Survival Endpoints Research Articles

Real-world effectiveness and safety of amivantamab for exon 20 EGFR-mutant non-small cell lung cancer (NSCLC) in French early access program: Amexon 20 GFPC.

e20529 Background: Amivantamab is a bispecific EGFR/MET antibody approved for patients with NSCLC with EGFR exon 20 insertions. In France, early access has made it possible to use it as a second or further-line treatment after at least one line of chemotherapy. Methods: This is a retrospective multicentric cohort of consecutive patients with an adenocarcinoma and EGFR exon 20 insertions, who received at least one dose of amivantamab as part of the French early access program from September 3th, 2021 (date of approval in the French early access program) until April 30th, 2022. Patients gave consent for the data collection and were enrolled in 28 sites. Primary end point was Real world Progression free Survival (Rw PFS) and secondary endpoints were description of clinical characteristics, best response, duration of treatment, overall survival (OS), CNS activity, toxicities and subsequent treatment. Results: Thirty-nine patients received amivantamab in early French access. Median age was 60 years [39-83]; 64 % of patients were female and 54% never smoker; 33% had PS 2 or more and 66 % had brain metastasis at inclusion. Amivantamab was administered as a second or third line and more in 30% and 70% of patients respectively. Patients received a median number of 10 injections [1-47]. Overall, 37 patients were evaluable for response; response rate was 35 % and disease control rate (DCR) was 62%. After a median follow-up of 11.3 months [8-16.7], median Rw PFS and OS were 3.5 months [2.6-5.8] and 11.3 months [8-17.8] respectively. Rw PFS and OS at 6 months were 33.3% and 74.4% respectively. Duration of response was 5.8 months [2.3-11.9]. Overall, 23 patients were evaluable for CNS response; 26 % had CNS response and CNS DCR was 69%. Main sites of progression were lung (43%), bone (28%), liver (12.8%) and brain (10%). Grade III/IV adverse events were reported in 11 patients (28.2%), mainly skin toxicity (n = 5), infusion reaction (n = 2), digestive disorders (n = 1), interstitial pneumonia (n = 1), neurological disorders (n = 1), fever (n = 1). Amivantamab was permanently discontinued due to toxicities in 4 patients (10.3%). Subsequent treatment was administered to 21 patients (53.8%) after amivantamab, mostly consisting of mono chemotherapy (n = 13). Conclusions: Our real-world multicenter analysis showed that amivantamab was an effective monotherapy treatment in second line and after for patients with EGFR exon 20 insertions. No new toxicities were reported.

A randomized phase 2 study of combination atezolizumab and varlilumab (CDX-1127) with or without addition of cobimetinib in previously treated unresectable biliary tract cancer.

4017 Background: Combined MEK inhibition (MEKi) and PD-L1 blockade significantly improved progression-free survival (PFS) versus PD-L1 monotherapy in patients with advanced biliary tract cancer (BTC) following first-line chemotherapy (NCT03201458). Further study has shown MEKi enhances tumor immunogenicity but impairs host T cell activation/priming. The addition of immune agonists, e.g. a CD27 agonist, can rescue T cell function in the setting of systemic MEKi in vivo and may be an effective strategy to optimize MEKi’s immunomodulatory potential when used in combination with checkpoint blockade. Methods: We conducted a randomized ph 2 trial (NCT04941287) evaluating Atezolizumab in combination with the CD27 immune agonist (CDX-1127 [Varlilumab]) with or without the addition of a MEKi (Cobimetinib) in patients with unresectable, previously treated, BTC. Adults with pathologically-confirmed BTC, s/p at least 1 prior line (no more than 2) of systemic therapy, measurable disease, and ECOG PS ≤1 were enrolled. Patients who had received previous PD-[L]1 were eligible. The study was planned for 64 evaluable subjects randomized in a 1:1 ratio, stratified by BTC site, to either AV (Atezolizumab [840mg IV days 1,15])+ Varlilumab [3mg/kg IV days 1, 15]) or CAV (Cobimetinib [60mg oral daily days 1-21, off 22-28] + Atezolizumab + Varlilumab). Overall response rate (ORR) and PFS were co-primary endpoints. The primary correlative outcome was treatment-related changes in CD8+ infiltrating T cells. Results: A preplanned interim analysis based on objective response rates did not meet the threshold for continuance in either treatment arm, so the trial was closed early for futility. In total, 57 patients were enrolled (n = 29 [CAV], n = 28 [AV], 67% had intrahepatic cholangiocarcinoma, and 33% had received previous PD-[L]1. Both CAV and AV regimens were well-tolerated without new safety signals. Adding CDX-1127 did not appear to meaningful increase immunotoxicity beyond established PD-L1 toxicity profiles. ORR was 0% (CAV) and 4% (AV). With a median follow up time of 6.2 months, median PFS was 2.2 (CAV) and 1.8 (AV) months (HR 0.71 [0.40,1.25]). In PD[L]1-experienced patients, median PFS was 3.6 (CAV) and 1.7 (AV) months (HR 0.44 [0.14,1.33]). Median OS was 10.2 (CAV) and 6.1 (AV) months (HR 0.76 [0.37,1.55]). In PD[L]1-experienced patients, median OS was 6.4 (CAV) and 4.4 (AV) months for CAV and AV (HR 0.68 [0.19,2.42]). Primary correlative endpoint and updated survival will be reported at time of meeting. Conclusions: The combination of Atezolizumab and Varlilumab with or without Cobimetinib was safe but did not improve clinical outcomes in advanced stage BTC patients treated in later lines. Though not statistically significant, treatment with CAV demonstrated numerically favorable PFS/OS compared AV among immunotherapy-experienced patients. Clinical trial information: NCT03201458 .

