Eflornithine as Postimmunotherapy Maintenance in High-Risk Neuroblastoma: Externally Controlled, Propensity Score-Matched Survival Outcome Comparisons.

PURPOSE: Long-term survival in high-risk neuroblastoma (HRNB) is approximately 50%, with mortality primarily driven by relapse. Eflornithine (DFMO) to reduce risk of relapse after completion of immunotherapy was investigated previously in a single-arm, phase II study (NMTRC003B; ClinicalTrials.gov identifier: NCT02395666) that suggested improved event-free survival (EFS) and overall survival (OS) compared with historical rates in a phase III trial (Children Oncology Group ANBL0032; ClinicalTrials.gov identifier: NCT00026312). Using patient-level data from ANBL0032 as an external control, we present new analyses to further evaluate DFMO as HRNB postimmunotherapy maintenance. PATIENTS AND METHODS: NMTRC003B (2012-2016) enrolled patients with HRNB (N = 141) after standard up-front or refractory/relapse treatment who received up to 2 years of continuous treatment with oral DFMO (750 ± 250 mg/m2 twice a day). ANBL0032 (2001-2015) enrolled patients with HRNB postconsolidation, 1,328 of whom were assigned to dinutuximab (ch.14.18) treatment. Selection rules identified 92 NMTRC003B patients who participated in (n = 87) or received up-front treatment consistent with (n = 5) ANBL0032 (the DFMO/treated group) and 852 patients from ANBL0032 who could have been eligible for NMTRC003B after immunotherapy, but did not enroll (the NO-DFMO/control group). The median follow-up time for DFMO/treated patients was 6.1 years (IQR, 5.2-7.2) versus 5.0 years (IQR, 3.5-7.0) for NO-DFMO/control patients. Kaplan-Meier and Cox regression compared EFS and OS for overall groups, 3:1 (NO-DFMO:DFMO) propensity score-matched cohorts balanced on 11 baseline demographic and disease characteristics with exact matching on MYCN, and additional sensitivity analyses. RESULTS: DFMO after completion of immunotherapy was associated with improved EFS (hazard ratio [HR], 0.50 [95% CI, 0.29 to 0.84]; P = .008) and OS (HR, 0.38 [95% CI, 0.19 to 0.76]; P = .007). The results were confirmed with propensity score-matched cohorts and sensitivity analyses. CONCLUSION: The externally controlled analyses presented show a relapse risk reduction in patients with HRNB treated with postimmunotherapy DFMO.

REMAP Periop: a randomised, embedded, multifactorial adaptive platform trial protocol for perioperative medicine to determine the optimal enhanced recovery pathway components in complex abdominal surgery patients within a US healthcare system.

INTRODUCTION: Implementation of enhanced recovery pathways (ERPs) has resulted in improved patient-centred outcomes and decreased costs. However, there is a lack of high-level evidence for many ERP elements. We have designed a randomised, embedded, multifactorial, adaptive platform perioperative medicine (REMAP Periop) trial to evaluate the effectiveness of several perioperative therapies for patients undergoing complex abdominal surgery as part of an ERP. This trial will begin with two domains: postoperative nausea/vomiting (PONV) prophylaxis and regional/neuraxial analgesia. Patients enrolled in the trial will be randomised to arms within both domains, with the possibility of adding additional domains in the future. METHODS AND ANALYSIS: In the PONV domain, patients are randomised to optimal versus supraoptimal prophylactic regimens. In the regional/neuraxial domain, patients are randomised to one of five different single-injection techniques/combination of techniques. The primary study endpoint is hospital-free days at 30 days, with additional domain-specific secondary endpoints of PONV incidence and postoperative opioid consumption. The efficacy of an intervention arm within a given domain will be evaluated at regular interim analyses using Bayesian statistical analysis. At the beginning of the trial, participants will have an equal probability of being allocated to any given intervention within a domain (ie, simple 1:1 randomisation), with response adaptive randomisation guiding changes to allocation ratios after interim analyses when applicable based on prespecified statistical triggers. Triggers met at interim analysis may also result in intervention dropping. ETHICS AND DISSEMINATION: The core protocol and domain-specific appendices were approved by the University of Pittsburgh Institutional Review Board. A waiver of informed consent was obtained for this trial. Trial results will be announced to the public and healthcare providers once prespecified statistical triggers of interest are reached as described in the core protocol, and the most favourable interventions will then be implemented as a standardised institutional protocol. TRIAL REGISTRATION NUMBER: NCT04606264.

