Modernizing adverse events analysis in oncology clinical trials using alternative approaches: rationale and design of the MOTIVATE trial

An Overview of Regression Models for Adverse Events Analysis

Over the last few years, several review articles described the adverse events analysis as sub-optimal in clinical trials. Indeed, the context surrounding adverse events analyses often imply an overwhelming number of events, a lack of power to find associations, but also a lack of specific training regarding those complex data. In randomized controlled trials or in observational studies, comparing the occurrence of adverse events according to a covariable of interest (e.g., treatment) is a recurrent question in the analysis of drug safety data, and adjusting other important factors is often relevant. This article is an overview of the existing regression models that may be considered to compare adverse events and to discuss model choice regarding the characteristics of the adverse events of interest. Many dimensions may be relevant to compare the adverse events between patients, (e.g., timing, recurrence, and severity). Recent efforts have been made to cover all of them. For chronic treatments, the occurrence of intercurrent events during the patient follow-up usually needs the modeling approach to be adapted (at least with regard to their interpretation). Moreover, analysis based on regression models should not be limited to the estimation of relative effects. Indeed, absolute risks stemming from the model should be presented systematically to help the interpretation, to validate the model, and to encourage comparison of studies.

Early Adverse Event Derived Biomarkers in Predicting Clinical Outcomes in Patients with Advanced Non-Small Cell Lung Cancer Treated with Immunotherapy

RATIONALE

Adverse events (AEs) have been shown to have clinical associations, in addition to patient safety assessments of drugs of interest. However, due to their complex content and associated data structure, AE evaluation has been restricted to descriptive statistics and small AE subset for efficacy analysis, limiting the opportunity for global discovery. This study takes a unique approach to utilize AE-associated parameters to derive a set of innovative AE metrics. Comprehensive analysis of the AE-derived biomarkers enhances the chance of discovering new predictive AE biomarkers of clinical outcomes.

METHODS

We utilized a set of AE-associated parameters (grade, treatment relatedness, occurrence, frequency, and duration) to derive 24 AE biomarkers. We further innovatively defined early AE biomarkers by landmark analysis at an early time point to assess the predictive value. Statistical methods included the Cox proportional hazards model for progression-free survival (PFS) and overall survival (OS), two-sample t-test for mean difference of AE frequency and duration between disease control (DC: complete response (CR) + partial response (PR) + stable disease (SD)) versus progressive disease (PD), and Pearson correlation analysis for relationship of AE frequency and duration versus treatment duration. Two study cohorts (Cohort A: vorinostat + pembrolizumab, and B: Taminadenant) from two immunotherapy trials in late-stage non-small cell lung cancer were used to test the potential predictiveness of AE-derived biomarkers. Data from over 800 AEs were collected per standard operating procedure in a clinical trial using the Common Terminology Criteria for Adverse Events v5 (CTCAE). Clinical outcomes for statistical analysis included PFS, OS, and DC.

RESULTS

An early AE was defined as event occurrence at or prior to day 30 from initial treatment date. The early AEs were then used to calculate the 24 early AE biomarkers to assess overall AE, each toxicity category, and each individual AE. These early AE-derived biomarkers were evaluated for global discovery of clinical association. Both cohorts showed that early AE biomarkers were associated with clinical outcomes. Patients previously experienced with low-grade AEs (including treatment related AEs (TrAE)) had improved PFS, OS, and were associated with DC. The significant early AEs included low-grade TrAE in overall AE, endocrine disorders, hypothyroidism (pembrolizumab's immune-related adverse event (irAE)), and platelet count decreased (vorinostat related TrAE) for Cohort A and low-grade AE in overall AE, gastrointestinal disorders, and nausea for Cohort B. In contrast, patients with early development of high-grade AEs tended to have poorer PFS, OS, and correlated with PD. The associated early AEs included high-grade TrAE in overall AE, gastrointestinal disorders with two members, diarrhea and vomiting, for Cohort A and high-grade AE in overall AE, three toxicity categories, and five related individual AEs for Cohort B. One low-grade TrAE, alanine aminotransferase increased (vorinostat + pembrolizumab related), was an irAE and correlated with worse OS in Cohort A.

CONCLUSIONS

The study demonstrated the potential clinical utility of early AE-derived biomarkers in predicting positive and negative clinical outcomes. It could be TrAEs or combination of TrAEs and nonTrAEs from overall AEs, toxicity category AEs, to individual AEs with low-grade event leaning to encouraging effect and high-grade event to undesirable impact. Moreover, the methodology of the AE-derived biomarkers could change current AE analysis practice from a descriptive summary into modern informative statistics. It modernizes AE data analysis by helping clinicians discover novel AE biomarkers to predict clinical outcomes and facilitate the generation of vast clinically meaningful research hypotheses in a new AE content to fulfill the demands of precision medicine.

