Competing Risk Outcomes Research Articles

Scalable log-ratio lasso regression for enhanced microbial feature selection with FLORAL.

Identifying predictive biomarkers of patient outcomes from high-throughput microbiome data is of high interest, while existing computational methods do not satisfactorily account for complex survival endpoints, longitudinal samples, and taxa-specific sequencing biases. We present FLORAL, an open-source tool to perform scalable log-ratio lasso regression and microbial feature selection for continuous, binary, time-to-event, and competing risk outcomes, with compatibility for longitudinal microbiome data as time-dependent covariates. The proposed method adapts the augmented Lagrangian algorithm for a zero-sum constraint optimization problem while enabling a two-stage screening process for enhanced false-positive control. In extensive simulation and real-data analyses, FLORAL achieved consistently better false-positive control compared to other lasso-based approaches and better sensitivity over popular differential abundance testing methods for datasets with smaller sample sizes. In a survival analysis of allogeneic hematopoietic cell transplant recipients, FLORAL demonstrated considerable improvement in microbial feature selection by utilizing longitudinal microbiome data over solely using baseline microbiome data.

Multiple imputation with competing risk outcomes

AbstractIn time-to-event analyses, a competing risk is an event whose occurrence precludes the occurrence of the event of interest. Settings with competing risks occur frequently in clinical research. Missing data, which is a common problem in research, occurs when the value of a variable is recorded for some, but not all, records in the dataset. Multiple Imputation (MI) is a popular method to address the presence of missing data. MI uses an imputation model to generate M (M > 1) values for each variable that is missing, resulting in the creation of M complete datasets. A popular algorithm for imputing missing data is multivariate imputation using chained equations (MICE). We used a complex simulation design with covariates and missing data patterns reflective of patients hospitalized with acute myocardial infarction (AMI) to compare three strategies for imputing missing predictor variables when the analysis model is a cause-specific hazard when there were three different event types. We compared two MICE-based strategies that differed according to which cause-specific cumulative hazard functions were included in the imputation models (the three cause-specific cumulative hazard functions vs. only the cause-specific cumulative hazard function for the primary outcome) with the use of the substantive model compatible fully conditional specification (SMCFCS) algorithm. While no strategy had consistently superior performance compared to the other strategies, SMCFCS may be the preferred strategy. We illustrated the application of the strategies using a case study of patients hospitalized with AMI.

Competing risks and multivariate outcomes in epidemiological and clinical trial research.

Data analysis methods for the study of treatments or exposures in relation to a clinical outcome in the presence of competing risks have a long history, often with inference targets that are hypothetical, thereby requiring strong assumptions for identifiability with available data. Here data analysis methods are considered that are based on single and higher dimensional marginal hazard rates, quantities that are identifiable under standard independent censoring assumptions. These lead naturally to joint survival function estimators for outcomes of interest, including competing risk outcomes, and provide the basis for addressing a variety of data analysis questions. These methods will be illustrated using simulations and Women's Health Initiative cohort and clinical trial data sets, and additional research needs will be described.

Efficient estimation for left-truncated competing risks regression for case-cohort studies.

The case-cohort study design provides a cost-effective study design for a large cohort study with competing risk outcomes. The proportional subdistribution hazards model is widely used to estimate direct covariate effects on the cumulative incidence function for competing risk data. In biomedical studies, left truncation often occurs and brings extra challenges to the analysis. Existing inverse probability weighting methods for case-cohort studies with competing risk data not only have not addressed left truncation, but also are inefficient in regression parameter estimation for fully observed covariates. We propose an augmented inverse probability-weighted estimating equationfor left-truncated competing risk data to address these limitations of the current literature. We further propose a more efficient estimator when extra information from the other causes is available. The proposed estimators are consistent and asymptotically normally distributed. Simulation studies show that the proposed estimator is unbiased and leads to estimation efficiency gain in the regression parameter estimation. We analyze the Atherosclerosis Risk in Communities study data using the proposed methods.

Gaussian Copula Regression Modeling for Marker Classification Metrics with Competing Risk Outcomes

Decisions regarding competing risks are usually based on a continuous-valued marker toward predicting a cause-specific outcome. The classification power of a marker can be summarized using the time-dependent receiver operating characteristic curve and the corresponding area under the curve (AUC). This paper proposes a Gaussian copula-based model to represent the joint distribution of the continuous-valued marker, the overall survival time, and the cause-specific outcome. Then, it is used to characterize the time-varying ROC curve in the context of competing risks. Covariate effects are incorporated by linking linear components to the skewed normal distribution for the margin of the marker, a parametric proportional hazards model for the survival time, and a logit model for the cause of failure. Estimation is carried out using maximum likelihood, and a bootstrap technique is implemented to obtain confidence intervals for the AUC. Information-criteria strategies are employed to find a parsimonious model. The performance of the proposed model is evaluated in simulation studies, considering different sample sizes and censoring distributions. The methods are illustrated with the reanalysis of a prostate cancer clinical trial. The joint regression strategy produces a straightforward and flexible model of the time-dependent ROC curve in the presence of competing risks, enhancing the understanding of how covariates may affect the discrimination of a marker.