Effect of Antiplatelet Therapy on Survival and Organ Support-Free Days in Critically Ill Patients With COVID-19: A Randomized Clinical Trial.

Importance: The efficacy of antiplatelet therapy in critically ill patients with COVID-19 is uncertain. Objective: To determine whether antiplatelet therapy improves outcomes for critically ill adults with COVID-19. Design, Setting, and Participants: In an ongoing adaptive platform trial (REMAP-CAP) testing multiple interventions within multiple therapeutic domains, 1557 critically ill adult patients with COVID-19 were enrolled between October 30, 2020, and June 23, 2021, from 105 sites in 8 countries and followed up for 90 days (final follow-up date: July 26, 2021). Interventions: Patients were randomized to receive either open-label aspirin (n = 565), a P2Y12 inhibitor (n = 455), or no antiplatelet therapy (control; n = 529). Interventions were continued in the hospital for a maximum of 14 days and were in addition to anticoagulation thromboprophylaxis. Main Outcomes and Measures: The primary end point was organ support-free days (days alive and free of intensive care unit-based respiratory or cardiovascular organ support) within 21 days, ranging from -1 for any death in hospital (censored at 90 days) to 22 for survivors with no organ support. There were 13 secondary outcomes, including survival to discharge and major bleeding to 14 days. The primary analysis was a bayesian cumulative logistic model. An odds ratio (OR) greater than 1 represented improved survival, more organ support-free days, or both. Efficacy was defined as greater than 99% posterior probability of an OR greater than 1. Futility was defined as greater than 95% posterior probability of an OR less than 1.2 vs control. Intervention equivalence was defined as greater than 90% probability that the OR (compared with each other) was between 1/1.2 and 1.2 for 2 noncontrol interventions. Results: The aspirin and P2Y12 inhibitor groups met the predefined criteria for equivalence at an adaptive analysis and were statistically pooled for further analysis. Enrollment was discontinued after the prespecified criterion for futility was met for the pooled antiplatelet group compared with control. Among the 1557 critically ill patients randomized, 8 patients withdrew consent and 1549 completed the trial (median age, 57 years; 521 [33.6%] female). The median for organ support-free days was 7 (IQR, -1 to 16) in both the antiplatelet and control groups (median-adjusted OR, 1.02 [95% credible interval {CrI}, 0.86-1.23]; 95.7% posterior probability of futility). The proportions of patients surviving to hospital discharge were 71.5% (723/1011) and 67.9% (354/521) in the antiplatelet and control groups, respectively (median-adjusted OR, 1.27 [95% CrI, 0.99-1.62]; adjusted absolute difference, 5% [95% CrI, -0.2% to 9.5%]; 97% posterior probability of efficacy). Among survivors, the median for organ support-free days was 14 in both groups. Major bleeding occurred in 2.1% and 0.4% of patients in the antiplatelet and control groups (adjusted OR, 2.97 [95% CrI, 1.23-8.28]; adjusted absolute risk increase, 0.8% [95% CrI, 0.1%-2.7%]; 99.4% probability of harm). Conclusions and Relevance: Among critically ill patients with COVID-19, treatment with an antiplatelet agent, compared with no antiplatelet agent, had a low likelihood of providing improvement in the number of organ support-free days within 21 days. Trial Registration: ClinicalTrials.gov Identifier: NCT02735707.

Association Between Administration of IL-6 Antagonists and Mortality Among Patients Hospitalized for COVID-19: A Meta-analysis.

