Bayesian Predictive Probability Research Articles

Some design considerations incorporating early futility for single-arm clinical trials with time-to-event primary endpoints using Weibull distribution.

Sample size calculation is an essential component of the planning phase of a clinical trial. In the context of single-arm clinical trials with time-to-event (TTE) endpoints, only a few options with limited design features are available. Motivated from ethical or practical considerations, two-stage designs are implemented for single-arm studies to obtain early evidence of futility. A major drawback of such designs is that early stopping may only occur at the conclusion of the first stage, even if lack of efficacy becomes apparent at any other time point over the course of the clinical trial. In this manuscript, we attempt to fill some existing gaps in the literature related to single-arm clinical trials with TTE endpoints. We propose a parametric maximum likelihood estimate-based test whose variance component accounts for the expected proportion of loss to follow-up and different accrual patterns (early, late, or uniform accrual). For the proposed method, we present three stochastic curtailment methods (conditional power, predictive power, Bayesian predictive probability) which can be employed for efficacy or futility testing purposes. Finally, we discuss the implementation of group sequential designs for obtaining an early evidence of efficacy or futility at pre-planned timings of interim analyses. Through extensive simulations, it is shown that our proposed method performs well for designing these studies with moderate to large sample sizes. Some examples are presented to demonstrate various aspects of the stochastic curtailment and repeated significance testing methods presented in this manuscript.

Bayesian predictive probabilities: a good way to monitor clinical trials

BackgroundBayesian methods, with the predictive probability (PredP), allow multiple interim analyses with interim posterior probability (PostP) computation, without the need to correct for multiple looks at the data. The objective of this paper was to illustrate the use of PredP by simulating a sequential analysis of a clinical trial. MethodsWe used data from the Laryngobloc trial that planned to include 480 patients to demonstrate the equivalence of success between a laryngoscopy performed with the Laryngobloc® device and a control device. A crossover Bayesian design was used. The success rates of the two laryngoscopy devices were compared. Interim analyses, computed from random numbers of subjects, were simulated. ResultsThe PostP of equivalence rapidly reached the predefined bound of 0.95. The PredP computed with an equivalence margin of 10% reached the efficacy bound between 352 and 409 of the 480 included patients. If a frequentist analysis had been made on the basis of 217 out of 480 subjects, the study would have been prematurely stopped for equivalence. The PredP indicated that this result was nonetheless unstable and that the equivalence was, thus far, not guaranteed. ConclusionsBased on these interim analyses, we can conclude with a sufficiently high probability that the equivalence would have been met on the primary outcome before the predetermined end of this particular trial. If a Bayesian approach using PredP had been used, it would have allowed an early termination of the trial by reducing the calculated sample size by 15–20%.

Abstract CT011: Evaluation of durvalumab in combination with olaparib and paclitaxel in high-risk HER2 negative stage II/III breast cancer: Results from the I-SPY 2 TRIAL

Abstract Background: I-SPY2 is a multicenter, phase 2 trial using response-adaptive randomization within molecular subtypes defined by receptor status and MammaPrint risk to evaluate novel agents as neoadjuvant therapy for breast cancer. The primary endpoint is pathologic complete response (pCR, ypT0/is ypN0)). DNA repair deficiency in cancer cells can lead to immunogenic neoantigens, activation of the STING pathway, and PARP inhibition can also upregulate PD-L1 expression. Based on these rationales we tested the combination of durvalumab (anti-PDL1), olaparib (PARP inhibitor) and paclitaxel in I-SPY2. Methods: Women with tumors ≥ 2.5 cm were eligible for screening. Only HER2 negative (HER2-) patients were eligible for this treatment, hormone receptor positive (HR+) patients had to have MammaPrint high molecular profile. Treatment included durvalumab 1500 mg every 4 weeks x 3, olaparib 100 mg twice daily through weeks 1-11 concurrent with paclitaxel 80 mg/m2 weekly x 12 (DOP) followed by doxorubicin/cyclophosphamide (AC) x 4. The control arm was weekly paclitaxel x 12 followed by AC x 4. All patients undergo serial MRI imaging and imaging response at 3 & 12 weeks combined with accumulating pCR data are used to estimate, and continuously update, predicted pCR rate for the trial arm. Regimens “graduation with success” when the Bayesian predictive probability of success in a 300-patient phase 3 neoadjuvant trial in the appropriate biomarker groups reaches > 85%. Results: A total of 73 patients received DOP treatment including 21 HR- tumors (i.e. triple-negative breast cancer, TNBC) and 52 HR+ tumors between May 2018 - June 2019. The control group included 299 patients with HER2- tumors. The DOP arm graduated in June 2019, 13 months after enrollment had started, for all HER2- negative and the HR+/HER2- cohorts with > 0.85% predictive probabilities of success. 72 patient completed surgery and evaluable for pCR, the final predicted probabilities of success in a future phase III trial to demonstrate higher pCR rate with DOP compared to control are 81% for all HER2- cancers (estimated pCR rate 37%), 80% for TNBC (estimated pCR rate 47%) and 74.5% for HR+/HER2- patients (estimated pCR rate 28%). Association between pCR and germline BRCA status and immune gene expression including PDL1 will be presented at the meeting. No unexpected toxicities were seen, but 10 patients (14%) had possibly immune or olaparib related grade 2/3 AEs (3 pneumonitis, 2 adrenal insufficiency, 1 colitis, 1 pancreatitis, 2 elevated LFT, 1 skin toxicity, 2 hypothyroidism, 1 hyperthyroidism, 1 esophagitis). Conclusion: I-SPY2 demonstrated a significant improvement in pCR with durvalumab and olaparib included with paclitaxel compared to chemotherapy alone in women with stage II/III high-risk, HER2-negative breast cancer, improvement was seen in both the HR+ and TNBC subsets. Citation Format: Lajos Pusztai, Hyo S. Han, Christina Yau, Denise Wolf, Anne M. Wallace, Rebecca Shatsky, Teresa Helsten, Judy C. Boughey, Tufia Haddad, Erica Stringer-Reasor, Carla Falkson, A. Jo Chien, Rita Mukhtar, Anthony Elias, Borges Virginia, Rita Nanda, Douglas Yee, Kevin Kalinsky, Kathy S. Albain, Aixa Soyano Muller, Kathleen Kemmer, Amy S. Clark, Claudine Isaacs, Alexandra Thomas, Nola Hylton, W. Fraser Symmans, Jane Perlmutter, Michelle Melisko, Hope S. Rugo, Richard Schwab, Amy Wilson, Amy Wilson, Ruby Singhrao, Smita Asare, Laura J. van't Veer, Angela M. DeMichele, Ashish Sanil, Donald A. Berry, Laura J. Esserman, Trial Consortium I-SPY 2. Evaluation of durvalumab in combination with olaparib and paclitaxel in high-risk HER2 negative stage II/III breast cancer: Results from the I-SPY 2 TRIAL [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr CT011.