Volume under the ROC surface for high-dimensional independent screening with ordinal competing risk outcomes.

We propose a screening method for high-dimensional data with ordinal competing risk outcomes, which is time-dependent and model-free. Existing methods are designed for cause-specific variable screening and fail to evaluate how a biomarker is associated with multiple competing events simultaneously. The proposed method utilizes the Volume under the ROC surface (VUS), which measures the concordance between values of a biomarker and event status at certain time points and provides an overall evaluation of the discrimination capacity of a biomarker. We show that the VUS possesses the sure screening property, i.e., true important covariates can be retained with probability tending to one, and the size of the selected set can be bounded with high probability. The VUS appears to be a viable model-free screening metric as compared to some existing methods in simulation studies, and it is especially robust to data contamination. Through an analysis of breast-cancer gene-expression data, we illustrate the unique insights into the overall discriminatory capability provided by the VUS.

Comparison of Allogeneic Hematopoietic Cell Transplantation Outcomes from Younger Matched Unrelated Donor Versus Older Sibling Donor for Acute Myeloid Leukemia

Background: Aging population has resulted in an increase in the incidence of acute myeloid leukemia (AML). The median age of diagnosis for AML is 68 years. Allogeneic hematopoietic cell transplant (alloHCT) is a curative modality for patients with AML, however, disease relapse remains a major issue. Human leukocyte antigen (HLA)-matched sibling donors (MSD) are considered the preferred donor type in clinical practice. However, with increasing median age of patients, sibling donors for these patients are likely to be older as well, often with higher comorbidities. Prior registry studies demonstrated that older donor age is associated with inferior survival. A recent study in older MDS patients demonstrated reduced relapse risk with improved disease-free survival (DFS) with younger matched unrelated donors (MUD). However, the impact of donor age and type on alloHCT outcomes in AML patients remains unknown. Methods: This retrospective cohort study accessed data from the Center for International Blood and Marrow Transplant Research database (CIBMTR) in AML patients 50 years or older undergoing alloHCT from older (>50) MSDs or younger (<35) MUDs between 2011 and 2018. The study included common allograft types, conditioning regimens, and graft-versus-host-disease (GVHD) prophylaxis strategies. Other exclusion criteria were recipients of ex vivo T-cell depleted grafts, mismatched unrelated donor, cord blood or identical twin transplants. The primary outcome, compared between the 2 donor age groups, was relapse risk, and secondary outcomes included non-relapse mortality (NRM), DFS, and overall survival (OS). Cumulative incidence estimates were calculated for competing risks outcomes. To evaluate potential risk factors, multivariable cox regression was used, and relevant transplant-related covariates were considered. Interactions between the main effect (donor age group) and significant risk factors were tested. Fine and gray model was used for NRM and relapse. Results: Among 4684 eligible patients in the study cohort, 1736 underwent alloHCT with an older MSD (median age, 61 [range: 50-80]) whereas 2948 received transplant from a younger MUD (median age, 63 [range: 50-84]). In multivariable analysis, younger MUDs conferred a significantly decreased risk of relapse (HR 0.86; 95% CI, 0.77-0.96; p=.005) as compared to older MSDs (Table 1; Figure 1). The adjusted cumulative incidence of relapse at 5 years was significantly lower with younger MUDs (35%; 95% CI, 33-37%) versus older MSDs (41%; 95% CI, 38%-43%) (p=.003) (Table 2). NRM was reported in 2 subgroups of transplant years (2011-2015 and 2016-2018) as there was significant interaction between the main effect (MUD vs MSD) and alloHCT year (p=.003). Younger MUDs conferred an increased NRM (HR 1.24; 95% CI, 1.04-1.49; p=.016) compared to older MSDs in the earlier period of 2011-2015, whereas the NRM was lower with younger MUDs in the latter period of 2016-2018 (HR, 0.78, 95% CI, 0.64-0.96; p=.017) (Table1; Figure 1). While there was no difference in the adjusted cumulative incidence of NRM at 5 years in the latter period (17% vs 20%; p=.147), NRM was significantly higher in AML patients who underwent alloHCT between 2011 and 2015 from a younger MUD (28%; 95% CI, 25-30%) compared to older MSDs (24%; 95% CI, 20%-27%; p=.045) (Table2). There was no difference in OS or DFS rates of alloHCT with younger MUDs vs older MSDs (OS: HR 1.02; 95% CI, 0.94-1.12; p=.607; DFS: HR 0.92; 95% CI, 0.85-1.01; p=.073) (Table 1; Figure 1). While the adjusted 5-year OS probability was 47% in each of the 2 cohorts (p=0.95), the adjusted 5-year DFS probability was 44% with younger MUDs (95% CI, 42%-46%) as compared to 41% with older MSDs (95% CI, 38%-43%; p=.04) (Table2). Conclusions: This large study from a prospectively collected registry data found significant relapse reduction in AML patients who underwent alloHCT with younger MUDs as compared to older MSDs and highlights that a younger MUD donor type exerts a stronger GVL effect and should be preferred when available. The higher NRM associated with younger MUDs has decreased in recent years, likely due to enhanced GVHD prophylaxis, infection preemption and preventative strategies, and supportive cares measures translating into improved leukemia-free survival. The results aid in donor selection and younger MUDs should be preferred in AML patients at a higher risk for post-alloHCT relapse. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Younger Matched Unrelated Donors Confer a Decreased Relapse Risk As Compared to Older Sibling Donors for Adult B-Cell ALL Patients Undergoing Allogeneic Hematopoietic Cell Transplantation

