Oncology Phase Research Articles

Disease progression, survival, and molecular disparities among Black and White patients with endometrioid endometrial carcinoma in real-world registries and legacy NRG oncology phase III trials

Disease progression, survival, and molecular disparities among Black and White patients with endometrioid endometrial carcinoma in real-world registries and legacy NRG oncology phase III trials

Prostate Cancer Risk Stratification in NRG Oncology Phase III Randomized Trials Using Multimodal Deep Learning With Digital Histopathology.

Current clinical risk stratification methods for localized prostate cancer are suboptimal, leading to over- and undertreatment. Recently, machine learning approaches using digital histopathology have shown superior prognostic ability in phase III trials. This study aims to develop a clinically usable risk grouping system using multimodal artificial intelligence (MMAI) models that outperform current National Comprehensive Cancer Network (NCCN) risk groups. The cohort comprised 9,787 patients with localized prostate cancer from eight NRG Oncology randomized phase III trials, treated with radiation therapy, androgen deprivation therapy, and/or chemotherapy. Locked MMAI models, which used digital histopathology images and clinical data, were applied to each patient. Expert consensus on cut points defined low-, intermediate-, and high-risk groups on the basis of 10-year distant metastasis rates of 3% and 10%, respectively. The MMAI's reclassification and prognostic performance were compared with the three-tier NCCN risk groups. The median follow-up for censored patients was 7.9 years. According to NCCN risk categories, 30.4% of patients were low-risk, 25.5% intermediate-risk, and 44.1% high-risk. The MMAI risk classification identified 43.5% of patients as low-risk, 34.6% as intermediate-risk, and 21.8% as high-risk. MMAI reclassified 1,039 (42.0%) patients initially categorized by NCCN. Despite the MMAI low-risk group being larger than the NCCN low-risk group, the 10-year metastasis risks were comparable: 1.7% (95% CI, 0.2 to 3.2) for NCCN and 3.2% (95% CI, 1.7 to 4.7) for MMAI. The overall 10-year metastasis risk for NCCN high-risk patients was 16.6%, with MMAI further stratifying this group into low-, intermediate-, and high-risk, showing metastasis rates of 3.4%, 8.2%, and 26.3%, respectively. The MMAI risk grouping system expands the population of men identified as having low metastatic risk and accurately pinpoints a high-risk subset with elevated metastasis rates. This approach aims to prevent both overtreatment and undertreatment in localized prostate cancer, facilitating shared decision making.

A single-centre study evaluating a geriatric screening tool in oncology phase I trial patients.

Though cancer is more prevalent in the older population, this patient group are underrepresented in phase I oncology trials. We evaluated the use of a geriatric screening tool (SAOP3) in patients of 70 years of age or older who attended a Phase I Clinical Trials Unit, with the aim of assessing the feasibility of the tool and identifying potential unmet needs in this patient group. Twenty-two patients over the age of 70 completed the SAOP3 questionnaire. Geriatric impairments and needs were analysed with descriptive statistics. Qualitative responses were grouped in themes using structured thematic analysis. All of patients triggered at least 1 geriatric domain, most commonly mobility. Six core themes were identified as being important to the patient including family, friends and positivity. On cognition assessment over 20% of patients triggered as requiring further cognitive assessment. The group had a relatively high screen fail risk. In conclusion, routine geriatric screening withSAOP3 was feasible and identified areas of patient need. Results highlight the prevalence of psychological distress and cognitive impairment. Geriatric screening offers an opportunity for prehabilitation prior to trial and support during trial participation to optimise safety and improve trial access.

Statistical fragility of findings from randomized phase 3 trials in pediatric oncology.