Comparison of statistical methods for the analysis of recurrent adverse events in the presence of non-proportional hazards and unobserved heterogeneity: a simulation study

Background

In preventive drug trials such as intermittent preventive treatment for malaria prevention during pregnancy (IPTp), where there is repeated treatment administration, recurrence of adverse events (AEs) is expected. Challenges in modelling the risk of the AEs include accounting for time-to-AE and within-patient-correlation, beyond the conventional methods. The correlation comes from two sources; (a) individual patient unobserved heterogeneity (i.e. frailty) and (b) the dependence between AEs characterised by time-dependent treatment effects. Potential AE-dependence can be modelled via time-dependent treatment effects, event-specific baseline and event-specific random effect, while heterogeneity can be modelled via subject-specific random effect. Methods that can improve the estimation of both the unobserved heterogeneity and treatment effects can be useful in understanding the evolution of risk of AEs, especially in preventive trials where time-dependent treatment effect is expected.

Methods

Using both a simulation study and the Chloroquine for Malaria in Pregnancy (NCT01443130) trial data to demonstrate the application of the models, we investigated whether the lognormal shared frailty models with restricted cubic splines and non-proportional hazards (LSF-NPH) assumption can improve estimates for both frailty variance and treatment effect compared to the conventional inverse Gaussian shared frailty model with proportional hazard (ISF-PH), in the presence of time-dependent treatment effects and unobserved patient heterogeneity. We assessed the bias, precision gain and coverage probability of 95% confidence interval of the frailty variance estimates for the models under varying known unobserved heterogeneity, sample sizes and time-dependent effects.

Results

The ISF-PH model provided a better coverage probability of 95% confidence interval, less bias and less precise frailty variance estimates compared to the LSF-NPH models. The LSF-NPH models yielded unbiased hazard ratio estimates at the expense of imprecision and high mean square error compared to the ISF-PH model.

Conclusion

The choice of the shared frailty model for the recurrent AEs analysis should be driven by the study objective. Using the LSF-NPH models is appropriate if unbiased hazard ratio estimation is of primary interest in the presence of time-dependent treatment effects. However, ISF-PH model is appropriate if unbiased frailty variance estimation is of primary interest.

Trial registration

ClinicalTrials.gov; NCT01443130

Quality-adjusted Time Without Symptoms of disease or Toxicity (Q-TWiST) analysis of CPX-351 versus 7 + 3 in older adults with newly diagnosed high-risk/secondary AML

BACKGROUND

CPX-351 (United States: Vyxeos^®; Europe: Vyxeos^® Liposomal), a dual-drug liposomal encapsulation of daunorubicin and cytarabine in a synergistic 1:5 molar ratio, is approved by the US FDA and the EMA for the treatment of adults with newly diagnosed therapy-related acute myeloid leukemia or acute myeloid leukemia with myelodysplasia-related changes. In a pivotal phase 3 study that evaluated 309 patients aged 60 to 75 years with newly diagnosed high-risk/secondary acute myeloid leukemia, CPX-351 significantly improved median overall survival versus conventional 7 + 3 chemotherapy (cytarabine continuous infusion for 7 days plus daunorubicin for 3 days), with a comparable safety profile. A Quality-adjusted Time Without Symptoms of disease or Toxicity (Q-TWiST) analysis of the phase 3 study was performed to compare survival quality between patients receiving CPX-351 versus conventional 7 + 3 after 5 years of follow-up.

METHODS

Patients were randomized 1:1 between December 20, 2012 and November 11, 2014 to receive induction with CPX-351 or 7 + 3. Survival time for each patient was partitioned into 3 health states: TOX (time with any grade 3 or 4 toxicity or prior to remission), TWiST (time in remission without relapse or grade 3 or 4 toxicity), and REL (time after relapse). Within each treatment arm, Q-TWiST was calculated by adding the mean time spent in each health state weighted by its respective quality-of-life, represented by health utility. The relative Q-TWiST gain, calculated as the difference in Q-TWiST between treatment arms divided by the mean survival of the 7 + 3 control arm, was determined in order to evaluate results in the context of other Q-TWiST analyses.

RESULTS

The relative Q-TWiST gain with CPX-351 versus 7 + 3 was 53.6% in the base case scenario and 39.8% among responding patients. Across various sensitivity analyses, the relative Q-TWiST gains for CPX-351 ranged from 48.0 to 57.6%, remaining well above the standard clinically important difference threshold of 15% for oncology.

CONCLUSIONS

This post hoc analysis demonstrates that CPX-351 improved quality-adjusted survival, further supporting the clinical benefit in patients with newly diagnosed high-risk/secondary acute myeloid leukemia. Trial registration This trial was registered on September 28, 2012 at www.clinicaltrials.gov as NCT01696084 ( https://clinicaltrials.gov/ct2/show/NCT01696084 ) and is complete.