Efficacy and toxicity monitoring via Bayesian predictive probabilities in phase II clinical trials

Bayesian monitoring strategies based on predictive probabilities are widely used in phase II clinical trials that involve a single efficacy binary variable. The essential idea is to control the predictive probability that the trial will show a conclusive result at the scheduled end of the study, given the information at the interim stage and the prior beliefs. In this paper, we present an extension of this approach to incorporate toxicity considerations in single-arm phase II trials. We consider two binary endpoints representing response and toxicity of the experimental treatment and define the result as successful at the conclusion of the study if the posterior probability of an high efficacy and that of a small toxicity are both sufficiently large. At any interim look, the Multinomial-Dirichlet distribution provides the predictive probability of each possible combination of future efficacy and toxicity outcomes. It is exploited to obtain the predictive probability that the trial will yield a positive outcome, if it continues to the planned end. Different possible interim situations are considered to investigate the behaviour of the proposed predictive rules and the differences with the monitoring strategies based on posterior probabilities are highlighted. Simulation studies are also performed to evaluate the frequentist operating characteristics of the proposed design and to calibrate the design parameters.

Optimality of testing procedures for survival data in the nonproportional hazards setting.

Most statistical tests for treatment effects used in randomized clinical trials with survival outcomes are based on the proportional hazards assumption, which often fails in practice. Data from early exploratory studies may provide evidence of nonproportional hazards, which can guide the choice of alternative tests in the design of practice-changing confirmatory trials. We developed a test to detect treatment effects in a late-stage trial, which accounts for the deviations from proportional hazards suggested by early-stage data. Conditional on early-stage data, among all tests that control the frequentist Type I error rate at a fixed α level, our testing procedure maximizes the Bayesian predictive probability that the study will demonstrate the efficacy of the experimental treatment. Hence, the proposed test provides a useful benchmark for other tests commonly used in the presence of nonproportional hazards, for example, weighted log-rank tests. We illustrate this approach in simulations based on data from a published cancer immunotherapy phase III trial.

Abstract P3-11-02: Evaluation of patritumab/paclitaxel/trastuzumab over standard paclitaxel/trastuzumab in early stage, high-risk HER2 positive breast cancer: Results from the neoadjuvant I-SPY 2 trial

Abstract Background: I-SPY2 is a multicenter, phase 2 trial using response-adaptive randomization within biomarker subtypes to evaluate novel agents as neoadjuvant therapy for high-risk breast cancer. The primary endpoint is pathologic complete response (pCR) at surgery. The goal is to identify (graduate) regimens with ≥ 85% Bayesian predictive probability of success (i.e., demonstrating superiority to control) in a future 300-patient phase 3 1:1 randomized neoadjuvant trial with pCR endpoint within signatures defined by hormone-receptor (HR), HER2, and MammaPrint (MP) status. Regimens may leave the trial for futility (< 10% probability of success), maximum sample size accrual (with probability of success ≥ 10% and < 85%), or safety concerns as recommended by the independent DSMB. For HER2+ patients, the I-SPY2 control arm was 12 weekly cycles of paclitaxel+trastuzumab (TH, control) followed by doxorubicin/cyclophosphamide (AC) q2-3 weeks x4 and surgery. Patritumab is a fully human monoclonal antibody that inhibits HER3. In this experimental arm for HER2+ patients, patritumab was given q3w x 4 cycles (18mg/kg loading dose followed by 9mg/kg/dose) concurrent with paclitaxel and trastuzumab q1w x 12 weeks (PTH, treatment), followed by AC q2-3w. Methods: Women with tumors ≥ 2.5cm were eligible for screening. MP low/HR+ tumors were ineligible. MRI scans (baseline, 3 weeks after start of therapy, prior to AC, and prior to surgery) were used in a longitudinal statistical model to predict pCR for individual patients. Analysis was intention to treat. Patients who switched to non-protocol therapy count as non-pCR. Patients on treatment arm therapy at the time of arm closure are non-evaluable. Graduation potential was in 3 of 10 pre-defined signatures: all HER2+, HR-/HER2+, and HR+/HER2+. Results: The PTH regimen was stopped at the recommendation of the Safety Working Group and DSMB based on a safety event (bilateral sensorineural hearing loss, Gr 3) observed in one patient. At the time of arm closure, N=31 patients had received PTH treatment; 4 patients receiving PTH were changed to non-protocol therapy and removed from the analysis. The final estimated pCR report will consider 27 PTH and 31 TH as evaluable patients. Accrual was insufficient to assess graduation, however, there appears to be good signal in the HER2+HR- but not HER2+HR+ signatures. I-SPY 2 TRIAL Est. pCR at time of arm closureSignaturesPTH (Treatment)N= 31TH (Control)N = 31All (HER2+)0.40 (0.22 - 0.59), n=310.23 (0.09 - 0.37), n=31HR-/HER2+0.64 (0.36 - 0.91), n=110.30 (0.12 - 0.47), n=12HR+/HER2+0.28 (0.08 - 0.48), n=200.20 (0.06 - 0.34), n=19 HR+/HER2+0.28 (0.08 - 0.48), n=200.20 (0.06 - 0.34), n=19The patient who developed Gr3 sensorineural hearing loss 6 days after the 2nd patritumab (and 4th paclitaxel/trastuzumab) treatment, did not recover her hearing after patritumab was stopped, and also reported Gr3 vulvovaginal pain, vulvitis, and vaginal inflammation. Other gynecological symptoms in the PTH arm include: 1 pt with Gr1 vaginal hemorrhage, and 1 pt with Gr2 dyspareunia. There was a higher frequency of Gr3 hypokalaemia (12.5% vs. 3.2%). One pt in the PTH arm reported Gr3 small intestinal obstruction which resolved with conservative management. Conclusion: The I-SPY 2 study aims to assess the probability that investigational regimens will be successful in a phase 3 neoadjuvant trial; PTH was stopped due to safety concerns, although there was activity in the HER2+ HR- signature. This is the first report of Gr3 hearing loss associated with patritumab/paclitaxel/trastuzumab, and thus attribution is uncertain. Citation Format: Teresa L Helsten, Shelly S Lo, Christina Yau, Kevin Kalinsky, Anthony D Elias, Anne M Wallace, A. Jo Chien, Janice Lu, Julie E Lang, Kathy S Albain, Erica Stringer-Reasor, Amy S Clark, Judy C Boughey, Erin D Ellis, Douglas Yee, Angela DeMichele, Claudine Isaacs, Jane Perlmutter, Hope S Rugo, Richard Schwab, Nola M. Hylton, W. Fraser Symmans, Michelle E Melisko, Laura J van't Veer, Amy Wilson, Ruby Singhrao, Smita M Asare, Ashish Sanil, Donald A Berry, Laura J Esserman. Evaluation of patritumab/paclitaxel/trastuzumab over standard paclitaxel/trastuzumab in early stage, high-risk HER2 positive breast cancer: Results from the neoadjuvant I-SPY 2 trial [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P3-11-02.