Importance: Clinical trials assessing the efficacy of IL-6 antagonists in patients hospitalized for COVID-19 have variously reported benefit, no effect, and harm. Objective: To estimate the association between administration of IL-6 antagonists compared with usual care or placebo and 28-day all-cause mortality and other outcomes. Data Sources: Trials were identified through systematic searches of electronic databases between October 2020 and January 2021. Searches were not restricted by trial status or language. Additional trials were identified through contact with experts. Study Selection: Eligible trials randomly assigned patients hospitalized for COVID-19 to a group in whom IL-6 antagonists were administered and to a group in whom neither IL-6 antagonists nor any other immunomodulators except corticosteroids were administered. Among 72 potentially eligible trials, 27 (37.5%) met study selection criteria. Data Extraction and Synthesis: In this prospective meta-analysis, risk of bias was assessed using the Cochrane Risk of Bias Assessment Tool. Inconsistency among trial results was assessed using the I2 statistic. The primary analysis was an inverse variance-weighted fixed-effects meta-analysis of odds ratios (ORs) for 28-day all-cause mortality. Main Outcomes and Measures: The primary outcome measure was all-cause mortality at 28 days after randomization. There were 9 secondary outcomes including progression to invasive mechanical ventilation or death and risk of secondary infection by 28 days. Results: A total of 10 930 patients (median age, 61 years [range of medians, 52-68 years]; 3560 [33%] were women) participating in 27 trials were included. By 28 days, there were 1407 deaths among 6449 patients randomized to IL-6 antagonists and 1158 deaths among 4481 patients randomized to usual care or placebo (summary OR, 0.86 [95% CI, 0.79-0.95]; P = .003 based on a fixed-effects meta-analysis). This corresponds to an absolute mortality risk of 22% for IL-6 antagonists compared with an assumed mortality risk of 25% for usual care or placebo. The corresponding summary ORs were 0.83 (95% CI, 0.74-0.92; P < .001) for tocilizumab and 1.08 (95% CI, 0.86-1.36; P = .52) for sarilumab. The summary ORs for the association with mortality compared with usual care or placebo in those receiving corticosteroids were 0.77 (95% CI, 0.68-0.87) for tocilizumab and 0.92 (95% CI, 0.61-1.38) for sarilumab. The ORs for the association with progression to invasive mechanical ventilation or death, compared with usual care or placebo, were 0.77 (95% CI, 0.70-0.85) for all IL-6 antagonists, 0.74 (95% CI, 0.66-0.82) for tocilizumab, and 1.00 (95% CI, 0.74-1.34) for sarilumab. Secondary infections by 28 days occurred in 21.9% of patients treated with IL-6 antagonists vs 17.6% of patients treated with usual care or placebo (OR accounting for trial sample sizes, 0.99; 95% CI, 0.85-1.16). Conclusions and Relevance: In this prospective meta-analysis of clinical trials of patients hospitalized for COVID-19, administration of IL-6 antagonists, compared with usual care or placebo, was associated with lower 28-day all-cause mortality. Trial Registration: PROSPERO Identifier: CRD42021230155.

Propensity score weighting for causal subgroup analysis.

A common goal in comparative effectiveness research is to estimate treatment effects on prespecified subpopulations of patients. Though widely used in medical research, causal inference methods for such subgroup analysis (SGA) remain underdeveloped, particularly in observational studies. In this article, we develop a suite of analytical methods and visualization tools for causal SGA. First, we introduce the estimand of subgroup weighted average treatment effect and provide the corresponding propensity score weighting estimator. We show that balancing covariates within a subgroup bounds the bias of the estimator of subgroup causal effects. Second, we propose to use the overlap weighting (OW) method to achieve exact balance within subgroups. We further propose a method that combines OW and LASSO, to balance the bias-variance tradeoff in SGA. Finally, we design a new diagnostic graph-the Connect-S plot-for visualizing the subgroup covariate balance. Extensive simulation studies are presented to compare the proposed method with several existing methods. We apply the proposed methods to the patient-centered results for uterine fibroids (COMPARE-UF) registry data to evaluate alternative management options for uterine fibroids for relief of symptoms and quality of life.

A subset analysis of a phase II trial evaluating the use of DFMO as maintenance therapy for high-risk neuroblastoma.

