Differential Treatment Effects of Subgroup Analyses in Phase 3 Oncology Trials From 2004 to 2020

Importance

Subgroup analyses are often performed in oncology to investigate differential treatment effects and may even constitute the basis for regulatory approvals. Current understanding of the features, results, and quality of subgroup analyses is limited.

Objective

To evaluate forest plot interpretability and credibility of differential treatment effect claims among oncology trials.

Design, Setting, and Participants

This cross-sectional study included randomized phase 3 clinical oncology trials published prior to 2021. Trials were screened from ClinicalTrials.gov.

Main Outcomes and Measures

Missing visual elements in forest plots were defined as a missing point estimate or use of a linear x-axis scale for hazard and odds ratios. Multiplicity of testing control was recorded. Differential treatment effect claims were rated using the Instrument for Assessing the Credibility of Effect Modification Analyses. Linear and logistic regressions evaluated associations with outcomes.

Results

Among 785 trials, 379 studies (48%) enrolling 331 653 patients reported a subgroup analysis. The forest plots of 43% of trials (156 of 363) were missing visual elements impeding interpretability. While 4148 subgroup effects were evaluated, only 1 trial (0.3%) controlled for multiple testing. On average, trials that did not meet the primary end point conducted 2 more subgroup effect tests compared with trials meeting the primary end point (95% CI, 0.59-3.43 tests; P = .006). A total of 101 differential treatment effects were claimed across 15% of trials (55 of 379). Interaction testing was missing in 53% of trials (29 of 55) claiming differential treatment effects. Trials not meeting the primary end point were associated with greater odds of no interaction testing (odds ratio, 4.47; 95% CI, 1.42-15.55, P = .01). The credibility of differential treatment effect claims was rated as low or very low in 93% of cases (94 of 101).

Conclusions and Relevance

In this cross-sectional study of phase 3 oncology trials, nearly half of trials presented a subgroup analysis in their primary publication. However, forest plots of these subgroup analyses largely lacked essential features for interpretation, and most differential treatment effect claims were not supported. Oncology subgroup analyses should be interpreted with caution, and improvements to the quality of subgroup analyses are needed.

A statistical framework for powerful multi-trait rare variant analysis in large-scale whole-genome sequencing studies

Large-scale whole-genome sequencing (WGS) studies have improved our understanding of the contributions of coding and noncoding rare variants to complex human traits. Leveraging association effect sizes across multiple traits in WGS rare variant association analysis can improve statistical power over single-trait analysis, and also detect pleiotropic genes and regions. Existing multi-trait methods have limited ability to perform rare variant analysis of large-scale WGS data. We propose MultiSTAAR, a statistical framework and computationally-scalable analytical pipeline for functionally-informed multi-trait rare variant analysis in large-scale WGS studies. MultiSTAAR accounts for relatedness, population structure and correlation among phenotypes by jointly analyzing multiple traits, and further empowers rare variant association analysis by incorporating multiple functional annotations. We applied MultiSTAAR to jointly analyze three lipid traits (low-density lipoprotein cholesterol, high-density lipoprotein cholesterol and triglycerides) in 61,861 multi-ethnic samples from the Trans-Omics for Precision Medicine (TOPMed) Program. We discovered new associations with lipid traits missed by single-trait analysis, including rare variants within an enhancer of NIPSNAP3A and an intergenic region on chromosome 1.

Prevalence and implications of significance testing for baseline covariate imbalance in randomised cancer clinical trials: The Table 1 Fallacy

BACKGROUND

The 'Table 1 Fallacy' refers to the unsound use of significance testing for comparing the distributions of baseline variables between randomised groups to draw erroneous conclusions about balance or imbalance. We performed a cross-sectional study of the Table 1 Fallacy in phase III oncology trials.

METHODS

From ClinicalTrials.gov, 1877 randomised trials were screened. Multivariable logistic regressions evaluated predictors of the Table 1 Fallacy.

