Robustness of an odds-ratio test in a stratified group sequential trial with a binary outcome measure

General Anesthetics in CAncer REsection Surgery (GA-CARES) randomized multicenter trial of propofol vs volatile inhalational anesthesia: protocol description

BACKGROUND

Studies indicate that patients can be "seeded" with their own cancer cells during oncologic surgery and that the immune response to these circulating cancer cells might influence the risk of cancer recurrence. Preliminary data from animal studies and some retrospective analyses suggest that anesthetic technique might affect the immune response during surgery and hence the risk of cancer recurrence. In 2015, experts called for prospective scientific inquiry into whether anesthetic technique used in cancer resection surgeries affects cancer-related outcomes such as recurrence and mortality. Therefore, we designed a pragmatic phase 3 multicenter randomized controlled trial (RCT) called General Anesthetics in Cancer Resection (GA-CARES).

METHODS

After clinical trial registration and institutional review board approval, patients providing written informed consent were enrolled at five sites in New York (NY) State. Eligible patients were adults with known or suspected cancer undergoing one of eight oncologic surgeries having a high risk of cancer recurrence. Exclusion criteria included known or suspected history of malignant hyperthermia or hypersensitivity to either propofol or volatile anesthetic agents. Patients were randomized (1:1) stratified by center and surgery type using REDCap to receive either propofol or volatile agent for maintenance of general anesthesia (GA). This pragmatic trial, which seeks to assess the potential impact of anesthetic type in "real world practice", did not standardize any aspect of patient care. However, potential confounders, e.g., use of neuroaxial anesthesia, were recorded to confirm the balance between study arms. Assuming a 5% absolute difference in 2-year overall survival rates (85% vs 90%) between study arms (primary endpoint, minimum 2-year follow-up), power using a two-sided log-rank test with type I error of 0.05 (no planned interim analyses) was calculated to be 97.4% based on a target enrollment of 1800 subjects. Data sources include the National Death Index (gold standard for vital status in the USA), NY Cancer Registry, and electronic harvesting of data from electronic medical records (EMR), with minimal manual data abstraction/data entry.

DISCUSSION

Enrollment has been completed (n = 1804) and the study is in the follow-up phase. This unfunded, pragmatic trial, uses a novel approach for data collection focusing on electronic sources.

TRIAL REGISTRATION

Registered (NCT03034096) on January 27, 2017, prior to consent of the first patient on January 31, 2017.

Statistical design of Phase II/III clinical trials for testing therapeutic interventions in COVID-19 patients

BACKGROUND

Because of unknown features of the COVID-19 and the complexity of the population affected, standard clinical trial designs on treatments may not be optimal in such patients. We propose two independent clinical trials designs based on careful grouping of patient and outcome measures.

METHODS

Using the World Health Organization ordinal scale on patient status, we classify treatable patients (Stages 3-7) into two risk groups. Patients in Stages 3, 4 and 5 are categorized as the intermediate-risk group, while patients in Stages 6 and 7 are categorized as the high-risk group. To ensure that an intervention, if deemed efficacious, is promptly made available to vulnerable patients, we propose a group sequential design incorporating four factors stratification, two interim analyses, and a toxicity monitoring rule for the intermediate-risk group. The primary response variable (binary variable) is based on the proportion of patients discharged from hospital by the 15^th day. The goal is to detect a significant improvement in this response rate. For the high-risk group, we propose a group sequential design incorporating three factors stratification, and two interim analyses, with no toxicity monitoring. The primary response variable for this design is 30 day mortality, with the goal of detecting a meaningful reduction in mortality rate.

RESULTS

Required sample size and toxicity boundaries are calculated for each scenario. Sample size requirements for designs with interim analyses are marginally greater than ones without. In addition, for both the intermediate-risk group and the high-risk group, the required sample size with two interim analyses is almost identical to analyses with just one interim analysis.

CONCLUSIONS

We recommend using a binary outcome with composite endpoints for patients in Stage 3, 4 or 5 with a power of 90% to detect an improvement of 20% in the response rate, and a 30 day mortality rate outcome for those in Stage 6 or 7 with a power of 90% to detect 15% (effect size) reduction in mortality rate. For the intermediate-risk group, two interim analyses for efficacy evaluation along with toxicity monitoring are encouraged. For the high-risk group, two interim analyses without toxicity monitoring is advised.

Statistical design of Phase II/III clinical trials for testing therapeutic interventions in COVID-19 patients

BACKGROUND

Because of unknown features of the COVID-19 and the complexity of the population affected, standard clinical trial designs on treatments may not be optimal in such patients. We propose two independent clinical trials designs based on careful grouping of patient and outcome measures.

METHODS

Using the World Health Organization ordinal scale on patient status, we classify treatable patients (Stages 3-7) into two risk groups. Patients in Stages 3, 4 and 5 are categorized as the intermediate-risk group, while patients in Stages 6 and 7 are categorized as the high-risk group. To ensure that an intervention, if deemed efficacious, is promptly made available to vulnerable patients, we propose a group sequential design incorporating four factors stratification, two interim analyses, and a toxicity monitoring rule for the intermediate-risk group. The primary response variable (binary variable) is based on the proportion of patients discharged from hospital by the 15th day. The goal is to detect a significant improvement in this response rate. For the high-risk group, we propose a group sequential design incorporating three factors stratification, and two interim analyses, with no toxicity monitoring. The primary response variable for this design is 30 day mortality, with the goal of detecting a meaningful reduction in mortality rate.

RESULTS

Required sample size and toxicity boundaries are calculated for each scenario. Sample size requirements for designs with interim analyses are marginally greater than ones without. In addition, for both the intermediate-risk group and the high-risk group, the required sample size with two interim analyses is almost identical to analyses with just one interim analysis.

CONCLUSIONS

We recommend using a binary outcome with composite endpoints for patients in Stage 3, 4 or 5 with a power of 90% to detect an improvement of 20% in the response rate, and a 30 day mortality rate outcome for those in Stage 6 or 7 with a power of 90% to detect 15% (effect size) reduction in mortality rate. For the intermediate-risk group, two interim analyses for efficacy evaluation along with toxicity monitoring are encouraged. For the high-risk group, two interim analyses without toxicity monitoring is advised.

Treatment of Middle East respiratory syndrome with a combination of lopinavir/ritonavir and interferon-β1b (MIRACLE trial): statistical analysis plan for a recursive two-stage group sequential randomized controlled trial

The MIRACLE trial (MERS-CoV Infection tReated with A Combination of Lopinavir/ritonavir and intErferon-β1b) investigates the efficacy of a combination therapy of lopinavir/ritonavir and recombinant interferon-β1b provided with standard supportive care, compared to placebo provided with standard supportive care, in hospitalized patients with laboratory-confirmed MERS. The MIRACLE trial is designed as a recursive, two-stage, group sequential, multicenter, placebo-controlled, double-blind randomized controlled trial. The aim of this article is to describe the statistical analysis plan for the MIRACLE trial. The primary outcome is 90-day mortality. The primary analysis will follow the intention-to-treat principle. The MIRACLE trial is the first randomized controlled trial for MERS treatment. TRIAL REGISTRATION: ClinicalTrials.gov, NCT02845843. Registered on 27 July 2016.