Background: The incidence of acute lymphoblastic leukemia (ALL) has increased due to aging population and the median age at diagnosis is 56 years among adults. Allogeneic hematopoietic cell transplant (alloHCT) offers cure for ALL patients, however, disease relapse remains a major issue. Human leukocyte antigen (HLA)-matched sibling donors (MSD) are considered the ideal donor type, mainly due to lower incidence and severity of graft-versus-host-disease (GVHD) and improved disease-free survival (DFS) in ALL. However, siblings are also likely to be older, often with higher comorbidities, and adult ALL patients typically undergo reduced-intensity conditioning and need novel strategies to reduce the post-alloHCT relapse risk. While recent data have suggested relapse reduction and survival benefit with younger matched unrelated donors (MUD) in AML and MDS, the impact of donor age on alloHCT outcomes in adult ALL patients remain unknown. Methods: This retrospective cohort study from the Center for International Blood and Marrow Transplant Research database (CIBMTR) data in B-cell ALL patients 50 years or older, undergoing alloHCT from older (>50) MSDs or younger (<35) MUDs between 2011 and 2018, included common allograft types (PB vs BM), conditioning regimens (RIC/NMA vs MAC), and graft-versus-host-disease (GVHD) prophylaxis strategies (FK-based vs CSA-based vs. others). Other exclusion criteria were recipients of ex vivo T-cell depleted grafts, recipients of mismatched unrelated donors, cord blood or identical twin transplants. The primary outcome, compared between the 2 donor age groups, was relapse risk whereas secondary outcomes included non-relapse mortality (NRM), DFS, and overall survival (OS). Cumulative incidence estimates were calculated for competing risks outcomes, including NRM and relapse. The Kaplan-Meier method was used to estimate the probabilities for survival. To evaluate potential risk factors, multivariable cox regression was used, and relevant transplant-related covariates were considered. Interactions between the main effect (donor age group) and significant risk factors were tested. Fine and Gray model was used for NRM and relapse. Given the multicenter nature of the database, center effect was tested for OS by fitting a gamma frailty model. Results: Among 925 eligible patients in the study cohort, 386 underwent alloHCT with an older MSD (median age, 59 [range: 50-75]) whereas 539 received transplant from a younger MUD (median age, 60 [range: 50-77]). In multivariable analysis, younger MUDs conferred a significantly decreased risk of relapse (HR 0.68; 95% CI, 0.53-0.87; p=.002) versus older MSDs (Table 1; Figure 1). The adjusted cumulative incidence of relapse at 5 years was significantly lower with younger MUDs compared to older MSDs (26% vs 37%; p=.001) (Table 2). Younger MUDs conferred an increased NRM compared to older MSDs (HR 1.38; 95% CI, 1.05-1.82; p=.02) (Table1; Figure 1). The adjusted cumulative incidence of NRM at 5 years was also significantly higher in B-ALL patients who underwent alloHCT from younger MUDs compared to older MSDs (31% vs 22%; p=.006) (Table2). There were no differences in OS or DFS rates of alloHCT with younger MUDs vs older MSDs (OS: HR 1.09; 95% CI, 0.90-1.32; p=.37; DFS: HR 0.95; 95% CI, 0.79-1.14; p=.57) (Table 1; Figure 1). The adjusted 5-year OS probability was 48% in younger MUDs (95% CI, 43%-52%) and 51% among older MSDs (95% CI, 45%-56%; p=0.37). The adjusted 5-year DFS probability was 44% with younger MUDs (95% CI, 40%-49%) compared to 42% with older MSDs (95% CI, 37%-47%; p=.59) (Table2). Conclusions: This is the largest study to date performed in a mature, prospectively collected registry cohort that aids in donor selection and found significant relapse reduction in adult B-ALL patients who underwent alloHCT with younger MUDs as compared to older MSDs. The results highlight that younger MUD donor type exerts a stronger GVL effect and should be preferred in patients at a higher risk for post-alloHCT relapse. The higher NRM associated with younger MUDs will need enhanced GVHD prophylaxis, infection preemption and preventative strategies, and supportive cares measures to consolidate the leukemia-free survival. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Impact of HLA-B Leader Mismatching on Outcomes after Haploidentical Transplantation with Antithymocyte Globulin