RESULTS

A total of 765 randomised controlled trials involving 553,405 patients were analysed. The Table 1 Fallacy was observed in 25% of trials (188 of 765), with 3% of comparisons deemed significant (59 of 2353), approximating the typical 5% type I error assertion probability. Application of trial-level multiplicity corrections reduced the rate of significant findings to 0.3% (six of 2345 tests). Factors associated with lower odds of the Table 1 Fallacy included industry sponsorship (adjusted odds ratio [aOR] 0.29, 95% confidence interval [CI] 0.18-0.47; multiplicity-corrected P < 0.0001), larger trial size (≥795 versus <280 patients; aOR 0.32, 95% CI 0.19-0.53; multiplicity-corrected P = 0.0008), and publication in a European versus American journal (aOR 0.06, 95% CI 0.03-0.13; multiplicity-corrected P < 0.0001).

CONCLUSIONS

This study highlights the persistence of the Table 1 Fallacy in contemporary oncology randomised controlled trials, with one of every four trials testing for baseline differences after randomisation. Significance testing is a suboptimal method for identifying unsound randomisation procedures and may encourage misleading inferences. Journal-level enforcement is a possible strategy to help mitigate this fallacy.

Unsupervised representation learning improves genomic discovery and risk prediction for respiratory and circulatory functions and diseases

High-dimensional clinical data are becoming more accessible in biobank-scale datasets. However, effectively utilizing high-dimensional clinical data for genetic discovery remains challenging. Here we introduce a general deep learning-based framework, REpresentation learning for Genetic discovery on Low-dimensional Embeddings (REGLE), for discovering associations between genetic variants and high-dimensional clinical data. REGLE uses convolutional variational autoencoders to compute a non-linear, low-dimensional, disentangled embedding of the data with highly heritable individual components. REGLE can incorporate expert-defined or clinical features and provides a framework to create accurate disease-specific polygenic risk scores (PRS) in datasets which have minimal expert phenotyping. We apply REGLE to both respiratory and circulatory systems: spirograms which measure lung function and photoplethysmograms (PPG) which measure blood volume changes. Genome-wide association studies on REGLE embeddings identify more genome-wide significant loci than existing methods and replicate known loci for both spirograms and PPG, demonstrating the generality of the framework. Furthermore, these embeddings are associated with overall survival. Finally, we construct a set of PRSs that improve predictive performance of asthma, chronic obstructive pulmonary disease, hypertension, and systolic blood pressure in multiple biobanks. Thus, REGLE embeddings can quantify clinically relevant features that are not currently captured in a standardized or automated way.

An allelic-series rare-variant association test for candidate-gene discovery

Allelic series are of candidate therapeutic interest because of the existence of a dose-response relationship between the functionality of a gene and the degree or severity of a phenotype. We define an allelic series as a collection of variants in which increasingly deleterious mutations lead to increasingly large phenotypic effects, and we have developed a gene-based rare-variant association test specifically targeted to identifying genes containing allelic series. Building on the well-known burden test and sequence kernel association test (SKAT), we specify a variety of association models covering different genetic architectures and integrate these into a Coding-Variant Allelic-Series Test (COAST). Through extensive simulations, we confirm that COAST maintains the type I error and improves the power when the pattern of coding-variant effect sizes increases monotonically with mutational severity. We applied COAST to identify allelic-series genes for four circulating-lipid traits and five cell-count traits among 145,735 subjects with available whole-exome sequencing data from the UK Biobank. Compared with optimal SKAT (SKAT-O), COAST identified 29% more Bonferroni-significant associations with circulating-lipid traits, on average, and 82% more with cell-count traits. All of the gene-trait associations identified by COAST have corroborating evidence either from rare-variant associations in the full cohort (Genebass, n = 400,000) or from common-variant associations in the GWAS Catalog. In addition to detecting many gene-trait associations present in Genebass by using only a fraction (36.9%) of the sample, COAST detects associations, such as that between ANGPTL4 and triglycerides, that are absent from Genebass but that have clear common-variant support.