e23003 Background: The fragility index (FI) and survival-inferred fragility index (SIFI) have been proposed as adjunctive metrics to facilitate interpretation of p-values in clinical trials. FI/SIFI is the number of patients on a positive (p < 0.05) trial’s experimental arm who would need to switch from “responder/non-event” to “non-responder/event” to cause p > 0.05. Fragility indices have been reported previously in medical oncology, with median values < 10 often seen in the literature. This metric has not been studied in pediatric oncology trials. Methods: PubMed was used to identify randomized phase 3 pediatric oncology trials published between 1980-2020. We report trial characteristics and calculate FI for trials with a binary outcome and SIFI for trials with a time-to-event outcome. We also report the fragility quotients (FQ and SFQ) to normalize FI and SIFI relative to trial size. Results: 113 trials included sufficient data for analysis. The median FI for trials with a binary outcome (n = 40) was 4.5 (range: 1-33). The median SIFI for trials with a time-to-event outcome (n = 73) was 13 (range: 0-61). The FI or SIFI was less than the number of patients lost to follow-up in 25% of 36 trials. Median FQ = 0.026 and SFQ = 0.03, indicating that revised outcome in 2.6% or 3% of trial participants would be sufficient to alter the trial conclusion. FQ and SFQ did not significantly vary according to trial characteristics. While sample sizes increased over time (mean 187, 1980s; mean 437, 2010s), FQ and SFQ remained stable between 2-3%. Conclusions: The statistical conclusions of pediatric oncology phase 3 trials hinge on a relatively small number and proportion of patients. Despite sample size limitations of low prevalence diseases, pediatric cancer trials appear similarly or less fragile than adult oncology trials. Fragility was similar for large versus small trials and remained constant over four decades. [Table: see text]

On the relative conservativeness of Bayesian logistic regression method in oncology dose-finding studies.

The Bayesian logistic regression method (BLRM) is a widely adopted and flexible design for finding the maximum tolerated dose in oncology phase I studies. However, the BLRM design has been criticized in the literature for being overly conservative due to the use of the overdose control rule. Recently, a discussion paper titled "Improving the performance of Bayesian logistic regression model with overall control in oncology dose-finding studies" in Statistics in Medicine has proposed an overall control rule to address the "excessive conservativeness" of the standard BLRM design. In this short communication, we discuss the relative conservativeness of the standard BLRM design and also suggest a dose-switching rule to further enhance its performance.

Differences in Absolute Neutrophil Count Eligibility Criteria of Pediatric Phase I and Phase I/II Oncology Clinical Trials According to Sponsorship

Introduction: Racial clinical trial enrollment disparities in oncology are well documented ( Pediatr Blood Cancer. 2020 10.1002/pbc.28296). Barriers to enrollment are multifactorial and include structural and clinical barriers like restrictive eligibility criteria ( Am Soc Clin Oncol Educ Book. 2016 10.1200/EDBK_156686). Many phase I/II oncology clinical trials have laboratory criteria such as a minimum threshold for absolute neutrophil count (ANC). The Common Terminology Criteria for Adverse Events (CTCAE) defines an ANC of less than 1500 cells/µL as the upper limit of a grade 2 toxicity ( J Clin Oncol. 2010 10.1200/JCO.2009.24.3881), and this has become the minimum eligibility threshold in many oncology trials ( J Clin Oncol. 2017 10.1200/JCO.2017.73.4186). Using a minimum ANC of 1500 cells/µL may contribute to further racial enrollment disparities due to the Duffy null phenotype, which has a prevalence of ~65% in Black/African American individuals ( PLoS Genet. 2011 10.1371/journal.pgen.1002108) and results in a lower number of circulating neutrophils compared to people without the Duffy null phenotype ( Blood. 2021 10.1182/blood.2020008600). Up to 23% of pediatric patients with the Duffy null phenotype have ANCs <1500 cells/µL ( Blood Adv. 2023 10.1182/bloodadvances.2022007680). A recent adult multiple myeloma study found that the most common reason for ineligibility among Black patients was failure to meet hematologic laboratory criteria ( Blood. 2023 10.1182/blood.2022018657). Less is known about eligibility criteria as a barrier to enrollment in the pediatric cancer population. We hypothesized industry-sponsored trials would have more restrictive eligibility criteria than non-industry-sponsored trials. To explore this further we describe ANC eligibility criteria of pediatric oncology phase I/II clinical trials according to sponsorship. Methods: We analyzed ANC eligibility criteria of interventional phase I and phase I/II oncology trials opened in the United States from January 2010 to May 2023 using the ClinicalTrials.gov registry. Trials that included pediatric participants only (up to age 18), pediatric and adolescent young adult participants (up to age 39), and adult studies that allowed pediatric participants were included. We excluded trials for non-malignant conditions and trials that did not report ANC eligibility criteria. ANC eligibility criteria were categorized as ≥1500 cells/µL and <1500 cells/µL. Trials were compared according to primary sponsorship (industry vs non-industry). Secondary objective was to describe language requirements explicitly mentioned for these trials. Distribution of trial characteristics were compared using Fisher's exact or Chi-squared tests. Logistic regression was used to evaluate potential changes in ANC criteria over time (in years), using ANC<1500 cells/µL as the reference group. Results: We identified 861 interventional trials that met our initial inclusion criteria. After review of the trials, 36 were excluded for enrolling participants with non-malignant diagnoses. Of the remaining 825 studies, 438 (53%) had ANC eligibility criteria reported. 89 (20%) studies were industry-sponsored (Table 1). Trial phase, age eligibility, diagnosis, and results reported differed statistically significantly by sponsor type. 76 (17%) studies required ANC ≥1500 cells/µL for enrollment. For industry sponsored trials, 26 (29%) required ANC ≥1500, while 50 (14%) non-industry sponsored trials required ANC ≥1500, with a risk ratio of 2.04 (95% CI 1.35-3.08, p=0.002) (Figure 1). Almost all studies (96%) did not explicitly state language requirements, and this did not vary by sponsor type. There was no significant change in odds of a trial requiring ANC ≥1500 cells/µL over time (odds ratio 1.05, 95% CI 0.98-1.12, p=0.138). Conclusion: A minority of trials (17%) required an ANC ≥1500 cells/µL for enrollment on a phase I or phase I/II interventional clinical oncology trial; however, industry-sponsored trials were twice as likely than non-industry trials to require this higher ANC threshold. Further studies are needed to investigate laboratory criteria as a possible reason for racial enrollment disparities in pediatric cancer trials.