Objective Haploidentical hematopoietic stem cell transplantation (haplo-HSCT) is increasingly used to treat hematologic diseases in the last 10 years. Continuous effort has been made to identify a high-risk HLA mismatch pattern to avoid detrimental factors to HSCT outcomes. However, whether HLA-B leader could be used to optimize donor selection in haplo-HSCT with antithymocyte globulin (ATG) is unknown. We thus sought to evaluate the role of HLA-B leader in haplo-HSCT with ATG which may help to optimize donor selection. Method We conducted a single-center retrospective analysis of 570 patients who received haplo-HSCT for the treatment of hematological malignancies between Sep, 2012 and May, 2021 in our institution to test a series of hypotheses for a better understanding of the significance of HLA-B leader in haplo-HSCT with ATG. HLA-B leader genotype and shared HLA-B leader was determined using HLA-B Leader Assessment Tool. Probabilities of survival were estimated by the Kaplan-Meier method. Cumulative incidences of competing risk outcomes were estimated by competing risk model. The Cox proportional hazard regression and the Fine-Gray competing risks regression model was used to define hazard ratio (HR) and conduct multivariate analysis. All statistical analysis was performed with R version 4.0.5. Results All patients received myeloablative conditioning regimen with ATG for graft-versus-host-disease prevention. Median follow-up was 649 days (interquartile range [IQR] 398-965 days) for surviving patients.Relative to HLA-B leader matched transplantation, HLA-B leader mismatched transplantation had lower overall-survival (OS) (HR, 1.44; 95% CI, 1.02-2.01; P=0.04; Figure 1A) and higher non-relapse mortality (NRM) (HR, 1.56; 95% CI, 1.04-2.35, P=0.03; Figure 1B). These results were further demonstrated by multivariable analysis. With respect to specific causes of NRM, leader mismatch group had a higher risk of infection-related death (HR, 2.30; 95% CI, 1.25-4.23; P=0.01). Additionally, we found that relapse was adversely affected by a shared M relative to a shared T allotype in the leader matched subgroup (HR, 1.92; 95% CI, 1.07-3.47, P=0.03; Figure 1C). In terms of disease-free-survival (DFS) and OS, no significant differences were identified between shared M and shared T in both leader matched subgroup and leader mismatched subgroup (Figure 1D; Figure 1E). Notably, patients benefited from HLA-B leader match with a shared T, which was associated with superior OS (HR, 0.71; 95% CI, 0.51-0.98; P=0.035) and modestly increased DFS (HR, 0.74; 95% CI, 0.55-1.00; P=0.051) when compared with the other three groups. Additionally, in the subgroup with shared T, HLA-B leader mismatch had a non-statistically significant trend of higher NRM (HR, 1.50; 95% CI, 0.97-2.31, P=0.07; Figure 1F). Similar association was observed in multivariable models that relative to leader match with shared T, leader match with shared M was associated with higher relapse (HR, 1.85; 95% CI, 1.02-3.36; P=0.045) and leader mismatch with shared T was associated with inferior OS (HR, 1.57; 95% CI, 1.08-2.28; P=0.017). Conclusion Our findings suggested that mismatched HLA-B leader informed increased risk of NRM and lower OS. Furthermore, patients benefited from HLA-B leader matched donors with shared T. A paradigm inclusive of HLA-B leader might help to optimize donor selection and improve clinical outcomes of haplo-HSCT with ATG. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

High-sensitive Troponin T is not associated with the progression of asymptomatic mild to moderate aortic stenosis: a post hoc substudy of the SEAS trial