Leveraging a surrogate outcome to improve inference on a partially missing target outcome

Sample sizes vary substantially across tissues in the Genotype-Tissue Expression (GTEx) project, where considerably fewer samples are available from certain inaccessible tissues, such as the substantia nigra (SSN), than from accessible tissues, such as blood. This severely limits power for identifying tissue-specific expression quantitative trait loci (eQTL) in undersampled tissues. Here we propose Surrogate Phenotype Regression Analysis (Spray) for leveraging information from a correlated surrogate outcome (eg, expression in blood) to improve inference on a partially missing target outcome (eg, expression in SSN). Rather than regarding the surrogate outcome as a proxy for the target outcome, Spray jointly models the target and surrogate outcomes within a bivariate regression framework. Unobserved values of either outcome are treated as missing data. We describe and implement an expectation conditional maximization algorithm for performing estimation in the presence of bilateral outcome missingness. Spray estimates the same association parameter estimated by standard eQTL mapping and controls the type I error even when the target and surrogate outcomes are truly uncorrelated. We demonstrate analytically and empirically, using simulations and GTEx data, that in comparison with marginally modeling the target outcome, jointly modeling the target and surrogate outcomes increases estimation precision and improves power.

Using a Clinically Interpretable End Point Composed of Multiple Outcomes to Evaluate Totality of Treatment Effect in Comparative Oncology Studies

This cohort study demonstrates how to use cumulative event count curves to create a clinically meaningful end point by simultaneously considering recurrence, progression, and survival times from the individual patient.

Effect of Dapagliflozin on Total Heart Failure Events in Patients With Heart Failure With Mildly Reduced or Preserved Ejection Fraction: A Prespecified Analysis of the DELIVER Trial

Importance

In the Dapagliflozin Evaluation to Improve the Lives of Patients With Preserved Ejection Fraction Heart Failure (DELIVER) trial, dapagliflozin reduced the risk of time to first worsening heart failure (HF) event or cardiovascular death in patients with HF with mildly reduced or preserved ejection fraction (EF).

Objective

To evaluate the effect of dapagliflozin on total (ie, first and recurrent) HF events and cardiovascular death in this population.

Design, Setting, and Participants

In this prespecified analysis of the DELIVER trial, the proportional rates approach of Lin, Wei, Yang, and Ying (LWYY) and a joint frailty model were used to examine the effect of dapagliflozin on total HF events and cardiovascular death. Several subgroups were examined to test for heterogeneity in the effect of dapagliflozin, including left ventricular EF. Participants were enrolled from August 2018 to December 2020, and data were analyzed from August to October 2022.

Interventions

Dapagliflozin, 10 mg, once daily or matching placebo.

Main Outcomes and Measures

The outcome was total episodes of worsening HF (hospitalization for HF or urgent HF visit requiring intravenous HF therapies) and cardiovascular death.

Results

Of 6263 included patients, 2747 (43.9%) were women, and the mean (SD) age was 71.7 (9.6) years. There were 1057 HF events and cardiovascular deaths in the placebo group compared with 815 in the dapagliflozin group. Patients with more HF events had features of more severe HF, such as higher N-terminal pro-B-type natriuretic peptide level, worse kidney function, more prior HF hospitalizations, and longer duration of HF, although EF was similar to those with no HF events. In the LWYY model, the rate ratio for total HF events and cardiovascular death for dapagliflozin compared with placebo was 0.77 (95% CI, 0.67-0.89; P < .001) compared with a hazard ratio of 0.82 (95% CI, 0.73-0.92; P < .001) in a traditional time to first event analysis. In the joint frailty model, the rate ratio was 0.72 (95% CI, 0.65-0.81; P < .001) for total HF events and 0.87 (95% CI, 0.72-1.05; P = .14) for cardiovascular death. The results were similar for total HF hospitalizations (without urgent HF visits) and cardiovascular death and in all subgroups, including those defined by EF.

