Avelumab first-line maintenance in advanced urothelial carcinoma: Complete screening for prognostic and predictive factors using machine learning in the JAVELIN Bladder 100 phase 3 trial.

BACKGROUND: Avelumab first-line (1 L) maintenance is a standard of care for advanced urothelial carcinoma (aUC) based on the JAVELIN Bladder 100 phase 3 trial, which showed that avelumab 1 L maintenance + best supportive care (BSC) significantly prolonged overall survival (OS) and progression-free survival (PFS) vs BSC alone in patients who were progression free after receiving 1 L platinum-containing chemotherapy. Here, we comprehensively screened JAVELIN Bladder 100 trial datasets to identify prognostic factors that define subpopulations of patients with longer or shorter OS irrespective of treatment, and predictive factors that select patients who could obtain a greater OS benefit from avelumab 1 L maintenance treatment. METHODS: We performed machine learning analyses to screen a large set of baseline covariates, including patient demographics, disease characteristics, laboratory values, molecular biomarkers, and patient-reported outcomes. Covariates were identified from previously reported analyses and established prognostic and predictive markers. Variables selected from random survival forest models were processed further in univariate Cox models with treatment interaction and visually inspected using correlation analysis and Kaplan-Meier curves. Results were summarized in a multivariable Cox model. RESULTS: Prognostic baseline covariates associated with OS included in the final model were assignment to avelumab 1 L maintenance treatment, Eastern Cooperative Oncology Group performance status, site of metastasis, sum of longest target lesion diameters, levels of C-reactive protein and alkaline phosphatase in blood, lymphocyte proportion in intratumoral stroma, tumor mutational burden, and tumor CD8+ T-cell infiltration. Potential predictive factors included site of metastasis, tumor mutation burden, and tumor CD8+ T-cell infiltration. An analysis in patients with PD-L1+ tumors had similar findings to those in the overall population. CONCLUSIONS: Machine learning analyses of data from the JAVELIN Bladder 100 trial identified potential prognostic and predictive factors for avelumab 1 L maintenance treatment in patients with aUC, which warrant further evaluation in other clinical datasets.

Comparison of tumor assessments using RECIST 1.1 and irRECIST, and association with overall survival.

BACKGROUND: Patients treated with immune checkpoint inhibitors (ICIs) may experience pseudoprogression, which can be classified as progressive disease (PD) by Response Evaluation Criteria in Solid Tumors (RECIST) V.1.1 and could lead to inappropriate treatment discontinuation. Immune-response criteria were developed to better capture novel response patterns seen with ICIs. METHODS: We pooled data from 1765 patients with 12 types of advanced solid tumors treated with avelumab (an anti-programmed death ligand 1 (PD-L1) monoclonal antibody) monotherapy in the JAVELIN Solid Tumor and JAVELIN Merkel 200 trials, conducted a comparative analysis of tumor assessments by investigators according to RECIST 1.1 and immune-related RECIST (irRECIST), and evaluated the correlation between progression-free survival (PFS) and overall survival (OS). RESULTS: In total, 147 patients (8.3%) had a best overall response (BOR) of PD by RECIST 1.1 but had immune-related disease control by irRECIST (defined as immune-related BOR (irBOR) of immune-related stable disease or better). This discordance was seen irrespective of PD-L1 status and observed across all tumor types. Overall, PFS and immune-related PFS showed similar imputed rank correlations with OS. CONCLUSIONS: The use of irRECIST identified a subset of patients with a BOR of PD by RECIST 1.1 but an irBOR of immune-related disease control by irRECIST with a distinctive survival curve, thereby providing more clinically relevant information than RECIST 1.1 alone. However, as a surrogate endpoint for OS in the whole population, immune-related PFS by irRECIST did not show improved predictive value compared with PFS by RECIST 1.1.

Long-term avelumab in advanced non-small-cell lung cancer: summaries and post hoc analyses from JAVELIN Solid Tumor.