Drug Repurposing for Cancers With Limited Survival: Protocol for a Retrospective Cohort Study.

Only 5% of the molecules tested in oncology phase 1 trials reach the market after an average of 7.5 years of waiting and at a cost of tens of millions of dollars. To reduce the cost and shorten the time of discovery of new treatments, "drug repurposing" (research with molecules already approved for another indication) and the use of secondary data (not collected for the purpose of research) have been proposed. Due to advances in informatics in clinical care, secondary data can, in some cases, be of equal quality to primary data generated through prospective studies. The objective of this study is to identify drugs currently marketed for other indications that may have an effect on the prognosis of patients with cancer. We plan to monitor a cohort of patients with high-lethality cancers treated in the public health system of Catalonia between 2006 and 2012, retrospectively, for survival for 5 years after diagnosis or until death. A control cohort, comprising people without cancer, will also be retrospectively monitored for 5 years. The following study variables will be extracted from different population databases: type of cancer (patients with cancer cohort), date and cause of death, pharmacological treatment, sex, age, and place of residence. During the first stage of statistical analysis of the patients with cancer cohort, the drugs consumed by the long-term survivors (alive at 5 years) will be compared with those consumed by nonsurvivors. In the second stage, the survival associated with the consumption of each relevant drug will be analyzed. For the analyses, groups will be matched for potentially confounding variables, and multivariate analyses will be performed to adjust for residual confounding variables if necessary. The control cohort will be used to verify whether the associations found are exclusive to patients with cancer or whether they also occur in patients without cancer. We anticipate discovering multiple significant associations between commonly used drugs and the survival outcomes of patients with cancer. We expect to publish the initial results in the first half of 2024. This retrospective study may identify several commonly used drugs as candidates for repurposing in the treatment of various cancers. All analyses are considered exploratory; therefore, the results will have to be confirmed in subsequent clinical trials. However, the results of this study may accelerate drug discovery in oncology. DERR1-10.2196/48925.

Validation of a Genomic Classifier in the NRG Oncology/RTOG 0521 Phase III Trial of Docetaxel with Androgen Suppression and Radiotherapy for Localized High-Risk Prostate Cancer