Abstract P3-09-02: Evaluation of a novel agent plus standard neoadjuvant therapy in early stage, high-risk HER2 negative breast cancer: Results from the I-SPY 2 TRIAL

Abstract Background: I-SPY2 is a multicenter, response-adaptive randomization phase 2 trial to evaluate novel agents when added to standard neoadjuvant therapy for women with high-risk stage II/III breast cancer - weekly paclitaxel + investigational treatment x 12 wks followed by doxorubicin & cyclophosphamide(AC) q3 wks x 4 vs. weekly paclitaxel/AC (control). The primary endpoint is pathologic complete response (pCR). The goal for all investigational arms is to identify/graduate regimens with ≥85% Bayesian predictive probability of success (i.e. demonstrating superiority to control) in a future 300-patient phase 3 1:1 randomized neoadjuvant trial with a pCR endpoint within signatures defined by hormone-receptor (HR) & HER2 status & MammaPrint (MP). Findings from the graduated, previously reported Pembro4 arm (Nanda et al, ASCO 2017) supported investigation of de-escalating therapy, and determining if pembrolizumab (an anti-PD-1 antibody) alone q3 wks x 4 after weekly paclitaxel x 12 wks + pembrolizumab q3 wks x 4 was sufficient to sustain response without AC. Methods: Women with tumors ≥2.5cm were eligible for screening. MP low/HR+ were ineligible. MRI scans (at baseline, 3 wks, 12 wks, and prior to surgery) were used in a longitudinal statistical model to predict pCR for individual patients (pts). Pts who receive non-protocol therapy (e.g., carboplatin or AC for the Pembro8-noAC arm) count as non-pCR. Pembro8-noAC was open to HER2- pts for evaluation in 3 of 10 pre-defined signatures: HER2-, HR+/HER2-, and HR-/HER2-. Regimens exit the trial for futility (<10% probability of success), maximum sample size accrual (10% <probability of success <85%), or safety as recommended by the independent DSMB. Results: Pembro8-noAC was randomized to 73 pts, 3 of whom progressed while receiving pembrolizumab alone on study. Randomization to this arm continued after the first report because the rate of progression during AC over the course of the trial was estimated to be 6.5% based on serial MRI studies. However, notification of the third case prompted the study team to ask the DSMB for the summary response for this arm. Although it did not meet formal stopping rules for either graduation or futility, Pembro8-noAC was not near the target threshold pCR rates of 60% for HR-/HER2- and 30% for HR+/HER2+. As a result of this information, combined with the on-treatment progressions, assignment to Pembro8-noAC was discontinued. Treatment with pembrolizumab alone was no longer allowed due to the potential concern for progression, and investigators were given the option to administer AC with pembrolizumab or proceed with definitive surgery following the 12 weeks of paclitaxel + pembrolizumab. 34 pts had surgery results at the time the study was closed. Of the remaining 39 pts, 34 pts have on-therapy MRI assessments. Estimated pCR rates were based on all pts with information at the time (see table). Immune-related adverse events included grade 3 colitis (n=2), grade 3 pneumonitis (n=1), grade 3 transaminitis (n=1), grade 3 hypothyroidism (n=1), and grade 1-2 adrenal insufficiency (n=5). Conclusion: Although Pembro8-noAC is performing at least as well as standard paclitaxel/AC, the likelihood is very low that the regimen would be successful in a phase 3 trial. Pembrolizumab alone following 12 weeks of paclitaxel + pembrolizumab was not sufficient to sustain a response. This was quickly assessed with a small number of patients. Estimated pCR rateSignature(95% prob interval)Pembro8-noACControlHER2-0.210.2(0.09-0.32)(0.15-0.25)HR-/HER2-0.270.27(0.09-0.45)(0.19-0.35)HR+/HER2-0.150.15(0.01-0.29)(0.09-0.20) Citation Format: Minetta C. Liu, Patricia A Robinson, Christina Yau, Anne M Wallace, A. Jo Chien, Erica Stringer-Reasor, Rita Nanda, Douglas Yee, Kathy S Albain, Judy C Boughey, Heather S Han, Anthony D Elias, Kevin Kalinsky, Amy S Clark, Kathleen Kemmer, Claudine Isaacs, Julie E Lang, Janice Lu, Tara Sanft, Angela DeMichele, Nola M Hylton, Michelle E Melisko, Jane Perlmutter, Hope S Rugo, Richard Schwab, W. Fraser Symmans, Laura J van't Veer, Patricia K Haugen, Amy Wilson, Ruby Singhrao, Smita Asare, Ashish Sanil, Donald A Berry, Laura J Esserman. Evaluation of a novel agent plus standard neoadjuvant therapy in early stage, high-risk HER2 negative breast cancer: Results from the I-SPY 2 TRIAL [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P3-09-02.