Background: This study examined patients with advanced non-small-cell lung cancer who received long-term avelumab (anti-PD-L1) in a large phase Ib trial (JAVELIN Solid Tumor). Methods: Patients receiving >2 years of avelumab were reviewed and exploratory descriptive analyses were conducted. Results: Individuals with varying baseline characteristics who had received up to 6 years of avelumab were reviewed. Overall, 37/340 (10.9%) had received ≥2 years of treatment; in this subgroup, best response was complete response in 5.4%, partial response in 59.5% and stable disease in 29.7%; 51.4% had continued treatment beyond disease progression. Conclusions: In this study, 11% of patients with advanced non-small-cell lung cancer received ≥2 years of avelumab treatment and experienced prolonged response or continued clinical benefit. Clinical Trial Registration: NCT02395172 (ClinicalTrials.gov).

Pharmacometric modeling and machine learning analyses of prognostic and predictive factors in the JAVELIN Gastric 100 phase III trial of avelumab.

Avelumab (anti-PD-L1) is an approved anticancer treatment for several indications. The JAVELIN Gastric 100 phase III trial did not meet its primary objective of demonstrating superior overall survival (OS) with avelumab maintenance versus continued chemotherapy in patients with advanced gastric cancer/gastroesophageal junction cancer; however, the OS rate was numerically higher with avelumab at timepoints after 12 months. Machine learning (random forests, SIDEScreen, and variable-importance assessments) was used to build models to identify prognostic/predictive factors associated with long-term OS and tumor growth dynamics (TGDs). Baseline, re-baseline, and longitudinal variables were evaluated as covariates in a parametric time-to-event model for OS and Gompertzian population model for TGD. The final OS model incorporated a treatment effect on the log-logistic shape parameter but did not identify a treatment effect on OS or TGD. Variables identified as prognostic for longer OS included older age; higher gamma-glutamyl transferase (GGT) or albumin; absence of peritoneal carcinomatosis; lower neutrophil-lymphocyte ratio, lactate dehydrogenase, or C-reactive protein (CRP); response to induction chemotherapy; and Eastern Cooperative Oncology Group performance status of 0. Among baseline and time-varying covariates, the largest effects were found for GGT and CRP, respectively. Liver metastasis at re-baseline predicted higher tumor growth. Tumor size after induction chemotherapy was associated with number of metastatic sites and stable disease (vs. response). Asian region did not impact OS or TGD. Overall, an innovative workflow supporting pharmacometric modeling of OS and TGD was established. Consistent with the primary trial analysis, no treatment effect was identified. However, potential prognostic factors were identified.

Estimands for overall survival in clinical trials with treatment switching in oncology.

An addendum of the ICH E9 guideline on Statistical Principles for Clinical Trials was released in November 2019 introducing the estimand framework. This new framework aims to align trial objectives and statistical analyses by requiring a precise definition of the inferential quantity of interest, that is, the estimand. This definition explicitly accounts for intercurrent events, such as switching to new anticancer therapies for the analysis of overall survival (OS), the gold standard in oncology. Traditionally, OS in confirmatory studies is analyzed using the intention-to-treat (ITT) approach comparing treatment groups as they were initially randomized regardless of whether treatment switching occurred and regardless of any subsequent therapy (treatment-policy strategy). Regulatory authorities and other stakeholders often consider ITT results as most relevant. However, the respective estimand only yields a clinically meaningful comparison of two treatment arms if subsequent therapies are already approved and reflect clinical practice. We illustrate different scenarios where subsequent therapies are not yet approved drugs and thus do not reflect clinical practice. In such situations the hypothetical strategy could be more meaningful from patient's and prescriber's perspective. The cross-industry Oncology Estimand Working Group (www.oncoestimand.org) was initiated to foster a common understanding and consistent implementation of the estimand framework in oncology clinical trials. This paper summarizes the group's recommendations for appropriate estimands in the presence of treatment switching, one of the key intercurrent events in oncology clinical trials. We also discuss how different choices of estimands may impact study design, data collection, trial conduct, analysis, and interpretation.

Impact of subsequent immune checkpoint inhibitor treatment on overall survival with avelumab vs docetaxel in platinum-treated advanced NSCLC: Post hoc analyses from the phase 3 JAVELIN Lung 200 trial.