Neuroblastoma is a sympathetic nervous system tumor, primarily presenting in children under 6 years of age. The long-term prognosis for patients with high-risk neuroblastoma (HRNB) remains poor despite aggressive multimodal therapy. This report provides an update to a phase II trial evaluating DFMO as maintenance therapy in HRNB. Event-free survival (EFS) and overall survival (OS) of 81 subjects with HRNB treated with standard COG induction, consolidation and immunotherapy followed by 2 years of DFMO on the NMTRC003/003b Phase II trial were compared to a historical cohort of 76 HRNB patients treated at Beat Childhood Cancer Research Consortium (BCC) hospitals who were disease-free after completion of standard upfront therapy and did not receive DFMO. The 2- and 5-year EFS were 86.4% [95% confidence interval (CI) 79.3%-94.2%] and 85.2% [77.8%-93.3%] for the NMTRC003/003b subset vs 78.3% [69.5%-88.3%] and 65.6% [55.5%-77.5%] for the historical control group. The 2- and 5-year OS were 98.8% [96.4-100%] and 95.1% [90.5%-99.9%] vs 94.4% [89.3%-99.9%] and 81.6% [73.0%-91.2%], respectively. DFMO maintenance for HRNB after completion of standard of care therapy was associated with improved EFS and OS relative to historical controls treated at the same institutions. These results support additional investigations into the potential role of DFMO in preventing relapse in HRNB.

Development and validation of machine learning models to identify high-risk surgical patients using automatically curated electronic health record data (Pythia): A retrospective, single-site study.

BACKGROUND: Pythia is an automated, clinically curated surgical data pipeline and repository housing all surgical patient electronic health record (EHR) data from a large, quaternary, multisite health institute for data science initiatives. In an effort to better identify high-risk surgical patients from complex data, a machine learning project trained on Pythia was built to predict postoperative complication risk. METHODS AND FINDINGS: A curated data repository of surgical outcomes was created using automated SQL and R code that extracted and processed patient clinical and surgical data across 37 million clinical encounters from the EHRs. A total of 194 clinical features including patient demographics (e.g., age, sex, race), smoking status, medications, comorbidities, procedure information, and proxies for surgical complexity were constructed and aggregated. A cohort of 66,370 patients that had undergone 99,755 invasive procedural encounters between January 1, 2014, and January 31, 2017, was studied further for the purpose of predicting postoperative complications. The average complication and 30-day postoperative mortality rates of this cohort were 16.0% and 0.51%, respectively. Least absolute shrinkage and selection operator (lasso) penalized logistic regression, random forest models, and extreme gradient boosted decision trees were trained on this surgical cohort with cross-validation on 14 specific postoperative outcome groupings. Resulting models had area under the receiver operator characteristic curve (AUC) values ranging between 0.747 and 0.924, calculated on an out-of-sample test set from the last 5 months of data. Lasso penalized regression was identified as a high-performing model, providing clinically interpretable actionable insights. Highest and lowest performing lasso models predicted postoperative shock and genitourinary outcomes with AUCs of 0.924 (95% CI: 0.901, 0.946) and 0.780 (95% CI: 0.752, 0.810), respectively. A calculator requiring input of 9 data fields was created to produce a risk assessment for the 14 groupings of postoperative outcomes. A high-risk threshold (15% risk of any complication) was determined to identify high-risk surgical patients. The model sensitivity was 76%, with a specificity of 76%. Compared to heuristics that identify high-risk patients developed by clinical experts and the ACS NSQIP calculator, this tool performed superiorly, providing an improved approach for clinicians to estimate postoperative risk for patients. Limitations of this study include the missingness of data that were removed for analysis. CONCLUSIONS: Extracting and curating a large, local institution's EHR data for machine learning purposes resulted in models with strong predictive performance. These models can be used in clinical settings as decision support tools for identification of high-risk patients as well as patient evaluation and care management. Further work is necessary to evaluate the impact of the Pythia risk calculator within the clinical workflow on postoperative outcomes and to optimize this data flow for future machine learning efforts.