Abstract Background Aortic stenosis (AS) and coronary artery disease (CAD) share pathophysiological pathways, as reflected by frequent concomitant revascularization in patients undergoing aortic valve replacement (AVR). High-sensitive Troponin T (hsTnT) is a proven biomarker of cardiomyocyte overload and injury, and predicts postoperative mortality after AVR. However, it is unknown if hsTnT can predict AVR, mortality or ischemic coronary events (ICE) in asymptomatic AS patients. Purpose To investigate the hypothesis that increased hsTnT is associated with more severe AS and a higher risk of adverse outcomes in asymptomatic AS patients without overt CAD. Methods hsTnT concentrations were examined at baseline and after 1-year follow-up in 1739 asymptomatic AS patients enrolled in the randomized, double-blind Simvastatin and Ezetimibe in Aortic Stenosis (SEAS) study. The main inclusion criteria were: left ventricular (LV) ejection fraction &amp;gt;55%, transaortic maximal velocity between 2.5–4.0 m/s, and no history of CAD. The primary exposure variable was increased hsTnT (&amp;gt;14 pg/mL according to the assay manufacturer, Roche). This study's primary endpoint was a composite of competing risk outcomes: all-cause mortality as the first event, AVR without revascularization, and ICE (defined as myocardial infarction before AVR, PCI before or combined with AVR, or any CABG). Multivariable regression examined associations between hsTnT and clinical variables. Cox proportional hazards regression models were adjusted for age, sex, creatinine, LV mass index, mean aortic pressure gradient (Pmean) and stratified by center and lipid-lowering treatment. We analyzed outcomes during 5-year follow-up from baseline. Results At baseline, 453 (26.0%) patients had increased hsTnT and 302 (17.4%) had moderate-severe AS with a mean (SD) aortic valve area of 0.8 (0.2) cm2 and Pmean of 33.2 (8.8) mmHg. The median annual hsTnT change from baseline to year 1 was 0.8 pg/mL (IQR, −0.4 to 2.3), regardless of AS severity (P=0.08). In adjusted models, log(hsTnT at baseline) was associated with age, sex, creatinine, and LV mass index (all P&amp;lt;0.05), but not with AS severity (P=0.36). The incidence rate ratio for ICE (Figure 1) in patients with increased vs normal baseline hsTnT concentrations was 2.32 (95% CI, 1.72–3.11, P&amp;lt;0.001). In adjusted Cox regression, increased hsTnT was associated with an increased 5-year ICE risk (HR 1.64; 95% CI, 1.18–2.29, P=0.003), but neither with AVR without revascularization nor death (Figure 1). Conclusion In these asymptomatic AS patients without overt CAD, hsTnT is often normal and remains stable during 1 year of follow-up regardless of AS severity. Increased hsTnT is associated with CAD-related events, but neither to AS severity nor AVR without concomitant revascularization. This analysis does not support routine hsTnT measurement in asymptomatic AS to predict AVR related to AS progression, although hsTnT could improve the risk assessment for ICE. Funding Acknowledgement Type of funding sources: Private company. Main funding source(s): Main sponsor (SEAS): Merck &amp; Co Inc, Whitehouse Station, New JerseyBlood analysis sponsor: Roche Diagnostics International Ltd, Switzerland

Marginal semiparametric transformation models for clustered multivariate competing risks data.

Multivariate survival models are often used in studying multiple outcomes for right-censored data. However, the outcomes of interest often have competing risks, where standard multivariate survival models may lead to invalid inferences. For example, patients who had stem cell transplantation may experience multiple types of infections after transplant while reconstituting their immune system, where death without experiencing infections is a competing risk for infections. Such competing risks data often suffer from cluster effects due to a matched pair design or correlation within study centers. The cumulative incidence function (CIF) is widely used to summarize competing risks outcomes. Thus, it is often of interest to study direct covariate effects on the CIF. Most literature on clustered competing risks data analyses is limited to the univariate proportional subdistribution hazards model with inverse probability censoring weighting which requires correctly specifying the censoring distribution. We propose a marginal semiparametric transformation model for multivariate competing risks outcomes. The proposed model does not require modeling the censoring distribution, accommodates nonproportional subdistribution hazards structure, and provides a platform for joint inference of all causes and outcomes.

Finite‐sample adjustments in variance estimators for clustered competing risks regression

The marginal Fine-Gray proportional subdistribution hazards model is a popular approach to directly study the association between covariates and the cumulative incidence function with clustered competing risks data, which often arise in multicenter randomized trials or multilevel observational studies. To account for the within-cluster correlations between failure times, the uncertainty of the regression parameters estimators is quantified by the robust sandwich variance estimator, which may have unsatisfactory performance with a limited number of clusters. To overcome this limitation, we propose four bias-corrected variance estimators to reduce the negative bias of the usual sandwich variance estimator, extending the bias-correction techniques from generalized estimating equations with noncensored exponential family outcomes to clustered competing risks outcomes. We further compare their finite-sample operating characteristics through simulations and two real data examples. In particular, we found the Mancl and DeRouen (MD) type sandwich variance estimator generally has the smallest bias. Furthermore, with a small number of clusters, the Wald -confidence interval with the MD sandwich variance estimator carries close to nominal coverage for the cluster-level effect parameter. The -confidence intervals based on the sandwich variance estimator with any one of the three types of multiplicative bias correction or the -confidence interval with the Morel, Bokossa and Neerchal (MBN) type sandwich variance estimator have close to nominal coverage for the individual-level effect parameter. Finally, we develop a user-friendly R package crrcbcv implementing the proposed sandwich variance estimators to assist practical applications.