Conclusions and Relevance

In the DELIVER trial, dapagliflozin reduced the rate of total HF events (first and subsequent HF hospitalizations and urgent HF visits) and cardiovascular death regardless of patient characteristics, including EF.

Trial Registration

ClinicalTrials.gov Identifier: NCT03619213.

Inference of chronic obstructive pulmonary disease with deep learning on raw spirograms identifies new genetic loci and improves risk models

Chronic obstructive pulmonary disease (COPD), the third leading cause of death worldwide, is highly heritable. While COPD is clinically defined by applying thresholds to summary measures of lung function, a quantitative liability score has more power to identify genetic signals. Here we train a deep convolutional neural network on noisy self-reported and International Classification of Diseases labels to predict COPD case-control status from high-dimensional raw spirograms and use the model's predictions as a liability score. The machine-learning-based (ML-based) liability score accurately discriminates COPD cases and controls, and predicts COPD-related hospitalization without any domain-specific knowledge. Moreover, the ML-based liability score is associated with overall survival and exacerbation events. A genome-wide association study on the ML-based liability score replicates existing COPD and lung function loci and also identifies 67 new loci. Lastly, our method provides a general framework to use ML methods and medical-record-based labels that does not require domain knowledge or expert curation to improve disease prediction and genomic discovery for drug design.

Abstract 5375: Learned phenotypic embeddings enable scalable imputation of high-content molecular data elucidating prognostic chromatin signatures

Emerging high-content data modalities like functional genomics and spatial proteomics have enormous potential to reveal determinants of phenotypic plasticity that underlie variability in clinical outcomes, but to date these modalities are only collected in modestly sized research cohorts (&lt; 200-400 patients), where we lack power to detect subtype-specific or prognostic signatures.

To study intertumor heterogeneity on a much larger scale (&gt;10,000 patients), we developed a machine learning framework based on self-supervised embeddings that allows scalable imputation of high-content data on large standard of care datasets. Our framework starts by learning a phenotypic embedding of tumor state based solely on H&E histology images, allowing the embedding to be trained on large patient cohorts regardless of availability of molecular covariates. It then learns to predict genomic or proteomic labels from the lower-dimensional phenotypic embeddings. This model can be used for imputation in much larger cohorts, where only clinical outcome and histology are available.

To demonstrate our method, we use the TCGA ATAC-seq data, which is available for 400 patients across 23 cancer types. By learning self-supervised embeddings of histology, our framework was able to impute ATAC-seq for 5,000 peaks in 11,000 patients across 31 cancer types, with high accuracy in held out samples (R2 = 0.61). To our knowledge, this represents the broadest available pan-cancer chromatin landscape.

The imputed ATAC-seq reveals a subset of peaks that are significantly associated with overall survival (OS) in multiple cancer types (e.g., Breast (BC) ATAC-only HR 1.75, p: 8.6E-3). Genes proximal to these peaks are strongly enriched for well characterized oncogenes, and also several novel genes with functions in cellular metabolism and chromatin remodeling whose expression is not known to be prognostic in our disease settings.

Finally, we developed models to predict OS from H&E slide embeddings and from imputed ATAC-seq, both pan-cancer and in specific tumor types. Both models significantly outperform baseline stage and molecular subtype clinical risk predictors (e.g., BC baseline HR: 2.44, p: 3E-6 vs. embedding/imputed ATAC HR 3.78, 2E-9, p for improvement 2E-9) and, interestingly, we find that adding an ATAC-seq based risk score to an embedding-based risk score significantly improves disease-specific survival prediction (HCC embedding-only HR: 2.13, p: 8E-5 vs. HCC embedding/imputed ATAC HR: 2.65, p: 1E-6). This suggests that histopathology images are a rich source of prognostic information beyond that which is captured by traditional pathologist grading.

Overall, our work highlights the ability to use self-supervised embeddings of histopathology to impute biological covariates on large, standard-of-care cohorts, empowering novel insights into disease mechanisms and patient outcome.

