Methods for non-proportional hazards in clinical trials: A systematic review.

For the analysis of time-to-event data, frequently used methods such as the log-rank test or the Cox proportional hazards model are based on the proportional hazards assumption, which is often debatable. Although a wide range of parametric and non-parametric methods for non-proportional hazards has been proposed, there is no consensus on the best approaches. To close this gap, we conducted a systematic literature search to identify statistical methods and software appropriate under non-proportional hazard. Our literature search identified 907 abstracts, out of which we included 211 articles, mostly methodological ones. Review articles and applications were less frequently identified. The articles discuss effect measures, effect estimation and regression approaches, hypothesis tests, and sample size calculation approaches, which are often tailored to specific non-proportional hazard situations. Using a unified notation, we provide an overview of methods available. Furthermore, we derive some guidance from the identified articles.

Clinical course and follow-up of pediatric patients with COVID-19 vaccine-associated myocarditis compared to non-vaccine-associated myocarditis within the prospective multicenter registry-"MYKKE".

BACKGROUND: Since the onset of widespread COVID-19 vaccination, increased incidence of COVID-19 vaccine-associated myocarditis (VA-myocarditis) has been noted, particularly in male adolescents. METHODS: Patients <18 years with suspected myocarditis following COVID-19 vaccination within 21 days were enrolled in the PedMYCVAC cohort, a substudy within the prospective multicenter registry for pediatric myocarditis "MYKKE." Clinical data at initial admission, 3- and 9-months follow-up were monitored and compared to pediatric patients with confirmed non-vaccine-associated myocarditis (NVA-myocarditis) adjusting for various baseline characteristics. RESULTS: From July 2021 to December 2022, 56 patients with VA-myocarditis across 15 centers were enrolled (median age 16.3 years, 91% male). Initially, 11 patients (20%) had mildly reduced left ventricular ejection fraction (LVEF; 45%-54%). No incidents of severe heart failure, transplantation or death were observed. Of 49 patients at 3-months follow-up (median (IQR) 94 (63-118) days), residual symptoms were registered in 14 patients (29%), most commonly atypical intermittent chest pain and fatigue. Diagnostic abnormalities remained in 23 patients (47%). Of 21 patients at 9-months follow-up (259 (218-319) days), all were free of symptoms and diagnostic abnormalities remained in 9 patients (43%). These residuals were mostly residual late gadolinium enhancement in magnetic resonance imaging. Patients with NVA-myocarditis (n=108) more often had symptoms of heart failure (P = .003), arrhythmias (P = .031), left ventricular dilatation (P = .045), lower LVEF (P < .001) and major cardiac adverse events (P = .102). CONCLUSIONS: Course of COVID-19 vaccine-associated myocarditis in pediatric patients seems to be mild and differs from non-vaccine-associated myocarditis. Due to a considerable number of residual symptoms and diagnostic abnormalities at follow-up, further studies are needed to define its long-term implications.

Development and validation of explainable machine learning models for risk of mortality in transcatheter aortic valve implantation: TAVI risk machine scores.

Aims: Identification of high-risk patients and individualized decision support based on objective criteria for rapid discharge after transcatheter aortic valve implantation (TAVI) are key requirements in the context of contemporary TAVI treatment. This study aimed to predict 30-day mortality following TAVI based on machine learning (ML) using data from the German Aortic Valve Registry. Methods and results: Mortality risk was determined using a random forest ML model that was condensed in the newly developed TAVI Risk Machine (TRIM) scores, designed to represent clinically meaningful risk modelling before (TRIMpre) and in particular after (TRIMpost) TAVI. Algorithm was trained and cross-validated on data of 22 283 patients (729 died within 30 days post-TAVI) and generalisation was examined on data of 5864 patients (146 died). TRIMpost demonstrated significantly better performance than traditional scores [C-statistics value, 0.79; 95% confidence interval (CI)] [0.74; 0.83] compared to Society of Thoracic Surgeons (STS) with C-statistics value 0.69; 95%-CI [0.65; 0.74]). An abridged (aTRIMpost) score comprising 25 features (calculated using a web interface) exhibited significantly higher performance than traditional scores (C-statistics value, 0.74; 95%-CI [0.70; 0.78]). Validation on external data of 6693 patients (205 died within 30 days post-TAVI) of the Swiss TAVI Registry confirmed significantly better performance for the TRIMpost (C-statistics value 0.75, 95%-CI [0.72; 0.79]) compared to STS (C-statistics value 0.67, CI [0.63; 0.70]). Conclusion: TRIM scores demonstrate good performance for risk estimation before and after TAVI. Together with clinical judgement, they may support standardised and objective decision-making before and after TAVI.