Sgnlva-005: Open-label, phase II study of ladiratuzumab vedotin (LV) for advanced aerodigestive tract malignancies.

TPS469 Background: LIV-1 is a transmembrane protein with putative zinc transporter and metalloproteinase activity. It has been linked to the epidermal-to-mesenchymal transition that leads to malignant progression and metastasis. Ladiratuzumab vedotin (LV), also known as SGN-LIV1A, is an investigational antibody-drug conjugate (ADC) targeting LIV-1 that is composed of a humanized IgG1 monoclonal antibody conjugated to monomethyl auristatin E (MMAE) via a protease-cleavable linker. MMAE-linked ADCs can induce mitotic arrest and immunogenic cell death. In a phase 1 study, LV was well tolerated and showed antitumor activity in heavily pretreated patients (pts) with metastatic breast cancer (Modi et al 2017). The current study was initiated to evaluate LV in previously treated pts with advanced upper aerodigestive tract malignancies. Methods: SGNLVA-005 (NCT04032704) is an open-label, phase 2 study evaluating LV monotherapy (2.5 mg/kg IV every 3 weeks) for pts with the following advanced malignancies: gastric and gastroesophageal junction (GEJ) adenocarcinoma, esophageal squamous cell carcinoma, small cell lung cancer, non-small cell lung cancer (NSCLC)-squamous, NSCLC-nonsquamous, and head and neck squamous cell carcinoma. Up to approximately 30 pts with unresectable locally advanced or metastatic disease, measurable disease per RECIST v1.1, an ECOG score of 0 or 1, and adequate organ function are enrolling in each of the cohorts. Pts in the gastric and GEJ adenocarcinoma and esophageal squamous cell carcinoma cohorts must have received no more than 1 prior line of platinum-based cytotoxic chemotherapy and pts in the gastric and GEJ adenocarcinoma cohort should have received prior anti-PD(L)1 therapy if indicated. The study consists of a 2-stage design that includes a Bayesian predictive probability of success approach to determine futility criteria. Study objectives include objective response rate (primary); safety and tolerability, disease control rate, duration of response, progression-free and overall survival, and pharmacokinetics and immunogenicity (all secondary); and pharmacodynamics. Study enrollment is ongoing in North America. Pts will also enroll in Europe and Asia. Clinical trial information: NCT04032704.

An optimal Bayesian predictive probability design for phase II clinical trials with simple and complicated endpoints.

Most existing phase II clinical trial designs focus on conventional chemotherapy with binary tumor response as the endpoint. The advent of novel therapies, such as molecularly targeted agents and immunotherapy, has made the endpoint of phase II trials more complicated, often involving ordinal, nested, and coprimary endpoints. We propose a simple and flexible Bayesian optimal phase II predictive probability (OPP) design that handles binary and complex endpoints in a unified way. The Dirichlet-multinomial model is employed to accommodate different types of endpoints. At each interim, given the observed interim data, we calculate the Bayesian predictive probability of success, should the trial continue to the maximum planned sample size, and use it to make the go/no-go decision. The OPP design controls the type I error rate, maximizes power or minimizes the expected sample size, and is easy to implement, because the go/no-go decision boundaries can be enumerated and included in the protocol before the onset of the trial. Simulation studies show that the OPP design has satisfactory operatingcharacteristics.

Application of Bayesian predictive probability for interim futility analysis in single-arm phase II trial

BackgroundBayesian predictive probability design, with a binary endpoint, is gaining attention for the phase II trial due to its innovative strategy. To make the Bayesian design more accessible, we elucidate this Bayesian approach with a R package to streamline a statistical plan, so biostatisticians and clinicians can easily integrate the design into clinical trial.MethodsWe utilize a Bayesian framework using Bayesian posterior probability and predictive probability to build a R package and develop a statistical plan for the trial design. With pre-defined sample sizes, the approach employs the posterior probability with a threshold to calculate the minimum number of responders needed at end of the study to claim efficacy. Then the predictive probability is applied to evaluate future success at interim stages and form stopping rule at each stage.ResultsAn R package, ‘BayesianPredictiveFutility’, with associated graphical interface is developed for easy utilization of the trial design. The statistical tool generates a professional statistical plan with comprehensive results including a summary, details of study design, a series of tables and figures from stopping boundary for futility, Bayesian predictive probability, performance [probability of early termination (PET), type I error, and power], PET at each interim analysis, sensitivity analysis for predictive probability, posterior probability, sample size, and beta prior distribution. The statistical plan presents the methodology in a readable language fashion while preserving rigorous statistical arguments. The output formats (Word or PDF) are available to communicate with physicians or to be incorporated in the trial protocol. Two clinical trials in lung cancer are used to demonstrate its usefulness.ConclusionsBayesian predictive probability method presents a flexible design in clinical trial. The statistical tool brings an added value to broaden the application.