Decipher is a prognostic genomic classifier (GC) validated in several prospective NRG Oncology Phase III trials. Herein, we validate the GC in pre-treatment biopsy samples for risk stratification in a cohort of high-risk men treated with definitive radiotherapy and androgen suppression with or without docetaxel chemotherapy. As per a pre-specified and approved NCI analysis plan (Navigator #1061), we obtained available formalin-fixed paraffin-embedded tissue from biopsy specimens from the NRG biobank from patients enrolled on the NRG/RTOG 0521 randomized phase III trial. After central review, the highest-grade tumors were profiled on clinical-grade whole-transcriptome arrays (Veracyte, San Diego, CA) and GC scores were obtained. Pre-specified categorical GC scores, adjusted for archival tissue analysis, were used to define higher (>0.46) and lower (≤0.46) risk groups. The primary objective was to validate the independent prognostic ability of GC for metastasis-free survival (MFS) with Cox multivariable analyses (MVA). Samples were obtained from 283 consented, evaluable patients with tissue (50% of trial) yielding 183 (65%) GC scores that passed quality metrics, 91 from control and 92 from the interventional arm. Median age was 66 years, median PSA was 19.3 ng/uL (IQR: 8.1-41.4), 81% had clinical stage ≥T2 and 80% had Gleason score ≥8 (47% ≥9). Median GC score was 0.55 (IQR: 0.38-0.78) and overall the arms were balanced for key covariates. With a median follow-up of 9.9 years (IQR: 9.3, 10.7), 67 MFS events including 34 distant metastases (DM) were observed. On MVA, only the GC (per 0.1 unit) was independently associated with MFS (HR 1.12, 95% CI 1.01-1.25) as well as DM (sHR 1.22, 95% CI 1.06-1.41), whereas the 4 pre-defined trial risk groups used for stratification (based on Gleason score, T-stage and PSA), randomization and patient age were not. For categorical GC, on MVA, higher-risk GC patients (65%) had worse DM (sHR 2.82, 95% CI 1.1-7.3) compared to those with lower GC. Cumulative incidence of DM at 10-years was 27% for higher GC vs 9% (95% CI 7-18%) for lower GC. No biomarker-by-treatment interaction with GC and the addition of docetaxel was detected. In pre-treatment biopsy samples from a randomized Phase 3 trial cohort, GC demonstrated its ability to further risk stratify clinically high-risk men demonstrating an independent association of GC score with DM and MFS. High-risk prostate cancer is a heterogeneous disease state and GC can improve risk stratification to help personalize shared decision-making. NRG-GU009/PREDICT-RT (NCT04513717) aims to determine the optimal therapy based on GC score for high-risk prostate cancer.

Reporting and impact of subsequent cycle toxicities in oncology phase I clinical trials.

3149 Background: In recent years, the landscape of clinical trial development has evolved to include immunotherapy and targeted therapies, which have different lengths of administration and toxicity profiles compared to standard chemotherapy. As treatments have evolved, the classic assumptions of toxicity associated with cytotoxic agents may be less relevant, requiring new strategies for design of trials intended to inform dosing strategies for agents that may be administered beyond a set number of defined cycles. We aimed to describe the overall incidence of dose-limiting toxicities (DLTs) during and after cycle 1, the frequency of reporting subsequent cycle toxicities, and the impact of post-cycle 1 DLTs on conclusions drawn from oncology phase I clinical trials. Methods: We conducted a systematic review of subsequent cycle toxicities in oncology phase I clinical trials published in the Journal of Clinical Oncology from 2000 to 2020. Only trials that included more than one cycle of therapy were included. Data elements captured included: year of publication, study design, trial sponsor type, intervention type, number of anti-cancer therapies involved, primary diagnosis, incidence and type of cycle 1 DLT, incidence of dose modification in cycle 1, incidence and type of post-cycle 1 DLTs, incidence of dose modification in post-cycle 1, presence of a recommended phase 2 dose (RP2D) and impact of subsequent cycle toxicity on RP2D. Results: From 2000 to 2020, we identified 489 full-text articles reporting on therapeutic phase I clinical trials. Of these, 421 (86%) reported data regarding cycle 1 DLTs and 170 (35%) reported cycle 1 dose modifications. The overall median cycle 1 DLT rate was 8.89%. Only 47 (9.6%) publications reported whether or not there were post-cycle 1 DLTs and only 92 (19%) reported whether or not there were dose modifications beyond cycle 1. The overall median post-cycle 1 DLT rate among trials that reported post-cycle 1 data was 14.8%. A RP2D was reported in 76% (371 of 489) trials. Among these 371 studies, 89% did not report whether post-cycle 1 toxicities impacted the RP2D. Independent predictors of reporting toxicity data beyond cycle 1 included year of publication, investigational agent type, novelty, disease type, trial size, and sponsor. Conclusions: Reporting of subsequent cycle DLTs and dose modification is uncommon in oncology phase I clinical trial publications. As newer oncology treatments trend towards longer duration of treatment, there will be an increased need to understand toxicities occurring post-cycle 1. Guidelines for reporting of phase I clinical trials should expand to include toxicity data beyond the first cycle.