OBJECTIVES: The JAVELIN Lung 200 phase 3 trial did not meet its primary endpoint of improving overall survival (OS) with avelumab vs docetaxel in patients with platinum-treated PD-L1+ NSCLC. We report post hoc analyses assessing the effects of subsequent immune checkpoint inhibitor (ICI) treatment on OS. MATERIAL AND METHODS: Patients with stage IIIB/IV NSCLC progressed following platinum-doublet therapy were randomized to receive avelumab or docetaxel. OS was analyzed in the PD-L1+ population (≥1% of tumor cells) and full analysis set (PD-L1+ or PD-L1-). Effects of subsequent ICI (after permanent discontinuation of study treatment) on OS were analyzed using a preplanned naive sensitivity analysis and post hoc inverse probability of censoring weighting (IPCW) analysis. Subgroups with or without subsequent ICI treatment were analyzed using descriptive statistics. RESULTS: In the avelumab and docetaxel arms, a subsequent ICI was received by 16/396 (4.0 %) and 104/396 (26.3 %) after a median of 10.5 months (range, 3.9-20.4) and 5.7 months (range, 0.1-24.4), respectively. Some subgroups showed trends for higher subsequent ICI treatment, including patients with non-squamous NSCLC (avelumab arm, 4.3 % vs docetaxel arm, 32.1 %) or with a baseline ECOG performance status of 0 (6.3 % vs 31.3 %); those enrolled in the early recruitment wave (11.6 % vs 54.3 %), or enrolled in the US/Western Europe (2.8 % vs 45.5 %) or Asia (11.0 % vs 35.4 %); and non-white patients (10.1 % vs 35.0 %). The hazard ratio for OS with avelumab vs docetaxel was lower in the IPCW analysis than in the naive sensitivity analysis (PD-L1+ population: 0.80 [95 % CI, 0.62-1.04] vs 0.86 [95 % CI, 0.68-1.09], respectively). CONCLUSION: In the JAVELIN Lung 200 trial, avelumab showed clinical activity as second-line treatment for patients with advanced NSCLC. Post hoc analyses suggest that the primary OS analysis may have been confounded by subsequent ICI use in the docetaxel arm. ClinicalTrials.gov identifier: NCT02395172.

Efficacy and immune-related adverse event associations in avelumab-treated patients.

BACKGROUND: Adverse events (AEs) of special interest that arise during treatment with immune checkpoint inhibitors, including immune-related AEs (irAEs), have been reported to be associated with improved clinical outcomes. We analyzed patients treated with avelumab from the JAVELIN Solid Tumor and Merkel 200 trials, examining the association between AEs and efficacy while adjusting for confounding factors such as treatment duration and event order. METHODS: We analyzed efficacy and safety data from 1783 patients treated with the programmed death ligand 1 inhibitor avelumab who were enrolled in expansion cohorts of the JAVELIN Solid Tumor and Merkel 200 trials. To analyze the association between irAEs and efficacy with regard to survival, we used a time-dependent Cox model with time-varying indicators for irAEs, as well as multistate models that accounted for competing risks and time inhomogeneity. RESULTS: 295 patients (16.5%) experienced irAEs and 454 patients (25.5%) experienced infusion-related reactions. There was a reduced risk of death in patients who experienced irAEs compared with those who did not (HR 0.71, 95% CI 0.59 to 0.85) using the time-dependent Cox model. The multistate model did not suggest that the occurrence of irAEs could predict response; however, it predicted a higher chance of irAEs occurring after a response. No association was observed between response and infusion-related reactions. CONCLUSIONS: Patients who experience irAEs showed improved survival. Although irAEs are not predictors for response to immune checkpoint inhibitors, increased vigilance for irAEs is needed after treatment with avelumab. TRIAL REGISTRATION NUMBERS: NCT01772004 and NCT02155647.

Avelumab as second-line therapy for metastatic, platinum-treated urothelial carcinoma in the phase Ib JAVELIN Solid Tumor study: 2-year updated efficacy and safety analysis.