Citation Format: Christopher S. Probert, Zachary R. McCaw, Navami Jain, Daphne Koller. Learned phenotypic embeddings enable scalable imputation of high-content molecular data elucidating prognostic chromatin signatures. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 5375.

Assessment of Median and Mean Survival Time in Cancer Clinical Trials

This cohort study assesses the relative stability of median and mean survival time estimates reported in cancer clinical trials.

Quantifying Treatment Effects in Trials with Multiple Event-Time Outcomes

BACKGROUND

Data on the occurrence times of multiple outcomes, reflecting the temporal profile of disease burden/progression, have been used to estimate treatment effects in various recent randomized trials. Most procedures for analyzing these data require specific model assumptions. When the assumptions are not met, the results may be misleading. Robust, model-free procedures for study design and analysis that enable clinically meaningful interpretations are warranted.

METHODS

For each treatment group, we constructed and summarized the estimated mean cumulative count of events over time by the area under the curve (AUC), which can be interpreted as the mean total event-free time lost from multiple undesirable outcomes. A higher curve, and resulting larger AUC, implies a worse treatment. The treatment effect is quantified by the ratio and/or difference of AUCs. The timing and occurrence of recurrent heart failure hospitalizations (HFHs) and cardiovascular (CV) death from Prospective Comparison of ARNI with ARB Global Outcomes in HF with Preserved Ejection Fraction (PARAGON-HF), comparing sacubitril/valsartan with valsartan, are presented for illustration. We also discuss the design of future studies on the basis of the proposed method.

RESULTS

With 48 months of follow-up, estimated AUCs, representing the total event-free time lost to HFHs and CV death, were 11.3 and 13.1 event-months for sacubitril/valsartan and valsartan, respectively. The ratio of these AUCs was 0.86 (95% confidence interval, 0.75 to 1.00; P=0.049), a 14% reduction of disease burden favoring combination therapy. A future study, similar to PARAGON-HF, designed using the new proposal would require fewer patients would than a conventional time-to-first-event analysis.

CONCLUSIONS

The proposed method is robust and model-free and provides a clinically interpretable, time-scale summary of the treatment effect. (Funded by National Institutes of Health.).

Fitting Gaussian mixture models on incomplete data

Background

Bioinformatics investigators often gain insights by combining information across multiple and disparate data sets. Merging data from multiple sources frequently results in data sets that are incomplete or contain missing values. Although missing data are ubiquitous, existing implementations of Gaussian mixture models (GMMs) either cannot accommodate missing data, or do so by imposing simplifying assumptions that limit the applicability of the model. In the presence of missing data, a standard ad hoc practice is to perform complete case analysis or imputation prior to model fitting. Both approaches have serious drawbacks, potentially resulting in biased and unstable parameter estimates.

Results

Here we present missingness-aware Gaussian mixture models (MGMM), an R package for fitting GMMs in the presence of missing data. Unlike existing GMM implementations that can accommodate missing data, MGMM places no restrictions on the form of the covariance matrix. Using three case studies on real and simulated ’omics data sets, we demonstrate that, when the underlying data distribution is near-to a GMM, MGMM is more effective at recovering the true cluster assignments than either the existing GMM implementations that accommodate missing data, or fitting a standard GMM after state of the art imputation. Moreover, MGMM provides an accurate assessment of cluster assignment uncertainty, even when the generative distribution is not a GMM.

Conclusion

Compared to state-of-the-art competitors, MGMM demonstrates a better ability to recover the true cluster assignments for a wide variety of data sets and a large range of missingness rates. MGMM provides the bioinformatics community with a powerful, easy-to-use, and statistically sound tool for performing clustering and density estimation in the presence of missing data. MGMM is publicly available as an R package on CRAN: https://CRAN.R-project.org/package=MGMM.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12859-022-04740-9.