Factors influencing the clinical added value (CAV) in France over more than a decade.

e23102 Background: The rise of cancer drug prices has raised alarm over the need to base pricing decisions on the clinical added value (CAV). The French National Authority for Health (HAS) is responsible for health technology assessment (HTA) in France. For each indication retained for reimbursement, the HAS assesses the CAV on a 5-point scale ranging from I (Major) to V (absence of clinical added value). This assessment is primarily based on medical evidence. The CAV score is one of the factors contributing to price negotiations between the French ministry and pharmaceutical companies. Methods: A retrospective and descriptive study analysis of all indications in oncology (hematology and solid tumors), excluding accelerated approvals for innovative treatments, was conducted by the HAS comparing the 2010-2015 and 2016-2022 periods. For each assessment, information relative to the level of evidence and the clinical effect of the drug from the HAS opinion were collected and focused on the study’s conclusive endpoints and design. Results: The global distribution of CAV score between the two studied periods (2010-2015 and 2016-2022) was similar. However, the distribution of CAV score based on study design differed between the two periods. Regarding non comparative studies, the proportion of indications that were attributed a CAV score V increased 1.7 times between 2010-2015 and 2016-2022 and those with a CAV score I-III was nearly divided by 3. Differences were equally observed for the distribution of CAV score between the two periods based on the conclusive endpoints in the pivotal study. For the studies with a conclusive endpoint on overall survival, the rate of CAV I-III was multiplied by 1.68 (an increase from 6 to 36 cases). Concerning the cases with a progression-free survival endpoint, an observed increase of 1.3 times was noted for the CAV I-III. Lastly, studies with a conclusive overall response rate (ORR) saw an increase of 1.4 times for the CAV V as well as a decrease of 0.76 times for the CAV I-III. Over the most recent period (2016-2022), the conclusive endpoints identified in the evaluated studies were predominantly the overall response rate (93%, n = 42/45) for phase II studies. For phase III studies, progression-free survival and overall survival were respectively identified in 45% and 30% of the cases. Conclusions: Thisstudy shows that the number of indications evaluated has increased and that the assessment of the CAV is closely correlated with the choices of the study design (comparative or not) and the primary endpoint (clinically relevant or not). However, other elements underpin and qualify its valuation such as the medical need. These elements are reviewed in the HAS Guidance which emphasises overall survival as a primary endpoint and comparative data.

Clinical validation of a prognostic 7-marker IHC assay (7-IHC) in 382 patients (pts) with stage IB/IIA cutaneous melanoma (CM; MELARISK-001).