Prognostic accuracy for predicting ordinal competing risk outcomes using ROC surfaces.

Many medical conditions are marked by a sequence of events in association with continuous changes in biomarkers. Few works have evaluated the overall accuracy of a biomarker in predicting disease progression. We thus extend the concept of receiver operating characteristic (ROC) surface and the volume under the surface (VUS) from multi-category outcomes to ordinal competing-risk outcomes that are also subject to noninformative censoring. Two VUS estimators are considered. One is based on the definition of the ROC surface and obtained by integrating the estimated ROC surface. The other is an inverse probability weighted U estimator that is built upon the equivalence of the VUS to the concordance probability between the marker and sequential outcomes. Both estimators have nice asymptotic results that can be derived using counting process techniques and U-statistics theory. We illustrate their good practical performances through simulations and applications to two studies of cognition and a transplant dataset.

Evaluating the association between latent classes and competing risks outcomes with multiphenotype data.

Latent class analysis is an intuitive tool to characterize disease phenotype heterogeneity. With data more frequently collected on multiple phenotypes in chronic disease studies, it is of rising interest to investigate how the latent classes embedded in one phenotype are related to another phenotype. Motivated by a cohort with mild cognitive impairment (MCI) from the Uniform Data Set (UDS), we propose and study a time-dependent structural model to evaluate the association between latent classes and competing risk outcomes that are subject to missing failure types. We develop a two-step estimation procedure which circumvents latent class membership assignment and is rigorously justified in terms of accounting for the uncertainty in classifying latent classes. The new method also properly addresses the realistic complications for competing risks outcomes, including random censoring and missing failure types. The asymptotic properties of the resulting estimator are established. Given that the standard bootstrapping inference is not feasible in the current problem setting, we develop analytical inference procedures, which are easy to implement. Our simulation studies demonstrate the advantages of the proposed method over benchmark approaches. We present an application to the MCI data from UDS, which uncovers a detailed picture of the neuropathological relevance of the baseline MCIsubgroups.

Joint analysis of longitudinal measurements and spatially clustered competing risks HIV/AIDS data.

The joint modeling of repeated measurements and time-to-event provides a general framework to describe better the link between the progression of disease through longitudinal measurements and time-to-event outcome. In the survival data, a sample of individuals is frequently grouped into clusters. In some applications, these clusters could be arranged spatially, for example, based on geographical regions. There are two benefits of considering spatial variation in these data, enhancing the efficiency and accuracy of the parameters estimations, and investigating the survivorship spatial pattern. On the other hand, in survival data, there is a situation that subjects are supposed to experience more than one type of event potentially, but the occurrence of one type of event prevents the occurrence of the others. In this article, we considered a Bayesian joint model of longitudinal and competing risks outcomes for spatially clustered HIV/AIDS data. The data were from a registry-based study carried in Hamadan Province, Iran, from December 1997 to June 2020. In this joint model, a linear mixed effects model was used for the longitudinal submodel and a cause-specific hazard model with spatial and spatial-risk random effects was used for the survival submodel. Also, a latent structure was defined by random effects to link both event times and longitudinal processes. We used a univariate intrinsic conditional autoregressive (ICAR) distribution and a multivariate ICAR distribution for modeling the areal spatial and spatial-risk random effects, respectively. The performance of our proposed model using simulation studies and analysis of HIV/AIDS data were assessed.

Instrumental variable estimation of early treatment effect in randomized screening trials.

The primary analysis of randomized screening trials for cancer typically adheres to the intention-to-screen principle, measuring cancer-specific mortality reductions between screening and control arms. These mortality reductions result from a combination of the screening regimen, screening technology and the effect of the early, screening-induced, treatment. This motivates addressing these different aspects separately. Here we are interested in the causal effect of early versus delayed treatments on cancer mortality among the screening-detectable subgroup, which under certain assumptions is estimable from conventional randomized screening trial using instrumental variable type methods. To define the causal effect of interest, we formulate a simplified structural multi-state model for screening trials, based on a hypothetical intervention trial where screening detected individuals would be randomized into early versus delayed treatments. The cancer-specific mortality reductions after screening detection are quantified by a cause-specific hazard ratio. For this, we propose two estimators, based on an estimating equation and a likelihood expression. The methods extend existing instrumental variable methods for time-to-event and competing risks outcomes to time-dependent intermediate variables. Using the multi-state model as the basis of a data generating mechanism, we investigate the performance of the new estimators through simulation studies. In addition, we illustrate the proposed method in the context of CT screening for lung cancer using the US National Lung Screening Trial data.