Abstract CT166: A Phase I/II, open-label, two part study of GSK3359609 in combination with tremelimumab in participants with selected, advanced solid tumors

Abstract Background: GSK3359609 (GSK609) is an inducible T cell co-stimulator (ICOS) agonist antibody and tremelimumab (treme) is a cytotoxic T-lymphocyte-antigen 4 (CTLA-4) antagonist antibody. Nonclinical and clinical studies have shown that ICOS expression on T cells is upregulated after anti-CTLA-4 treatment (tx). ICOS+ T cells have a positive association with overall survival (OS) in treme treated chemotherapy-resistant malignant mesothelioma and melanoma patients (pts). Therefore, the combination of these 2 drugs may provide greater antitumoral response than either drug alone (Coutzac C: AACR 2019 meeting abstract). Methods: This is a Phase I/II, open-label, 2-part study. In Part 1, N=~24 pts will be enrolled. The study is designed to assess treme starting at 75 mg followed by GSK609 starting at 8 mg. At dose level (DL) 1, one pt who will receive the lowest doses of both drugs will be enrolled. If no dose limiting toxicity (DLT) is observed in the DLT observation period of 28 days, then 1-3 pts will be enrolled in DL2 - DL6 in a zone-based fashion (zones include the DLs with the next highest dose of one of the two drugs - 2DL are included in each zone in this study) with different dose combinations of the drugs being evaluated. The dose of 1 of the 2 drugs will be escalated with each new DL up to the highest dose of each. If ≥ grade 2 tx-related AEs occur at DL1 or 2, additional pts will be enrolled to collect more safety information. If DLTs are observed, the bivariate Continual Reassessment Method (CRM) model will be used to guide dose recommendations. The primary objective of Part 1 will be to assess safety. Additional pts may be enrolled in Part 1 to further evaluate pharmacokinetics/pharmacodynamics (PK/PD). Part 2 is a randomized expansion enrolling recurrent/metastatic head and neck squamous cell carcinoma pts who have disease progression after platinum chemotherapy and anti-programmed death receptor protein-1 (PD-1)/anti-programmed death-ligand 1 (PD-L1) therapy, in combination or separately. Pts may not have had ≥4 lines of prior tx in either Part 1 or 2. In Part 2, pts will be randomized 2:1 to the doses of GSK609 and treme (n=60) selected from Part 1 or to the Investigator’s choice of SOC single-agent therapy (paclitaxel, docetaxel or cetuximab) (n=30). The primary objective in Part 2 is to evaluate overall survival and the primary analysis is to perform Bayesian predictive probability of success for a future hypothetical Phase III study based on survival data. Secondary endpoints include safety, ORR, PFS, DOR, and PK while pharmacodynamic and other biomarker analyses constitute exploratory endpoints. Pts will be treated for up to 2 years with survival follow-up for up to 2 years post-tx. This study has enrolled 1 pt. Study NCT03693612 funded by GSK. Citation Format: Aaron Hansen, Raghad Abdul-Karim, Naiyer Rizvi, Danny Rischen, John Hilton, Zujun Li, Patrick Ott, Natalie Karpinich-Fedoriw, Sapna Yadavilli, Xiaowei Wang, Laurel Adams, Melody Wyres, Charles Ferte, Marc Ballas, Axel Hoos, Daniel Zandberg. A Phase I/II, open-label, two part study of GSK3359609 in combination with tremelimumab in participants with selected, advanced solid tumors [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr CT166.

Abstract CT136: Evaluation of talazoparib in combination with irinotecan in early stage, high-risk HER2 negative breast cancer: Results from the I-SPY 2 TRIAL

Abstract Background: I-SPY2 is a multicenter, Phase II trial using response-adaptive randomization within biomarker subtypes to evaluate novel agents as neoadjuvant therapy for high-risk at least T2N0 breast cancer. The primary endpoint is pathologic complete response (pCR) at surgery. The goal is to identify regimens that have ≥ 85% Bayesian predictive probability of success in a 300-patient phase 3 neoadjuvant trial defined by hormone-receptor (HR) and HER2 status, and MammaPrint (MP). Regimens may leave the trial for futility (< 10% probability of success), maximum sample size accrual (with probability of success ≥ 10% and < 85%), or as recommended by the independent DSMB. For HER2- subjects the control arm is weekly paclitaxel x12 then doxorubicin & cyclophosphamide (AC) q2-3 weeks x4. For this arm, paclitaxel was omitted and replaced with maximum tolerated dose PARPi talazoparib with synergy dosed irinotecan (25mg/m2). Paclitaxel could be given adjuvantly for these subjects and non-responding subjects could be taken off of experimental therapy. Methods: Women with tumors ≥ 2.5cm were eligible for screening. MP low/HR+ tumors were ineligible. MRI scans (baseline, 3 cycles after start of therapy, prior to AC, and prior to surgery) were used in a longitudinal statistical model to predict pCR for individual patients. Talazoparib was given at 1mg daily with 25mg/m2 irinotecan q2wks. Analysis was intention to treat. Subjects who switched to non-protocol therapy count as non-pCR. Subjects on experimental therapy at time of arm closure are non-evaluable. Talazoparib/irinotecan (TI) was open only to HER2- tumors and eligible for graduation in 3 of 10 pre-defined signatures: HER2-, HR+HER2- and HR-/HER2-. Results: TI did not meet criteria for graduation and was stopped at the recommendation of the DSMB based on expectations of limited activity beyond that seen with standard treatment. Maximum sample size had been reached at the time of this recommendation and subjects currently receiving TI were allowed to continue or change to standard therapy. Exploratory “as treated” analysis for response in gBRCA mutation carriers showed 6/10 gBRCA carriers attained a pCR in the TI arm. Except for 1 patient these gBRCA pCR subjects had >90% tumor reduction by MRI after TI and prior to AC (range: 68-96%). In the TI arm pCR rates were also higher in subjects with a PARPi7-High/MP2 gene expression signature (0.344 vs 0.146). Expected differences in toxicity were seen between arms including g3/4 peripheral neuropathy on control therapy which included paclitaxel (2.6% vs none) and g3/4 neutropenia with TI (30.2% vs 8.2%). Notably gBRCA mutation carriers receiving TI had higher rates of g3/4 neutropenia (60% vs 25.9%). Conclusion: The I-SPY2 study finds the probability that investigational regimens will be successful in a Phase III neoadjuvant trial; TI did not reach the efficacy threshold of 85% probability of success in Phase III in any of the 3 signatures. However by adding talazoparib with synergy dosed irinotecan we were able to omit paclitaxel and observe similar estimated pCR rates. This informs current work to evolve the I-SPY2 trial design to reduce toxicity without compromising outcomes and develop successful combinations targeted to biology, including DNA repair deficiency. Citation Format: Richard Schwab, Amy S. Clark, Christina Yau, Nola Hylton, Wen Li, Denise Wolfe, A Jo Chien, Anne M. Wallace, Andres Forero-Torres, Erica Stringer-Reasor, Rita Nanda, Nora Jaskowiak, Judy Boughey, Tufia Haddad, Heather S. Han, Catherine Lee, Kathy Albain, Claudine Isaacs, Anthony D. Elias, Erin D. Ellis, Payal Shah, Julie E. Lang, Janice Lu, Debasish Tripathy, Kathleen Kemmer, Douglas Yee, Barbara Haley, Melanie Majure, Erin Roesch, Christos Vaklavas, Cheryl Ewing, Teresa Helsten, W Fraser Symmans, Jane Perlmutter, Hope S. Rugo, Michelle Melisko, Amy Wilson, Ruby Singhrao, Laura van 't Veer, Angela DeMichele, Smita Asare, Don Berry, Laura J. Esserman. Evaluation of talazoparib in combination with irinotecan in early stage, high-risk HER2 negative breast cancer: Results from the I-SPY 2 TRIAL [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr CT136.