BACKGROUND: Anti-programmed cell death ligand 1 (PD-L1)/programmed cell death 1 antibodies have shown clinical activity in platinum-treated metastatic urothelial carcinoma, resulting in regulatory approval of several agents, including avelumab (anti-PD-L1). We report ≥2-year follow-up data for avelumab treatment and exploratory subgroup analyses in patients with urothelial carcinoma. METHODS: Patients with previously treated advanced/metastatic urothelial carcinoma, pooled from two cohorts of the phase Ib JAVELIN Solid Tumor trial, received avelumab 10 mg/kg every 2 weeks until disease progression, unacceptable toxicity or withdrawal. End points included best overall response and progression-free survival (PFS) per RECIST V.1.1, overall survival (OS) and safety. Post hoc analyses included objective response rates (ORRs) in subgroups defined by established high-risk/poor-prognosis characteristics and association between time to response and outcome. RESULTS: 249 patients received avelumab; efficacy was assessed in 242 postplatinum patients. Median follow-up was 31.9 months (range 24-43), and median treatment duration was 2.8 months (range 0.5-42.8). The confirmed ORR was 16.5% (95% CI 12.1% to 21.8%; complete response in 4.1% and partial response in 12.4%). Median duration of response was 20.5 months (95% CI 9.7 months to not estimable). Median PFS was 1.6 months (95% CI 1.4 to 2.7 months) and the 12-month PFS rate was 16.8% (95% CI 11.9% to 22.4%). Median OS was 7.0 months (95% CI 5.9 to 8.5 months) and the 24-month OS rate was 20.1% (95% CI 15.2% to 25.4%). In post hoc exploratory analyses, avelumab showed antitumor activity in high-risk subgroups, including elderly patients and those with renal insufficiency or upper tract disease; ORRs were numerically lower in patients with liver metastases or low albumin levels. Objective response achieved by 3 months versus later was associated with longer OS (median not reached (95% CI 18.9 months to not estimable) vs 7.1 months (95% CI 5.2 to 9.0 months)). Safety findings were consistent with previously reported 6-month analyses. CONCLUSIONS: After ≥2 years of follow-up, avelumab showed prolonged efficacy and acceptable safety in patients with platinum-treated advanced/metastatic urothelial carcinoma, including high-risk subgroups. Survival appeared longer in patients who responded within 3 months. Long-term safety findings were consistent with earlier reports with avelumab treatment in this patient population.

Five-Factor Prognostic Model for Survival of Post-Platinum Patients with Metastatic Urothelial Carcinoma Receiving PD-L1 Inhibitors.

PURPOSE: A prognostic model for overall survival of post-platinum patients with metastatic urothelial carcinoma receiving PD-1/PD-L1 inhibitors is necessary as existing models were constructed in the chemotherapy setting. MATERIALS AND METHODS: Patient level data were used from phase I/II trials evaluating PD-L1 inhibitors following platinum based chemotherapy for metastatic urothelial carcinoma. The derivation data set consisted of 2 phase I/II trials evaluating atezolizumab (405). Two phase I/II trials that evaluated avelumab (242) and durvalumab (198) comprised the validation data sets. Cox regression analyses evaluated the association of candidate prognostic factors with overall survival. Stepwise selection was used to select an optimal model using the derivation data set. Discrimination and calibration were assessed in the avelumab and durvalumab data sets. RESULTS: The 5 prognostic factors identified in the optimal model using the atezolizumab derivation data set were ECOG-PS (1 vs 0, HR 1.80, 95% CI 1.36-2.36), liver metastasis (HR 1.55, 95% CI 1.20-2.00), platelet count (HR 2.22; 95% CI 1.54-3.18), neutrophil-to-lymphocyte ratio (HR 1.94, 95% CI 1.57-2.40) and lactate dehydrogenase (HR 1.60, 95% CI 1.28-1.99). There was robust discrimination of survival between low, intermediate and high risk groups. The c-statistic was 0.692 in the derivation and 0.671 and 0.773 in the avelumab and durvalumab validation data sets, respectively. A web based interactive tool was developed to calculate the expected survival probabilities based on risk factors. CONCLUSIONS: A validated 5-factor model has satisfactory prognostic performance for survival across 3 PD-L1 inhibitors to treat metastatic urothelial carcinoma after platinum therapy and may assist in stratification, interpreting and designing trials incorporating PD-1/PD-L1 inhibitors in the post-platinum setting.