OUP accepted manuscript

When designing a comparative oncology trial for an overall or progression-free survival endpoint, investigators often quantify the treatment effect using a difference in median survival times. However, rather than directly designing the study to estimate this difference, it is almost always converted to a hazard ratio (HR) to determine the study size. At the analysis stage, the hazard ratio is utilized for formal analysis, yet because it may be difficult to interpret clinically, especially when the proportional hazards assumption is not met, the observed medians are also reported descriptively. The hazard ratio and median difference contrast different aspects of the survival curves. Whereas the hazard ratio places greater emphasis on late-occurring separation, the median difference focuses locally on the centers of the distributions and cannot capture either short- or long-term differences. Having 2 sets of summaries (a hazard ratio and the medians) may lead to incoherent conclusions regarding the treatment effect. For instance, the hazard ratio may suggest a treatment difference whereas the medians do not, or vice versa. In this commentary, we illustrate these commonly encountered issues using examples from recent oncology trials. We present a coherent alternative strategy that, unlike relying on the hazard ratio, does not require modeling assumptions and always results in clinically interpretable summaries of the treatment effect.

Practical Recommendations on Quantifying and Interpreting Treatment Effects in the Presence of Terminal Competing Risks: A Review

Importance

In a comparative trial, the time to a clinical event is often a key end point. However, the occurrence of a terminal event, such as death or premature study discontinuation, may preclude observation of this outcome. Although various methods for handling competing risks are available, no specific recommendations have been made for scenarios encountered in practice, especially when the terminal event profiles of the study arms are dissimilar. Moreover, appropriate methods for a desirable outcome, such as live hospital discharge, have seldom been discussed.

Observations

Several of the most commonly used methods are reviewed. The first regards the terminal event as censoring and applies standard survival analysis to the event of interest. The between-group difference is usually summarized by the cause-specific hazard ratio. This summary measure is inappropriate when the new therapy markedly prolongs time to the terminal event. Moreover, the corresponding Kaplan-Meier curve for the end point of interest is uninterpretable. The second method is to use the cumulative incidence curve, which is the probability of experiencing the event of interest by each time point, acknowledging that patients who have died will never experience the event. However, the resulting pseudo hazard ratio is difficult to interpret. With a proper alternative summary measure, this approach works well for a desirable outcome but may not for an undesirable outcome. The third method focuses on the event-free survival time by combining information from occurrences of the terminal event and the event of interest simultaneously. This clinically interpretable method naturally accounts for differences in terminal event rates when comparing treatments with respect to the time to an undesirable outcome.

Conclusions and Relevance

This article enhances our understanding of each method's advantages and shortcomings and assists practitioners in choosing appropriate methods for handling competing risk problems in practice.

Multitrait GWAS to connect disease variants and biological mechanisms

Genome-wide association studies (GWASs) have uncovered a wealth of associations between common variants and human phenotypes. Here, we present an integrative analysis of GWAS summary statistics from 36 phenotypes to decipher multitrait genetic architecture and its link with biological mechanisms. Our framework incorporates multitrait association mapping along with an investigation of the breakdown of genetic associations into clusters of variants harboring similar multitrait association profiles. Focusing on two subsets of immunity and metabolism phenotypes, we then demonstrate how genetic variants within clusters can be mapped to biological pathways and disease mechanisms. Finally, for the metabolism set, we investigate the link between gene cluster assignment and the success of drug targets in randomized controlled trials.

Large-scale machine-learning-based phenotyping significantly improves genomic discovery for optic nerve head morphology