Effect of Minocycline on Depressive Symptoms in Patients With Treatment-Resistant Depression: A Randomized Clinical Trial.

Importance: Insufficient treatment response and resulting chronicity constitute a major problem in depressive disorders. Remission rates range as low as 15% to 40% and treatment-resistant depression (TRD) is associated with low-grade inflammation, suggesting anti-inflammatory interventions as a rational treatment strategy. Minocycline, which inhibits microglial activation, represents a promising repurposing candidate in the treatment of TRD. Objective: To determine whether 6 weeks of minocycline as add-on to antidepressant treatment as usual can significantly reduce depressive symptoms in patients with TRD. Design, Setting, and Participants: The study was conducted in Germany and designed as a multicenter double-blind randomized clinical trial (RCT) of 200 mg/d minocycline treatment over a course of 6 weeks with a 6-month follow-up. Participants were recruited from January 2016 to August 2020 at 9 university hospitals that served as study sites. Key inclusion criteria were a diagnosis of major depressive disorder (according to Diagnostic and Statistical Manual of Mental Disorders [Fifth Edition] criteria), severity of depressive symptoms on the Hamilton Depression Rating Scale (HAMD-17) greater than or equal to 16 points, aged 18 to 75 years, body mass index 18 to 40, Clinical Global Impression Scale (CGI-S) greater than or equal to 4, failure to adequately respond to an initial antidepressant standard medication as per Massachusetts General Hospital Antidepressant Treatment History Questionnaire, and stable medication for at least 2 weeks. A total of 258 patients were screened, of whom 173 were randomized and 168 were included into the intention-to-treat population. Statistical analysis was performed from April to November 2020. Interventions: Participants were randomized (1:1) to receive adjunct minocycline (200 mg/d) or placebo for 6 weeks. Main Outcomes and Measures: Primary outcome measure was the change in Montgomery-Åsberg Depression Rating Scale (MADRS) score from baseline to week 6 analyzed by intention-to-treat mixed model repeated measures. Secondary outcome measures were response, remission, and various other clinical rating scales. Results: Of 173 eligible and randomized participants (84 randomized to minocycline and 89 randomized to placebo), 168 formed the intention-to-treat sample (79 [47.0%] were women, 89 [53.0%] were men, 159 [94.6%] were White, 9 [6.4%] were of other race and ethnicity, including Asian and unknown ethnicity), with 81 in the minocycline group and 87 in the placebo group. The mean (SD) age was 46.1 (13.1) years, and the mean (SD) MADRS score at baseline was 26.5 (5.0). There was no difference in rates of completion between the minocycline (83.3% [70 of 81]) and the placebo group (83.1% [74 of 87]). Minocycline treatment did not alter the course of depression severity compared with placebo as assessed by a decrease in MADRS scores over 6 weeks of treatment (1.46 [-1.04 to 3.96], P = .25). Minocycline treatment also exhibited no statistically significant effect on secondary outcomes. Conclusions and Relevance: In this large randomized clinical trial with minocycline at a dose of 200 mg/d added to antidepressant treatment as usual for 6 weeks, minocycline was well tolerated but not superior to placebo in reducing depressive symptoms in patients with TRD. The results of this RCT emphasize the unmet need for therapeutic approaches and predictive biomarkers in TRD. Trial Registration: EU Clinical Trials Register Number: EudraCT 2015-001456-29.

Classification of Companion Diagnostics: A New Framework for Biomarker-Driven Patient Selection.

BACKGROUND: Modern personalized medicine strategies builds on therapy companion diagnostics to stratify patients into subgroups with differential benefit/risk. In general, stratification for drug response implies a treatment-by-subgroup interaction. This interaction is usually suggested by the drug's mechanism of action and investigated in pharmacological research or in clinical studies. In these candidate genes or pathway approaches, either biological reasons for a differential benefit/risk or statistical interaction regarding a pharmacological or clinical endpoint or both may be given. For successful drug approval, demonstration of a positive benefit/risk balance in the intended patient population is required. This also applies to situations with biomarker-selected populations. However, further regulatory considerations relate to the usefulness and plausibility of the selected patients and benefit/risk extrapolations or alternative therapy options in biomarker-negative populations. METHODS: To facilitate the specification of regulatory requirements and support the design of clinical development programmes, a systematic classification of biomarker-drug pairs is needed, in particular with regard to the expected underlying molecular mechanism and the clinical evidence. RESULTS: A classification of five biomarker-drug categories is proposed related to increasing evidence on the biomarker's predictive value in relation to a specific drug. We classified biomarkers into five ascending categories with increasing evidence on the predictive nature of the biomarker in relation to a specific drug according to the comparative pharmacological and clinical evidence. CONCLUSIONS: The proposed classification will facilitate regulatory decision-making and support drug development with respect to biomarker-related subgrouping, both, during clinical programme and at the time of marketing authorization application, since the grade of evidence on the differential power of the biomarker can be considered as an indicator for the usefulness of a biomarker-related subgrouping.