Patient factors associated with oral generic olanzapine initiation and substitution among Medicaid beneficiaries: a new user cohort study

Objective Generic drugs typically are less expensive than branded products; however, several factors can limit generic drug utilization. This study assesses the associations of patient factors with generic olanzapine initiation and substitution. Methods A retrospective new user cohort study was conducted using the 2011–2012 Medicaid administrative claims data. Beneficiaries continuously enrolled during the 6 month washout period prior to their initial oral brand or generic olanzapine prescription were included and followed up to 12 months. Among brand olanzapine new users, time to generic substitution and competing risk outcomes was estimated using the Fine–Gray cumulative incidence function. Patient demographic and health service utilization factors were assessed in the multivariate cause-specific hazards model. Results Among olanzapine new users, 70.7% patients initiated generic treatment. Beneficiaries aged ≥21, and living in the Midwest and West regions were more likely to initiate generic olanzapine. Among brand new users, 28.2% switched to generic olanzapine, 23.6% switched to an alternative atypical antipsychotic treatment and 38.0% discontinued within 12 months. Beneficiaries who resided in urban areas (adjusted hazard ratio [AHR) = 0.53, 95% CI = 0.37–0.75) and had prior hospitalizations (AHR = 0.85, 95% CI = 0.75–0.96) had lower rates of generic substitution, whereas those with emergency department (ED) visits (AHR = 1.06, 95% CI = 1.02–1.10) had a higher rate of generic substitution. In addition, beneficiaries in different age subgroups also had different rates of generic substitution in different regions. Conclusion Medicaid beneficiaries’ age, geographic region, prior hospitalization and ED utilization were associated with generic olanzapine initiation and substitution. Tailored educational outreach targeting these patient subgroups might improve generic olanzapine utilization.

Autologous Vs. Allogeneic Stem Cell Transplantation in Double-Expressor Lymphoma

Purpose: Dual expression of MYC and BCL2 proteins (Double Expressor Lymphoma-[DEL]) is an important prognostic factor in patients (pts) with diffuse large B-cell lymphoma (DLBCL) who are treated with standard chemo-immunotherapy. Studies have suggested that pts with these biomarkers have inferior survival compared to pts who were non-DEL after autologous stem cell transplantation (autoSCT). For this reason, allogeneic stem cell transplantation (alloSCT) has been advocated for its potential graft-versus-lymphoma effect. Data are limited however regarding the outcomes in pts with relapsed DEL who were treated with autoSCT vs. alloSCT. Methods: Data from pts with relapsed DEL/DLBCL who underwent auto- and alloSCT as their first transplant at our center and in whom archived tumor material was available were analyzed. Cutoff values of 40% for MYC and 70% for BCL2 were established by immunohistochemistry (IHC). The majority of autoSCT pts (84%) underwent chemo-mobilization of stem cells with rituximab (R) for in-vivo purging; rituximab was also given on days+1 and +8 with BEAM conditioning (J Clin Oncol 2005; 23:2240-7; Clin Cancer Res 2018; 24:2304-11). Conditioning for the alloSCT was myeloablative (n=16, 50%), reduced-intensity (n=6, 19%), and nonmyelablative (n=10,31%). The study was IRB-approved at our center. Kaplan-Meier analysis and corresponding log-rank/Gray's tests were used to estimate and compare overall survival (OS), progression-free survival (PFS), and cumulative incidence (CI) of non-relapse mortality and relapse by SCT group. Standard multivariable Cox regression was used to evaluate associations between SCT type and OS/PFS after controlling for differences in baseline characteristics and cause-specific multivariable Cox regression was used for competing-risk outcomes [non-relapse mortality (NRM) and relapse]. Results: The study included 161 autoSCT and 32 alloSCT pts treated between 2000-2018. Median age was 59 (range,18-80) and 55 years (range, 21-66), respectively (P=0.009). Male gender was 62.7 vs 50.0%, respectively (P= 0.234). Fifty-one (32.5%) autoSCT pts and 9 (28.1%) alloSCT pts had an elevated serum LDH (P= 0.683) and 41/148 (27.7%) autoSCT and 11/25 (44.0%) alloSCT pts were PET-positive at study entry (P=0.106). The time from diagnosis to transplant was 20.0 and 25.2 months, respectively, (P=0.068), and the distribution of the years of transplant was similar between both groups (P=0.317). Significant differences between treatment groups were observed for prior number of chemotherapies, disease status, stage, and HCT-CI score at transplant. Pts. receiving alloSCT were more heavily pretreated [median 3 prior therapies, (range 1-8) vs. a median of 2, (range 1-6); P=0.0001], had more advanced stage III-IV disease (P=0.035), and more refractory disease (P=0.010) at transplant. In addition, a higher percentage of alloSCT pts had an HCT-CI of &amp;gt;4 (28.1% vs. 10.2%, P=0.018) at study entry. Allogeneic transplant characteristics included matched siblings (n=18; 56.3%), matched unrelated (n=13, 40.6%) and haplo-identical (n=1, 3.1%) donors. With a median follow-up for autoSCT surviving pts of 65 months (range, 5-217 months) and 93 months (range, 34--124 months) for alloSCT pts, the OS at 5-years were 59% and 34%, respectively (P= 0.0001) (Figure1). The PFS rates at 5-years were 49% and 31%, respectively (P = 0.002). AutoSCT had a similar rate of relapse to alloSCT but a lower rate of NRM. The 5-year CI of relapse was 37% vs 28%, respectively (P = 0.611) and the 5-year NRM rates were 14% vs 41 % (P=0.0001), respectively (Figure 2). The associations between SCT group and outcomes persisted after adjusting for differences between groups (alloSCT vs. autoSCT HR= 3.46/p=0.0004 for OS; HR=2.67/p=0.005 for PFS; HR=4.74/p=0.009 for NRM; HR=2.22/p=0.080 for relapse). The 100-day CI of grade II-IV and III-IV acute GVHD in the alloSCT group was 37.5 and 9.4%, respectively. The 1-year CI of chronic GVHD was 25.0%. C onclusions: This study is the first to show that autoSCT confers a superior survival in pts with relapsed DEL/DLBCL compared to alloSCT. Our conclusions are supported by long-term follow-up. The use of novel nonmyeloablative regimens in a larger number of pts receiving an alloSCT and ongoing studies at our center with targeted investigational agents after autoSCT may improve results. Disclosures Khouri: Pfizer: Research Funding; Bristol Myers Squibb: Research Funding. Iyer:CRISPR: Research Funding; Spectrum: Research Funding; Daiichi Sankyo: Consultancy; Trillium: Research Funding; Curio Biosciences: Honoraria; Target Oncology: Honoraria; Afffimed: Research Funding; Merck: Research Funding; Legend Biotech: Consultancy; Rhizen: Research Funding; Seattle Genetics, Inc.: Research Funding. Popat:Bayer: Research Funding; Novartis: Research Funding. Qazilbash:Angiocrine: Research Funding; Amgen: Research Funding; Bioclinica: Consultancy; Janssen: Research Funding; Bioline: Research Funding. Champlin:DKMS America: Membership on an entity's Board of Directors or advisory committees; Actinium: Consultancy; Johnson and Johnson: Consultancy; Omeros: Consultancy; Cytonus: Consultancy; Takeda: Patents &amp; Royalties; Genzyme: Speakers Bureau.