Bayesian predictive distribution for a Poisson model with a parametric restriction

Predictive probability estimation for a Poisson distribution is addressed when the parameter space is restricted. The Bayesian predictive probability against the prior on the restricted space is compared with the non-restricted Bayes predictive probability. It is shown that the former predictive probability dominates the latter under some conditions when the predictive probabilities are evaluated by the risk function relative to the Kullback-Leibler divergence. This result is proved by first showing the corresponding dominance result for estimating the restricted parameter and then translating it into the framework of predictive probability estimation.

Abstract PD5-04: An open-label, phase II basket study of olaparib and durvalumab (MEDIOLA): Updated results in patients with germline BRCA-mutated (gBRCAm) metastatic breast cancer (MBC)

Abstract Background: Olaparib (Lynparza®) is approved for the treatment of gBRCAm HER2-negative MBC based on the OlympiAD study results. Olaparib-induced DNA repair defects may attract tumor-infiltrating lymphocytes, upregulate programmed cell death ligand-1 (PD-L1) and release tumor neo-antigens upon cell death. Here, the objective was to assess the efficacy and safety of olaparib in combination with durvalumab (Imfinzi), a PD-L1 agent, in gBRCAm HER2-negative MBC (NCT02734004). Methods: Patients (pts) with HER2-negative, gBRCAm MBC were eligible; prior platinum therapy without disease progression was allowed. Pts received olaparib 300 mg tablet BID for a 4-wk run-in, followed by a combination of olaparib 300 mg BID and durvalumab 1.5 g IV q 4 wks. The combination was continued until progressive disease by RECIST 1.1. Tumor assessments were done at baseline, 4 wks, and every 8 wks thereafter. The primary endpoints were disease control rate (DCR) at 12 wks, safety and tolerability. The secondary endpoints included DCR at 28 wks, objective response rate (ORR), duration of response (DoR), progression-free survival (PFS), overall survival (OS) and a biomarker analysis. A Bayesian predictive probability design was used for the statistical analysis of the primary endpoint. Results: Thirty-four pts were included in the safety, and 32 pts in the efficacy analysis. At data cut-off (DCO) on 30 Mar 18, the 12-wk DCR was 81% (90% CI: 66%, 92%) and 28-wk DCR 47% (90% CI: 32%, 63%). The ORR for the overall cohort was 56% (95% CI: 38%, 74%), with 1 (3%) complete response (CR), 17 (53%) partial response (PR), 8 (25%) stable disease (SD), and 6 (19%) progressive disease (PD). Median DoR was 9.2 months (interquartile range, 5.5 to 12.9 months). First- and second-line pts showed higher response rates (70% and 73%, respectively) compared to pts with 3 or more lines of prior chemotherapy (27%) (Table 1). The median PFS for the overall cohort was 6.7 months (95% CI of 4.6, 11.7 months). With median follow-up of 14.5 months, the median OS has not been reached. At the time of DCO, 8 (25%) pts remained on combination therapy. The most common Grade 3 or 4 adverse events reported were anemia (11.8%), neutropenia (8.8%), and pancreatitis (5.9%). Efficacy and safety results with longer follow-up will be presented. Biomarker analysis on CD3, CD8 and PD-L1 expression in the archival tissue and in paired biopsies obtained at baseline and after olaparib monotherapy run-in, will also be presented. Conclusions: The addition of durvalumab to olaparib showed promising activity with the ORR in the early line gBRCAm MBC pts, comparing favorably with olaparib monotherapy (OlympiAD) (>70% in 1–2L MEDIOLA vs 60% overall in OlympiAD). In addition, the median DoR appeared longer with the combination (9.2 months in MEDIOLA vs 6.4 months in OlympiAD). Approximately one-third of the pts in MEDIOLA were heavily pretreated (3L+) and as expected, the ORR was lower in these pts. Due to the small number of pts, these results warrant confirmation of activity in the early-line setting. Further MBC cohorts are in development. Best responses by line of therapy 1L2L3L+n101111CR010PR773SD125PD213ORR, %707327 Citation Format: Domchek SM, Postel-Vinay S, Im S-A, Hee Park Y, Delord J-P, Italiano A, Alexandre J, You B, Bastian S, Krebs MG, Waqar S, Lanasa M, Angell HK, Tang M, Gresty C, Opincar L, Herbolsheimer P, Kaufman B. An open-label, phase II basket study of olaparib and durvalumab (MEDIOLA): Updated results in patients with germline BRCA-mutated (gBRCAm) metastatic breast cancer (MBC) [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr PD5-04.