Incidence of Primary End Point Changes Among Active Cancer Phase 3 Randomized Clinical Trials

Primary end point (PEP) changes to an active clinical trial raise questions regarding trial quality and the risk of outcome reporting bias. It is unknown how the frequency and transparency of the reported changes depend on reporting method and whether the PEP changes are associated with trial positivity (ie, the trial met the prespecified statistical threshold for PEP positivity). To assess the frequency of reported PEP changes in oncology randomized clinical trials (RCTs) and whether these changes are associated with trial positivity. This cross-sectional study used publicly available data for complete oncology phase 3 RCTs registered in ClinicalTrials.gov from inception through February 2020. The main outcome was change between the initial PEP and the final reported PEP, assessed using 3 methods: (1) history of tracked changes on ClinicalTrials.gov, (2) self-reported changes noted in the article, and (3) changes reported within the protocol, including all available protocol documents. Logistic regression analyses were performed to evaluate whether PEP changes were associated with US Food and Drug Administration approval or trial positivity. Of 755 included trials, 145 (19.2%) had PEP changes found by at least 1 of the 3 detection methods. Of the 145 trials with PEP changes, 102 (70.3%) did not have PEP changes disclosed within the manuscript. There was significant variability in rates of PEP detection by each method (χ2 = 72.1; P < .001). Across all methods, PEP changes were detected at higher rates when multiple versions of the protocol (47 of 148 [31.8%]) were available compared with 1 version (22 of 134 [16.4%]) or no protocol (76 of 473 [16.1%]) (χ2 = 18.7; P < .001). Multivariable analysis demonstrated that PEP changes were associated with trial positivity (odds ratio, 1.86; 95% CI, 1.25-2.82; P = .003). This cross-sectional study revealed substantial rates of PEP changes among active RCTs; PEP changes were markedly underreported in published articles and mostly occurred after reported study completion dates. Significant discrepancies in the rate of detected PEP changes call into question the role of increased protocol transparency and completeness in identifying key changes occurring in active trials.

Outcomes of Older Adults With AML Treated in Community Versus Academic Centers: An Analysis of Alliance Trials.

Clinical trials are important for managing older patients with AML. We investigated differences in outcomes of older patients with AML on the basis of whether patients participated in intensive chemotherapy trials at community versus academic cancer centers. We used data from the Alliance for Clinical Trials in Oncology phase III trials that enrolled patients age ≥ 60 years with newly diagnosed AML between 1998 and 2002 in the Cancer and Leukemia Group B (CALGB) 9720 trial and between 2004 and 2006 in the CALGB 10201 trial. Centers funded by the NCI Community Oncology Research Program were identified as community cancer centers; others were designated as academic cancer centers. Logistic regression models and Cox proportional hazards models were used to compare 1-month mortality and overall survival (OS) by center type. Seventeen percent of the 1,170 patients were enrolled in clinical trials in community cancer centers. The study results demonstrated comparable rates of grade ≥3 adverse events (97% v 93%), 1-month mortality (19.1% v 16.1%), and OS (43.9% v 35.7% at 1 year) between community versus academic cancer centers, respectively. After adjusting for covariates, 1-month mortality (odds ratio, 1.40; 95% CI, 0.92 to 2.12; P = .11) and OS (hazard ratio, 1.04; 95% CI, 0.88 to 1.22; P = .67) were not statistically different among patients treated in community versus academic cancer centers. An older patient population, who have complex health care needs, can be successfully treated on intensive chemotherapy trials in select community cancer centers with outcomes comparable with that achieved at academic cancer centers.

EPID-07. A SYSTEMIC REVIEW OF PHASE II SINGLE-ARM TWO-STAGE DESIGNS IN GLIOBLASTOMA

Abstract Even though two-stage designs have been popularly in oncology Phase 2 single-arm trials during last decade, none, to our knowledge, has examined the usage for published articles that had applied the two-stage designs in Phase 2 single-arm trials in glioblastoma. It might be valuable to perform a comprehensive review on the phase 2 single-arm two-stage trials in glioblastoma for examination of appropriate application on the two-stage designs. Systematic review has been performed under the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA)-statement. The search was performed using following key words for the literature search as index terms or free-text words: “phase II trials”, “glioblastoma”, and “two-stage design”. After identifying 81 articles based on online search, we had 32 studies included in review due to exclusion of 49 articles such as duplicates (n = 10), randomized trials (n = 7), single-stage designs (n = 5), and insufficient information (n = 27). We extracted general and two-stage design information (tumor type and setting, drugs, error types, response rates, primary endpoint, two-stage sample size data, study termination status etc). Almost 50% of single-arm two-stage studies failed to provide key information related to study designs and sample size calculation, even wrong power analysis and interpretation. Response rate and progression-free survival are used as popular surrogate endpoints for overall survival. It is time to evaluate other potential endpoints like post-progress survival and hazard ratio, and usefulness of co-primary endpoint. Studies less than 10% provided the historical data source. Investigators should keep transparency of the trials by fully providing and citing historical data source and do conservative choice for historical rates. Meta-analysis can be a good option to accurately estimate the historical control. The appropriate selection on primary endpoint, historical control data, incorporation of adaptive designs, right application of two-stage designs might help overcoming the challenges on glioblastoma therapeutic trials.