A Bayesian adaptive phase I/II clinical trial design with late‐onset competing risk outcomes

Early-phase dose-finding clinical trials are often subject to the issue of late-onset outcomes. In phase I/II clinical trials, the issue becomes more intractable because toxicity and efficacy can be competing risk outcomes such that the occurrence of the first outcome will terminate the other one. In this paper, we propose a novel Bayesian adaptive phase I/II clinical trial design to address the issue of late-onset competing risk outcomes. We use the continuation-ratio model to characterize the trinomial response outcomes and the cause-specific hazard rate method to model the competing-risk survival outcomes. We treat the late-onset outcomes as missing data and develop a Bayesian data augmentation method to impute the missing data from the observations. We also propose an adaptive dose-finding algorithm to allocate patients and identify the optimal biological dose during the trial. Simulation studies show that the proposed design yields desirable operatingcharacteristics.

Sensitivity analysis of treatment effect to unmeasured confounding in observational studies with survival and competing risks outcomes.

No unmeasured confounding is often assumed in estimating treatment effects in observational data, whether using classical regression models or approaches such as propensity scores and inverse probability weighting. However, in many such studies collection of confounders cannot possibly be exhaustive in practice, and it is crucial to examine the extent to which the resulting estimate is sensitive to the unmeasured confounders. We consider this problem for survival and competing risks data. Due to the complexity of models for such data, we adapt the simulated potential confounder approach of Carnegie et al(2016), which provides a general tool for sensitivity analysis due to unmeasured confounding. More specifically, we specify one sensitivity parameter to quantify the association between an unmeasured confounder and the exposure or treatment received, and another set of parameters to quantify the association between the confounder and the time-to-event outcomes. By varying the magnitudes of the sensitivity parameters, we estimate the treatment effect of interest using the stochastic expectation-maximization (EM) and the EM algorithms. We demonstrate the performance of our methods on simulated data, and apply them to a comparative effectiveness study in inflammatory bowel disease. An R package "survSens" is available on CRAN that implements the proposed methodology.