Abstract OT3-05-03: MEDIOLA: An open-label, phase I/II basket study of olaparib (PARP inhibitor) and durvalumab (anti-PD-L1 antibody)–Additional breast cancer cohorts

Abstract Background: Olaparib (Lynparza®) is a PARP inhibitor that alters the repair of single-strand DNA breaks. Durvalumab (Imfinzi®) is a monoclonal antibody against programmed cell death ligand 1 (anti-PD-L1) that promotes antitumor immune responses. MEDIOLA (NCT02734004) is a Ph I/II open-label, multicenter study enrolling patients (pts) across several tumor types (small-cell lung cancer, gastric cancer, germline BRCA-mutated [gBRCAm] BC, or platinum sensitive relapsed gBRCAm ovarian cancer). 34 pts with gBRCAm BC received olaparib 300 mg po bid for a 4-wk run-in, followed by olaparib 300 mg po bid and durvalumab 1.5 g IV q4 wks. Encouraging preliminary results support an expansion cohort in a BRCAm popn. Evidence suggests that mutations in other homologous recombination repair (HRR) genes may confer a BRCA-like phenotype, warranting an expansion of the study population beyond BRCAm. Inhibition of vascular endothelial growth factor (VEGF) has been reported to enhance the efficacy of chemotherapy in TNBC. In addition, VEGF inhibition potentiates PARP inhibitor activity, particularly in pts who do not carry BRCAm. Combinations of immune checkpoint inhibitors and bevacizumab, an anti-VEGF-A antibody, have shown promising results in other tumor types. Thus, MEDIOLA will additionally explore the efficacy and safety of olaparib + durvalumab in combination with bevacizumab in TNBC pts. Trial design: Pts in the additional MBC cohorts will receive combination olaparib and durvalumab with no olaparib run-in. Pts in the TNBC cohort will also receive concurrent bevacizumab 10 mg/kg q 2 wks. Tumor assessments will be performed at baseline and every 8 wks thereafter. Eligibility criteria: Pts with histologically confirmed, locally advanced or metastatic HER2-neg BC, who are PARP-inhibitor and immunotherapy naïve. Prior anthracycline and/or taxane therapy in early or MBC is required. Prior platinum therapy is allowed, if there was no disease progression while receiving treatment and at least 12 mths has elapsed since the last dose. Pts will undergo BRCA and HRR mutation testing and will be assigned to a cohort based on their mutation status, as illustrated in Table 1. Specific aims: Cohort-specific primary efficacy endpoint targets were calculated aiming for superiority to standard of care treatment (Table 1). Other primary outcomes are safety and tolerability. Secondary endpoints are PK, DCR at 24 wks, objective response rate, duration of response, progression-free survival and overall survival. Exploratory endpoints include the analysis of tumor-infiltrating lymphocytes and PD-L1 expression. Statistical methods: The expansion cohort will have a single-stage statistical design. Bayesian predictive probability design will be used for the analysis of the HRRm and TNBC triplet cohorts. Accrual: Accrual targets are shown in Table 1. First subject will be enrolled in Nov 2018. New MEDIOLA BC cohortsCohort nameTarget accrualDrugsPopnPrimary endpointBreast BRCAm expansion80Olap+DurvHER2-neg, tBRCAmORR 70%Breast HRRm cohort29Olap+DurvHER2-neg, non BRCAm, HRRm-posDCR at 16 wks 80%TNBC triplet cohort30Olap+Durv+BevTNBC, non BRCAm, non HRRmDCR at 16 wks 80%Bev, bevacizumab; durv, durvalumab; olap, olaparib Citation Format: Domchek SM, Postel-Vinay S, Im S-A, Hee Park Y, Delord J-P, Italiano A, Alexandre J, You B, Bastian S, Krebs MG, Waqar S, Lanasa M, Angell HK, Tang M, Gresty C, Opincar L, Herbolsheimer P, Kaufman B. MEDIOLA: An open-label, phase I/II basket study of olaparib (PARP inhibitor) and durvalumab (anti-PD-L1 antibody)–Additional breast cancer cohorts [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr OT3-05-03.

An open-label, phase II basket study of olaparib and durvalumab (MEDIOLA): Results in patients with relapsed gastric cancer.

140 Background: Olaparib is a PARP inhibitor that has shown activity in relapsed gastric cancer (GC) when combined with chemotherapy. MEDIOLA assesses the efficacy and safety of a chemo-free combination of olaparib and durvalumab, an anti-programmed cell death ligand-1 (PD-L1) agent in patients (pts) with solid tumors, including relapsed GC (NCT02734004). Methods: Eligible pts had GC that relapsed following platinum-containing therapy. Pts received olaparib 300 mg PO BID for a 4-wk run-in to allow for serial biopsies, followed by combination olaparib and durvalumab (1.5 g IV q 4 wks) until progressive disease (PD) as measured by RECIST 1.1. Tumor measurements were taken at baseline, 4 wks and q 8 wks thereafter. Primary endpoints were disease control rate (DCR) at 12 wks, safety, and tolerability. The secondary endpoints included DCR at 28 wks, objective response rate (ORR), duration of response (DoR), progression-free survival, and OS. Biomarker endpoints included PD-L1 expression and evaluation of tumor infiltrating lymphocytes. Bayesian predictive probability design was used for statistical analysis. Results: Forty pts were included in the safety and 39 in the efficacy analysis. Among 39 pts, median age was 57 yrs (range 28–77). Nineteen pts (48%) had a grade 3 AE; three pts (8%) a grade 4 AE. Most common AEs were anemia, lipase increase, fatigue, dysphagia, hyponatremia, and alkaline phosphatase increase. Ten pts (25%) had immune-mediated AEs, most commonly rash. The ORR was 10%; two pts had complete response; two had partial response. Median DoR was 11.1 months. DCR at 12 wks was 26%. Further efficacy and biomarker data will be presented. Conclusions: The combination of olaparib and durvalumab was tolerable, with no unexpected AEs. All responses occurred after the addition of durvalumab and were durable, suggesting synergistic treatment effect of the combination in some pts. Furthermore, several pts with early PD showed unexpectedly long survival. DCR at 12 wks was below the target (70%) due to a high rate of early PDs following the olaparib run-in. To address the early treatment failures, an upfront addition of more aggressive therapies to the combination should be explored. Clinical trial information: NCT02734004.