Evidence synthesis analysis with prioritized benefit outcomes in oncology clinical trials

ABSTRACT Overall survival, progression-free survival, objective response/complete response, and duration of (complete) response are frequently used as the primary and secondary efficacy endpoints for designs and analyses of oncology clinical trials. However, these endpoints are typically analyzed separately. In this article, we introduce an evidence synthesis approach to prioritize the benefit outcomes by applying the generalized pairwise comparisons (GPC) method, and use win statistics (win ratio, win odds and net benefit) to quantify treatment benefit. Under the framework of GPC, the main advantage of this evidence synthesis approach is the ability to combine relevant outcomes of various types into a single summary statistic without relying on any parametric assumptions. It is particularly relevant since health authorities and the pharmaceutical industry are increasingly incorporating structured quantitative methodologies in their benefit-risk assessment. We apply this evidence synthesis approach to an oncology phase 3 study in first-line renal cell carcinoma to assess the overall effect of an investigational treatment by ranking the most clinically relevant endpoints in cancer drug development. This application and a simulation study demonstrate that the proposed approach can synthesize the evidence of treatment effect from multiple prioritized benefit outcomes, and has substantial advantage over conventional methods that analyze each individual endpoint separately. We also introduce a newly developed R package WINS for statistical inference based on win statistics.

Designing phase I oncology dose escalation using dose-exposure-toxicity models as a complementary approach to model-based dose-toxicity models.

One of the objectives of oncology phase I dose‐escalation studies has been to determine the maximum tolerated dose (MTD). Although MTD is no longer set as the dose for further development in contemporary oncology drug development, MTD determination is still important for informing the therapeutic index. Bayesian adaptive model‐based designs are becoming mainstream in oncology first‐in‐human trials. Herein, we illustrate via simulations the use of systemic exposure in Bayesian adaptive dose–toxicity models to estimate MTD. We extend traditional dose–toxicity models to incorporate pharmacokinetic exposure, which provides information on exposure–toxicity relationships. We pursue dose escalation until the maximum tolerated exposure (corresponding to the MTD) is reached. By leveraging pharmacokinetics, dose escalation considers exposure and interindividual variability on a continuous rather than discrete domain, offering additional information for dose‐escalation decisions. To demonstrate this, we generated 1000 simulations (starting dose of 1/25th the reference dose and six dose levels) for several different scenarios. Both rule‐based and model‐based designs were compared using metrics of potential safety, accuracy, and reliability. The mean results over simulations and different toxicity scenarios showed that model‐based designs were better than rule‐based methods and that exposure–toxicity model‐based methods have the potential to valuably complement dose–toxicity model‐based methods. Exposure–toxicity model‐based methods had decreased underdose risk accompanied by a relatively smaller increase in overdose risk, resulting in improved net reliability. MTD estimation accuracy was compromised when exposure variability was large, emphasizing the importance of appropriate control of pharmacokinetic variability in phase I dose‐escalation studies.

Inefficiency of two-stage designs in phase II oncology clinical trials with high proportion of inevaluable patients

BackgroundTwo-stage designs are commonly used for oncology Phase II clinical trials with a binary response endpoint. An issue that has not gained sufficient attention is the potential inefficiency in the usage of two-stage designs due to multiple enrollment suspensions when the proportion of patients inevaluable for response is high. MethodsSimulation studies were used to assess the performance of Simon's two-stage designs, two-stage designs with a proposed modification, and a single-stage design in the context of Phase II clinical trials with a high proportion of patients inevaluable for response. ResultsTwo-stage designs can require multiple enrollment disruptions when the inevaluable proportion is high, which can result in unacceptable inefficiency. The proposed modification provides a practical solution to this issue by enrolling an extra number of patients towards the end of the 1st stage, anticipating that a proportion of the patients pending response evaluation could be inevaluable. Single-stage designs with interim monitoring of futility that require no interim accrual suspension can be more efficient than two-stage designs, especially when the accrual and inevaluable rates are high. ConclusionsPlanning of Phase II trials should consider the issue of inefficiency of the two-stage designs, especially for trials with a high inevaluable proportion. Designs with monitoring rules that do not require accrual suspensions may be given more considerations, especially in trials of agents that have already had some evidence for safety and efficacy in other populations.