Pathway-Based Kernel Boosting for the Analysis of Genome-Wide Association Studies.

The analysis of genome-wide association studies (GWAS) benefits from the investigation of biologically meaningful gene sets, such as gene-interaction networks (pathways). We propose an extension to a successful kernel-based pathway analysis approach by integrating kernel functions into a powerful algorithmic framework for variable selection, to enable investigation of multiple pathways simultaneously. We employ genetic similarity kernels from the logistic kernel machine test (LKMT) as base-learners in a boosting algorithm. A model to explain case-control status is created iteratively by selecting pathways that improve its prediction ability. We evaluated our method in simulation studies adopting 50 pathways for different sample sizes and genetic effect strengths. Additionally, we included an exemplary application of kernel boosting to a rheumatoid arthritis and a lung cancer dataset. Simulations indicate that kernel boosting outperforms the LKMT in certain genetic scenarios. Applications to GWAS data on rheumatoid arthritis and lung cancer resulted in sparse models which were based on pathways interpretable in a clinical sense. Kernel boosting is highly flexible in terms of considered variables and overcomes the problem of multiple testing. Additionally, it enables the prediction of clinical outcomes. Thus, kernel boosting constitutes a new, powerful tool in the analysis of GWAS data and towards the understanding of biological processes involved in disease susceptibility.

A network-based kernel machine test for the identification of risk pathways in genome-wide association studies.

Biological pathways provide rich information and biological context on the genetic causes of complex diseases. The logistic kernel machine test integrates prior knowledge on pathways in order to analyze data from genome-wide association studies (GWAS). In this study, the kernel converts the genomic information of 2 individuals into a quantitative value reflecting their genetic similarity. With the selection of the kernel, one implicitly chooses a genetic effect model. Like many other pathway methods, none of the available kernels accounts for the topological structure of the pathway or gene-gene interaction types. However, evidence indicates that connectivity and neighborhood of genes are crucial in the context of GWAS, because genes associated with a disease often interact. Thus, we propose a novel kernel that incorporates the topology of pathways and information on interactions. Using simulation studies, we demonstrate that the proposed method maintains the type I error correctly and can be more effective in the identification of pathways associated with a disease than non-network-based methods. We apply our approach to genome-wide association case-control data on lung cancer and rheumatoid arthritis. We identify some promising new pathways associated with these diseases, which may improve our current understanding of the genetic mechanisms.

Two different sides of 'chemobrain': determinants and nondeterminants of self-perceived cognitive dysfunction in a prospective, randomized, multicenter study.

OBJECTIVE: Complaints of cognitive dysfunction are frequent among cancer patients. Many studies have identified neuropsychological compromise associated with cancer and cancer therapy; however, the neuropsychological compromise was not related to self-reported cognitive dysfunction. In this prospective study, the authors examined if confounding factors masked an underlying association of self-perceived cognitive function with actual cognitive performance. Determinants of self-perceived cognitive dysfunction were investigated. METHODS: Self-perceived cognitive function and cognitive performance were assessed before treatment, at the end of treatment, and 1 year after baseline in 101 breast cancer patients randomized to standard versus intensified chemotherapy. Linear mixed-effects models were applied to test the relationships of performance on neuropsychological tests, patient characteristics, and treatment variables to self-reported cognitive function. Change of cognitive performance was tested as a predictor of change in self-reports. RESULTS: Self-perceived cognitive function deteriorated during chemotherapy and had partially recovered 1 year after diagnosis. The personality trait negative affectivity, current depression, and chemotherapy regimen were consistently related to cognitive self-reports. No significant associations with performance in any of the 12 cognitive tests emerged. Change of cognitive performance was not reflected in self-reports of cognitive function. CONCLUSIONS: Neuropsychological compromise and self-perceived cognitive dysfunction are independent phenomena in cancer patients. Generally, cancer-associated neuropsychological compromise is not noticed by affected patients, but negative affectivity and treatment burden induce pessimistic self-appraisals of cognitive functioning regardless of the presence of neuropsychological compromise. Clinicians should consider this when determining adequate therapy for patients who complain of 'chemobrain'.