Applications of artificial intelligence/machine learning approaches in cardiovascular medicine: a systematic review with recommendations.

Aims: Artificial intelligence (AI) and machine learning (ML) promise vast advances in medicine. The current state of AI/ML applications in cardiovascular medicine is largely unknown. This systematic review aims to close this gap and provides recommendations for future applications. Methods and results: Pubmed and EMBASE were searched for applied publications using AI/ML approaches in cardiovascular medicine without limitations regarding study design or study population. The PRISMA statement was followed in this review. A total of 215 studies were identified and included in the final analysis. The majority (87%) of methods applied belong to the context of supervised learning. Within this group, tree-based methods were most commonly used, followed by network and regression analyses as well as boosting approaches. Concerning the areas of application, the most common disease context was coronary artery disease followed by heart failure and heart rhythm disorders. Often, different input types such as electronic health records and images were combined in one AI/ML application. Only a minority of publications investigated reproducibility and generalizability or provided a clinical trial registration. Conclusions: A major finding is that methodology may overlap even with similar data. Since we observed marked variation in quality, reporting of the evaluation and transparency of data and methods urgently need to be improved.

A comparison of subgroup identification methods in clinical drug development: Simulation study and regulatory considerations.

With advancement of technologies such as genomic sequencing, predictive biomarkers have become a useful tool for the development of personalized medicine. Predictive biomarkers can be used to select subsets of patients, which are most likely to benefit from a treatment. A number of approaches for subgroup identification were proposed over the last years. Although overviews of subgroup identification methods are available, systematic comparisons of their performance in simulation studies are rare. Interaction trees (IT), model-based recursive partitioning, subgroup identification based on differential effect, simultaneous threshold interaction modeling algorithm (STIMA), and adaptive refinement by directed peeling were proposed for subgroup identification. We compared these methods in a simulation study using a structured approach. In order to identify a target population for subsequent trials, a selection of the identified subgroups is needed. Therefore, we propose a subgroup criterion leading to a target subgroup consisting of the identified subgroups with an estimated treatment difference no less than a pre-specified threshold. In our simulation study, we evaluated these methods by considering measures for binary classification, like sensitivity and specificity. In settings with large effects or huge sample sizes, most methods perform well. For more realistic settings in drug development involving data from a single trial only, however, none of the methods seems suitable for selecting a target population. Using the subgroup criterion as alternative to the proposed pruning procedures, STIMA and IT can improve their performance in some settings. The methods and the subgroup criterion are illustrated by an application in amyotrophic lateral sclerosis.

Redirecting public oral health fluoride varnish intervention to low socio-economic status children in Alberta.

SETTING: Dental decay is most prevalent among low socio-economic status (SES) groups where cost limits access to dental care. To address inequities in oral health outcomes, Alberta Health Services (AHS) Oral Health Action Plan encompasses a population health approach that redirects fluoride varnish (FV) applications to low SES children. Using low SES measures to establish the eligibility criteria is fundamental to the delivery of FV applications to the target population. INTERVENTION: A series of four FV applications over two years is directed to children age 12-35 months and two applications per year to children in Kindergarten and grades 1 and 2, using low SES measures for eligibility criteria. The provincial objective for children receiving the first FV application is 10%-20% of the population age. Additional objectives are set for rates of subsequent FV applications for each population group. OUTCOMES: From 2015 to 2016, the rate of first FV applications for eligible target populations is below the provincial objective for children age 12-35 months (5%) and within the objective for children in Kindergarten and grades 1 and 2 (16%). Rates of subsequent FV applications in the school setting are being met. IMPLICATIONS: Encompassing a population health approach to deliver standardized fluoride varnish applications to low SES children better targets inequities in oral health outcomes in Alberta. Challenges of redirecting the FV intervention include creating the eligibility criteria and engaging the target population, particularly for the preschool population. Achieving population objectives are challenged by unequal distribution of resources across the province.