Genome-wide association studies (GWASs) require accurate cohort phenotyping, but expert labeling can be costly, time intensive, and variable. Here, we develop a machine learning (ML) model to predict glaucomatous optic nerve head features from color fundus photographs. We used the model to predict vertical cup-to-disc ratio (VCDR), a diagnostic parameter and cardinal endophenotype for glaucoma, in 65,680 Europeans in the UK Biobank (UKB). A GWAS of ML-based VCDR identified 299 independent genome-wide significant (GWS; p ≤ 5 × 10-8) hits in 156 loci. The ML-based GWAS replicated 62 of 65 GWS loci from a recent VCDR GWAS in the UKB for which two ophthalmologists manually labeled images for 67,040 Europeans. The ML-based GWAS also identified 93 novel loci, significantly expanding our understanding of the genetic etiologies of glaucoma and VCDR. Pathway analyses support the biological significance of the novel hits to VCDR: select loci near genes involved in neuronal and synaptic biology or harboring variants are known to cause severe Mendelian ophthalmic disease. Finally, the ML-based GWAS results significantly improve polygenic prediction of VCDR and primary open-angle glaucoma in the independent EPIC-Norfolk cohort.

Survival Analysis of Treatment Efficacy in Comparative Coronavirus Disease 2019 Studies

For survival analysis in comparative coronavirus disease 2019 trials, the routinely used hazard ratio may not provide a meaningful summary of the treatment effect. The mean survival time difference/ratio is an intuitive, assumption-free alternative. However, for short-term studies, landmark mortality rate differences/ratios are more clinically relevant and should be formally analyzed and reported.

How to Quantify and Interpret Treatment Effects in Comparative Clinical Studies of COVID-19

Clinical trials of treatments for coronavirus disease 2019 (COVID-19) draw intense public attention. More than ever, valid, transparent, and intuitive summaries of the treatment effects, including efficacy and harm, are needed. In recently published and ongoing randomized comparative trials evaluating treatments for COVID-19, time to a positive outcome, such as recovery or improvement, has repeatedly been used as either the primary or key secondary end point. Because patients may die before recovery or improvement, data analysis of this end point faces a competing risk problem. Commonly used survival analysis techniques, such as the Kaplan-Meier method, often are not appropriate for such situations. Moreover, almost all trials have quantified treatment effects by using the hazard ratio, which is difficult to interpret for a positive event, especially in the presence of competing risks. Using 2 recent trials evaluating treatments (remdesivir and convalescent plasma) for COVID-19 as examples, a valid, well-established yet underused procedure is presented for estimating the cumulative recovery or improvement rate curve across the study period. Furthermore, an intuitive and clinically interpretable summary of treatment efficacy based on this curve is also proposed. Clinical investigators are encouraged to consider applying these methods for quantifying treatment effects in future studies of COVID-19.

Selecting appropriate endpoints for assessing treatment effects in comparative clinical studies for COVID-19

To evaluate the efficacy and safety of a new treatment for COVID-19 vs. standard care, certain key endpoints are related to the duration of a specific event, such as hospitalization, ICU stay, or receipt of supplemental oxygen. However, since patients may die in the hospital during study follow-up, using, for example, the duration of hospitalization to assess treatment efficacy can be misleading. If the treatment tends to prolong patients' survival compared with standard care, patients in the new treatment group may spend more time in hospital. This can lead to a "survival bias" issue, where a treatment that is effective for preventing death appears to prolong an undesirable outcome. On the other hand, by using hospital-free survival time as the endpoint, we can circumvent the survival bias issue. In this article, we use reconstructed data from a recent, large clinical trial for COVID-19 to illustrate the advantages of this approach. For the analysis of ICU stay or oxygen usage, where the initiating event is potentially an outcome of treatment, standard survival analysis techniques may not be appropriate. We also discuss issues with analyzing the durations of such events.

Analysis of Response Data for Assessing Treatment Effects in Comparative Clinical Studies

In comparative studies, treatment effect is often assessed using a binary outcome that indicates response to the therapy. Commonly used summary measures for response include the cumulative and current response rates at a specific time point. The current response rate is sometimes called the probability of being in response (PBIR), which regards a patient as a responder only if they have achieved and remain in response at present. The methods used in practice for estimating these rates, however, may not be appropriate. Moreover, whereas an effective treatment is expected to achieve a rapid and sustained response, the response at a fixed time point does not provide information about the duration of response (DOR). As an alternative, a curve constructed from the current response rates over the entire study period may be considered, which can be used for visualizing how rapidly patients responded to therapy and how long responses were sustained. The area under the PBIR curve is the mean DOR. This connection between response and DOR makes this curve attractive for assessing the treatment effect. In contrast to the conventional method for analyzing the DOR data, which uses responders only, the above procedure includes all patients in the study. Although discussed extensively in the statistical literature, estimation of the current response rate curve has garnered little attention in the medical literature. This article illustrates how to construct and analyze such a curve using data from a recent study for treating renal cell carcinoma. Clinical trialists are encouraged to consider this robust and clinically interpretable procedure as an additional tool for evaluating treatment effects in clinical studies.