A straightforward meta‐analysis approach for oncology phase I dose‐finding studies

Phase I early-phase clinical studies aim at investigating the safety and the underlying dose-toxicity relationship of a drug or combination. While little may still be known about the compound's properties, it is crucial to consider quantitative information available from any studies that may have been conducted previously on the same drug. A meta-analytic approach has the advantages of being able to properly account for between-study heterogeneity, and it may be readily extended to prediction or shrinkage applications. Here we propose a simple and robust two-stage approach for the estimation of maximum tolerated dose(s) utilizing penalized logistic regression and Bayesian random-effects meta-analysis methodology. Implementation is facilitated using standard R packages. The properties of the proposed methods are investigated in Monte Carlo simulations. The investigations are motivated and illustrated by two examples from oncology.

Prostate cancer risk in African American men evaluated via digital histopathology multi-modal deep learning models developed on NRG Oncology phase III clinical trials.

108 Background: Artificial intelligence (AI) tools can display racial bias as a result of existing systemic health inequities and biased datasets. We have previously developed multi-modal AI (MMAI) prognostic models based on digital pathology images from five phase III randomized radiotherapy prostate cancer trials that outperform NCCN risk groups for prediction of distant metastasis (DM), biochemical failure (BF), prostate cancer-specific mortality (PCSM) and all-cause mortality (OS). In this study, we assessed the algorithmic fairness of the locked MMAI models between African American (AA) and non-AA populations in the five randomized trials. Methods: Patients enrolled in NRG/RTOG 9202, 9408, 9413, 9910, and 0126 with digitized biopsy histopathology slides were included in this study. The locked MMAI models were applied, and subgroup analyses were conducted by comparing distributions of clinical variables and MMAI scores (medians for continuous variables and proportions for categorical variables reported), and evaluating MMAI models’ prognostic ability among AA and non-AA men. The performance of the models were compared using DM as the primary endpoint and secondary endpoints of BF, PCSM, OS (death without an event as a competing risk) with Fine-Gray or Cox Proportional Hazards models. Either Kaplan Meier or cumulative incidence estimates were computed and compared using log-rank or Gray’s test. Results: This study included 5,624 men: 932 (17%) AA, 4503 (80%) white, and 189 (3%) other races. AA had younger median age (69 vs 71 year [yr]), higher median baseline PSA (12 vs 10 ng/mL), more T1-T2a (62% vs 57%), more Gleason &lt; 7 (42% vs 36%) and 8-10 (15% vs 12%), and more NCCN low and high risk (12% vs 10% and 41% vs 33%). AA and non-AA had estimated 5-yr BF rates 27% and 27%, 5-yr DM rates 5% and 5%, 10-yr PCSM 5% and 7%, and 10-yr OS 58% and 60%, respectively. The median (interquartile range) score of the model optimizing for 5-yr DM (5-yr DM MMAI) was 0.044 (0.037–0.059) in AA and 0.043 (0.036–0.057) in non-AA. Similarly, all other MMAI models had differences in the medians between AA and non-AA ranging from 0.001 to 0.02. For all endpoints, the 5-yr DM MMAI model showed strong prognostic signal (hazard ratio [HR] per one standard deviation increase: 1.6 for DM, 1.4 for BF, 1.6 for PCSM and 1.3 for OS, all p-values &lt; 0.001) and had comparable trends within AA vs. non-AA in the entire cohort (e.g., HR for DM 1.4 vs 1.6). Similar results were observed for the MMAI model optimizing for 10-yr PCSM. Conclusions: To our knowledge, this represents the first comparative analyses of a digital pathology AI prognostic model in AA vs. non-AA prostate cancer patients. The prognostic performance of the AI models was found to be comparable between subgroups. Our data supports the use of these models across racial groups, though further validation in AA cohorts is ongoing.

Real time monitoring and prediction of time to endpoint maturation in clinical trials.

