-BMT TKI experienced SMRs of 6.4 and 6.4, respectively (P=.8); and the SMRs were 11.6 and 8.1, respectively, for those with high-risk disease (P=.2). Independent predictors of non-CML-related mortality included chronic graft-versus-host disease (hazard ratio [HR], 2.8; 95% CI, 1.8-4.4) and busulfan/cyclophosphamide conditioning (HR, 0.5; 95% CI, 0.3-0.9; reference, total body irradiation/cyclophosphamide conditioning). The 20-year cumulative incidence of CML-related and non-CML-related mortality was 6% and 36%, respectively, for the entire cohort. Both CML-related mortality (HR, 1.0; 95% CI, 0.1-12.6) and non-CML-related mortality (HR, 1.3; 95% CI, 0.6-3.1) were comparable for those with and without pre-BMT TKI therapy. The similar late mortality experienced by patients with CML who undergo transplantation with or without pre-BMT TKIs suggests that allogeneic BMT can be considered in the context of TKI intolerance or nonadherence. The prevention of post-BMT non-CML-related mortality could favorably affect long-term survival.

Prognostic factors in patients with gastric adenocarcinoma using competing-risk analysis: a study of cases in the SEER database

Background: Accurate prognostic factors for gastric adenocarcinoma are still lacking in clinical practice, which contributes to inappropriate treatment. Applying the widely used Cox-proportional hazards model to describe survival trends and identify prognostic factors has limitations that result in a risk of bias. A competing-risk model was therefore adopted in this study to identify the significant prognostic factors and evaluate the cumulative incidence of cause-specific death for gastric adenocarcinoma, which can be used to guide clinical treatments.Methods: All of the cases analyzed in this study were extracted from the SEER (Surveillance, Epidemiology, and End Results) database. Using the competing risk approach, we calculated the cumulative incidence function (CIF) for cause-specific death and death from other causes at each time point. The Fine and Gray’s proportional subdistribution hazard model was then applied in the univariate analysis and multivariate analysis to test the differences in CIF between different groups and obtain independent prognostic factors.Results: The univariate analysis showed that patients with characteristics of advanced pathology grade, lymph node involvement, and metastasis, were at risk of increasing cancer-specific mortality. Primary-site surgery, radiation with surgery, and chemotherapy, were associated with decreased cancer-specific mortality. The multivariate analysis showed that pathology grade, primary-site surgery, radiation with surgery, and chemotherapy, could significantly affect the cancer-specific mortality and were independent prognostic factors in patients with gastric adenocarcinoma.Conclusions: Using a competing-risk model, this study obtained more-accurate estimates for the cumulative incidence of cancer-specific death and identified the prognostic factors more accurately for gastric adenocarcinoma.

Competing risks and cause-specific mortality in patients with pancreatic neuroendocrine tumors.

Currently, there are no competing risk analyses of cause-specific mortality in patients with pancreatic neuroendocrine tumors. We estimated a cumulative incidence function for cause-specific mortality. The first nomogram for predicting cause-specific mortality was constructed using a proportional subdistribution hazard model, validated using bootstrap cross-validation, and evaluated with decision curve analysis. Sex, age, positive lymph node status, metastasis, surveillance, epidemiology, and end results historic stage, grade, and surgery strongly predicted cause-specific mortality. The discrimination performance of Fine-Gray models was evaluated using the c-index, which was 0.864. In addition, the calibration plot of the developed nomogram demonstrated good concordance between the predicted and actual outcomes. Decision curve analysis yielded a range of threshold probabilities (0.014-0.779) at which the clinical net benefit of the risk model was greater than that in hypothetical all-screening or no-screening scenarios. Our nomogram allows selection of a patient population at high risk for cancer-specific mortality and thus facilitates the design of prevention trials for the affected population.

Cause-specific death assessment of patients with stage I small-cell lung cancer: a competing risk analysis.

Aim: Stage I small-cell lung cancer (SCLC) is a potentially curable disease that needs timely and multidisciplinary management. The aim of this study was to evaluate the probability of cause-specific mortality for patients with stage I SCLC. Material & methods: We identified patients in the SEER database and constructed a proportional subdistribution hazard model to evaluate cancer-specific mortality. A nomogram was built based on Fine and Gray competing risk regression model. Results:A total of864 stage I SCLC patients were identified. The 5-year cumulative incidence of SCLC-specific mortality was 56.2%, while that for other causes of death was 17.3%. The c-index for the prognostic prediction model was 0.66. Besides, the nomogram was well calibrated. Conclusion: Our nomogram might serve as a reference for clinicians when evaluating the prognosis of stage I SCLC.