Operating characteristics of the rank-based inverse normal transformation for quantitative trait analysis in genome-wide association studies

Quantitative traits analyzed in Genome-Wide Association Studies (GWAS) are often nonnormally distributed. For such traits, association tests based on standard linear regression are subject to reduced power and inflated type I error in finite samples. Applying the rank-based inverse normal transformation (INT) to nonnormally distributed traits has become common practice in GWAS. However, the different variations on INT-based association testing have not been formally defined, and guidance is lacking on when to use which approach. In this paper, we formally define and systematically compare the direct (D-INT) and indirect (I-INT) INT-based association tests. We discuss their assumptions, underlying generative models, and connections. We demonstrate that the relative powers of D-INT and I-INT depend on the underlying data generating process. Since neither approach is uniformly most powerful, we combine them into an adaptive omnibus test (O-INT). O-INT is robust to model misspecification, protects the type I error, and is well powered against a wide range of nonnormally distributed traits. Extensive simulations were conducted to examine the finite sample operating characteristics of these tests. Our results demonstrate that, for nonnormally distributed traits, INT-based tests outperform the standard untransformed association test, both in terms of power and type I error rate control. We apply the proposed methods to GWAS of spirometry traits in the UK Biobank. O-INT has been implemented in the R package RNOmni, which is available on CRAN.

Toll-like receptor 4-mediated respiratory syncytial virus disease and lung transcriptomics in differentially susceptible inbred mouse strains

Respiratory syncytial virus (RSV) causes severe lower respiratory tract disease in infants, young children, and susceptible adults. The pathogenesis of RSV disease is not fully understood, although toll-like receptor 4 (TLR4)-related innate immune response is known to play a role. The present study was designed to determine TLR4-mediated disease phenotypes and lung transcriptomics and to elucidate transcriptional mechanisms underlying differential RSV susceptibility in inbred strains of mice. Dominant negative Tlr4 mutant (C3H/HeJ, HeJ, Tlr4^Lps-d) and its wild-type (C3H/HeOuJ, OuJ, Tlr4^Lps-n) mice and five genetically diverse, differentially responsive strains bearing the wild-type Tlr4^Lps-n allele were infected with RSV. Bronchoalveolar lavage, histopathology, and genome-wide transcriptomics were used to characterize the pulmonary response to RSV. RSV-induced lung neutrophilia [1 day postinfection (pi)], epithelial proliferation (1 day pi), and lymphocytic infiltration (5 days pi) were significantly lower in HeJ compared with OuJ mice. Pulmonary RSV expression was also significantly suppressed in HeJ than in OuJ. Upregulation of immune/inflammatory (Cxcl3, Saa1) and heat shock protein (Hspa1a, Hsph1) genes was characteristic of OuJ mice, while cell cycle and cell death/survival genes were modulated in HeJ mice following RSV infection. Strain-specific transcriptomics suggested virus-responsive (Oasl1, Irg1, Mx1) and epidermal differentiation complex (Krt4, Lce3a) genes may contribute to TLR4-independent defense against RSV in resistant strains including C57BL/6J. The data indicate that TLR4 contributes to pulmonary RSV pathogenesis and activation of cellular immunity, the inflammasome complex, and vascular damage underlies it. Distinct transcriptomics in differentially responsive Tlr4-wild-type strains provide new insights into the mechanism of RSV disease and potential therapeutic targets.