Bayesian Predictive Distribution for a Negative Binomial Model

Estimation of the predictive probability function of a negative binomial distribution is addressed under the Kullback—Leibler risk. An identity that relates Bayesian predictive probability estimation to Bayesian point estimation is derived. Such identities are known in the cases of normal and Poisson distributions, and the paper extends the result to the negative binomial case. By using the derived identity, a dominance property of a Bayesian predictive probability is studied when the parameter space is restricted.

Abstract CT003: Preoperative pembrolizumab (pembro) before radical cystectomy (RC) for muscle-invasive urothelial bladder carcinoma (MIUC): Interim clinical and biomarker findings from the phase II PURE-01 study

Abstract Introduction: MIUC is an aggressive and cisplatin-based neoadjuvant chemotherapy (CT) is recommended, but it is administered in a minority of pts. Pembro is an EMA and FDA-approved therapy for metastatic UC after platinum failure or for cisplatin-ineligible pts. Methods: PURE-01 (NCT02736266) is an open-label, single-arm, phase 2 study to evaluate pembro before RC. Inclusion criteria: predominant UC histology, ≤T3bN0 stage, and eligibility to cisplatin or carboplatin CT. Pts were staged with PET/CT scan and multiparametric bladder MRI (mpMRI), and received 3 cycles of pembro 200mg 3 weekly before RC. No additional anti PD-1/PD-L1 therapy is allowed post-operatively. Pathologic complete response (pT0) is the primary endpoint (with ITT principle for analyses). The H1 is pT0 ≥20% and H0 pT0≤10%. Stopping rules based on Bayesian predictive probability will be applied over the planned total sample size of 90 pts (a0=0.5, b0=0.5, θT=0.90, θL=0.10, θU=0.90, p=0.20). Biomarker analyses on pre- (TURB) and post-therapy (RC) lesions include: IHC PD-L1 combined positive score (CPS, Dako 22C3) and genomic sequencing with FoundationOne assay (Foundation Medicine Inc., Cambridge, MA, USA), with tumor mutational burden (TMB) quantification. A first interim analysis is presented. Results: From 02/17-01/18, 34 pts were enrolled (28M, 6F). 21 had cT3, 13 cT2, 6 had hydronephrosis. All pts had evident disease at mpMRI before starting pembro. One pt (2.9%) had G3 irAE (ALT increase) and suspended pembro; 5 (14.7%) had reversible G2 irAE. At the time of interim analysis, 22 pts are evaluable for the primary endpoint. There were 7/22 pT0 (31.8%) and 3 pTa/is (total pT≤2 rate: 45.4%). In TURB samples, 100% pT0 pts had PD-L1 CPS>20%, versus all non-pT0 pts who had CPS≤10%. Median TMB in pts with pT≤2 response was 12.28 mut/mb versus 7.02 mut/mb in non-responding pts. On 6 evaluable pts, substantial differences were found in pre- versus post-therapy mutation profile (mean: 44.5% overlapping mutations), including FGFR alterations. Conclusion: Pembro is safe and endowed with a potent antitumor and biological activity in MIUC. Signals toward a strong association between CPS>20% and pT0 response, coupled with higher TMB and pT≤2 response, yield the potential for practice changing, pending full translational findings in this cohort that will be presented at the meeting. Citation Format: Andrea Necchi, Alberto Briganti, Marco Bianchi, Daniele Raggi, Patrizia Giannatempo, Massimo Freschi, Maurizio Colecchia, Nicola Fossati, Giorgio Gandaglia, Renzo Colombo, Andrea Gallina, Andrea Salonia, Roberto Salvioni, Siraj M. Ali, Jon H. Chung, Francesco Montorsi. Preoperative pembrolizumab (pembro) before radical cystectomy (RC) for muscle-invasive urothelial bladder carcinoma (MIUC): Interim clinical and biomarker findings from the phase II PURE-01 study [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr CT003.

Bayesian probability of predictive agreement for comparing the outcome of two separate regressions

AbstractThe concept of a Bayesian probability of agreement was recently introduced to give the posterior probabilities that the response surfaces for two different groups are within δ of one another. For example, a difference of less than δ in the mean response at fixed levels of the predictor variables might be thought to be practically unimportant. In such a case, we would say that the mean responses are in agreement. The posterior probability of this is called the Bayesian probability of agreement. In this article, we quantify the probability that new response observations from two groups will be within δ for a continuous response, and the probability that the two responses agree completely for categorical cases such as logistic regression and Poisson regression. We call these Bayesian comparative predictive probabilities, with the former being the predictive probability of agreement. We use Markov chain Monte Carlo simulation to estimate the posterior distribution of the model parameters and then the predictive probability of agreement. We illustrate the use of this methodology with three examples and provide a freely available R Shiny app that automates the computation and estimation associated with the methodology.

Predictive probability methods for interim monitoring in clinical trials with longitudinal outcomes.

In clinical research and development, interim monitoring is critical for better decision-making and minimizing the risk of exposing patients to possible ineffective therapies. For interim futility or efficacy monitoring, predictive probability methods are widely adopted in practice. Those methods have been well studied for univariate variables. However, for longitudinal studies, predictive probability methods using univariate information from only completers may not be most efficient, and data from on-going subjects can be utilized to improve efficiency. On the other hand, leveraging information from on-going subjects could allow an interim analysis to be potentially conducted once a sufficient number of subjects reach an earlier time point. For longitudinal outcomes, we derive closed-form formulas for predictive probabilities, including Bayesian predictive probability, predictive power, and conditional power and also give closed-form solutions for predictive probability of success in a future trial and the predictive probability of success of the best dose. When predictive probabilities are used for interim monitoring, we study their distributions and discuss their analytical cutoff values or stopping boundaries that have desired operating characteristics. We show that predictive probabilities utilizing all longitudinal information are more efficient for interim monitoring than that using information from completers only. To illustrate their practical application for longitudinal data, we analyze 2 real data examples from clinical trials.