In clinical trials, almost all key milestone dates can be defined in terms of time to endpoint maturation (TTEM). The real time monitoring and accurate prediction of TTEM have a significant impact on clinical trial planning and execution and can bring significant value to clinical trial practitioners. TTEM is defined as the time to achieve or observe a certain number or percentage of some endpoint of interest. It is a combination of time to site initiation, time to subject enrollment after site initiation and time to event of interest after subject enrollment. To better predict TTEM during the trial, the future site initiation and subject enrollment have to be taken into account while predicting the number of events. In this article, we propose a novel simulation-based framework combining time to site initiation, time to subject enrollment and time to event in order to predict TTEM. A nonhomogeneous Poisson process with a quadratic time-varying rate function is used to model site initiation and subject enrollment and more advanced time to event models had been explored and integrated on top of them, such as Weibull, piecewise exponential, and model averaging which is equivalent to a Bayesian model selection strategy. To evaluate the predictive performance of the proposed methodology, we conducted extensive simulations and applied the methodology to 14 randomly selected real oncology phase 2 and phase 3 studies in both solid tumor and hematology with a total 31 study-endpoint combinations. The predictive performance of the proposed methodology was then compared with popular and commonly available commercial software, for example, East (Cytel, Cambridge, MA, USA). From both simulation and real data, the proposed methodology can significantly improve the prediction accuracy by up to 54% compared to the commonly available method.

Utilizing restricted mean duration of response for efficacy evaluation of cancer treatments.

In oncology clinical trials, response-based endpoints (time to response, objective response, duration of response [DOR]) are commonly used to detect therapeutic effect to support proof-of-concept or submission decisions. The restricted mean DOR (RMDOR) was recently proposed as a composite nonparametric method to efficiently quantify the treatment effect related to tumor reductions, which offers an intuitive way to perform statistical inference in cross-arm comparison and has since been applied in some Phase III studies. In this paper, we provide further technical details and asymptotic properties of the RMDOR method and discuss the selection of the truncation time. A simulation study is conducted comparing the performance of the proposed method with existing standard methods in hypothesis testing and quantification of treatment efficacy. We use two oncology Phase III examples to illustrate the method. An R package PBIR and a SAS macro are available to perform statistical inference based on the RMDOR.

Development and validation of a prognostic AI biomarker using multi-modal deep learning with digital histopathology in localized prostate cancer on NRG Oncology phase III clinical trials.

222 Background: Prognostication in localized prostate cancer is reliant on non-specific tools, an issue that leads to the over- and under-treatment of patients. Various tissue-based molecular biomarkers have attempted to fill this unmet need, but most lack prospective randomized trial validation. Herein, we train and validate prognostic biomarkers in localized prostate cancer using five phase III randomized trials, by leveraging multi-modal deep learning on digital histopathology. Methods: Histopathology image data was generated from pre-treatment biopsy slides in five NRG Oncology phase III randomized radiotherapy prostate cancer trials (RTOG 9202, 9408, 9413, 9910, and 0126). The trials were randomly split into training (80%) and validation (20%) cohorts. A multi-modal artificial intelligence (MMAI) architecture was developed to take clinicopathologic and image-based (histopathology) data as input and predict binary outcomes. Using this architecture, various models were trained to predict relevant clinical endpoints: biochemical recurrence (BCR), distant metastasis (DM), prostate cancer-specific survival (PCaSS), and overall survival (OS). These models were then validated for measures of prognostic discrimination using the time-based area under the curve (AUC) method. Results: Clinicopathologic and histopathology image data was available for 5,654 of 7,957 eligible patients (71.1%), yielding 16.1 TB of data from 16,204 histopathology slides of pretreatment biopsy samples. After training the models, locking them, and evaluating them on the validation cohort, we found that the MMAI prognostic model had superior discrimination compared to the NCCN model (PSA, T-stage, and Gleason score) for 5-year DM (AUC of 0.84 vs 0.73), 5-year BCR (AUC of 0.69 vs 0.58), 10-year PCaSS (AUC of 0.79 vs 0.66), and 10-year OS (AUC of 0.65 vs 0.58). Within each of the individual trials in the validation cohort, the MMAI-model had superior performance compared to NCCN risk groups for all clinical endpoints. Conclusions: This represents the first ever development and validation of prognostic biomarkers in localized prostate cancer using multiple large phase III clinical trials. We have successfully validated that our MMAI-prognostic biomarkers are superior to standard clinical and pathologic variables in identifying future BCR, DM, PCaSS, and OS. This massively scalable technology is feasible and can help personalize the management of prostate cancer patients. Funding: This project was supported by grants U10CA180868 (NRG Oncology Operations), U10CA180822 (NRG Oncology SDMC), UG1CA189867 (NCORP), U24CA196067 (NRG Specimen Bank) from the National Cancer Institute (NCI).