9572 Background: For pts with resected CM, adjuvant anti-PD1 is approved for AJCC v8 stage IIB–IV and BRAFi/MEKi for stage III–IV disease. Nonetheless, there remains a subgroup of pts within stages IB/IIA, which is at high risk of relapse and death, accounts for a considerable share of overall CM mortality (1), and lacks access to such therapy. Thus, adjuvant clinical trials in this “early”-stage setting may be merited. However, this subgroup is not detectable via AJCC staging alone. A prognostic 7-IHC assay that was developed (2), and analytically (3) and prospectively clinically validated (4) to identify stage IB/IIA pts at high risk of relapse and death, may be useful in selecting pts for adjuvant trials (2,4). Here, we report results from the MELARISK-001 multicenter study of that assay. Methods: MELARISK-001 enrolled consecutive pts diagnosed with stage IB/IIA CM from 2000–16, with available formalin-fixed paraffin-embedded primary melanoma. Specimens were analyzed by 7-IHC and pts classified as either high-risk or low-risk. Kaplan-Meier survival analysis and multivariate Cox regression analysis (MVA) against Breslow thickness, ulceration, age, and sex were performed. Results: MELARISK-001 included 382 pts; 247 (65%) pts in stage IB, 135 (35%) in stage IIA, all sentinel node-negative. Median Breslow thickness was 1.6 mm, median age 60Y. 212 pts (55%) were classified as 7-IHC high-risk, 170 pts (45%) as 7-IHC low-risk. Median follow-up for recurrence-free survival (RFS) was 90 mos, for melanoma-specific survival (MSS) 98 mos. Comparing 7-IHC high-risk vs low-risk pts, 5Y RFS was 79% vs 100%, 10Y RFS, 72% vs 99%, 5Y MSS, 92% vs 100%, and 10Y MSS 82% vs 100% (Table; logrank p<0.001). In MVA, 7-IHC risk class (hazard ratio [HR] 22.3; 95% CI 7.0–70.9), Breslow thickness (1.5; 1.1–1.9), and age (1.61; 1.0–2.5) were significant independent prognosticators of RFS. 7-IHC HR for MSS could not be computed due to 7-IHC's 100% sensitivity. Prognostic performance variables (95% CI) for RFS were: sensitivity 98% (91%–100%), positive predictive value 26% (20%–33%), negative predictive value 99% (97%–100%). Conclusions: 7-IHC risk categorization identified 98% of relapses and 100% of CM-related deaths in stage IB/IIA pts. In MVA, 7-IHC risk class was the strongest independent prognosticator of survival endpoints. 7-IHC high-risk pts have a relapse rate comparable to that of later-stage pts for whom adjuvant therapy is approved. MELARISK-001 provides further evidence supporting use of 7-IHC to identify early-stage pts at high risk of relapse and death who may benefit from adjuvant therapy trials. 1. Whiteman et al J Invest Dermatol 2015. 2. Meyer et al PloS ONE 2012. 3. Ziemer et al Diagnostics2023. 4. Meyer et al Eur J Cancer 2023. [Table: see text]

Mid- to long-term periprosthetic bone density changes after cementless short stem hip arthroplasty in elderly: A clinical and radiological analysis

IntroductionShort stem prostheses were originally designed for younger and more active patients. In recent years, they have been increasingly offered to older patients. This study evaluates the mid-to long-term survival of a short stem prosthesis and the changes in periprosthetic bone density following implantation of a cementless short hip stem in patients over 60 years of age. Methods118 patients aged over 60 received short stem prostheses. Clinical examination included Harris Hip Score (HHS) and Hip Disability and Osteoarthritis Outcome Score (HOOS). 93 patients were followed clinically for at least five years. 53 patients underwent dual-energy x-ray absorptiometry (DXA) and radiographic evaluation. Follow-up intervals were preoperative and postoperative (t0), at approximately six months (t1), at approximately two years (t2), and at approximately five years or later (t3). ResultsOver a mean 6.7-year observation period for all 118 patients, one stem revision occurred due to a traumatic periprosthetic stem fracture. The five-year survival rate for the endpoint survival of the Metha® stem in 95 at-risk patients is 99.2%. HHS improved significantly from t0 55.3 ± 11.5 (range 30–79) to t3 95.3 ± 8.6 (range 57–100) at a mean of 8.0 years (p < 0.001). HOOS improved significantly in each subscale (p < 0.001). Bone mineral density (BMD) was available for review in 53 patients after a mean of 7.1 years. BMD increased from t0 to t3 in region of interest (ROI) 3 (+0.4%) and ROI 6 (+2.9%) and decreased in ROI 1 (−10.3%), ROI 2 (−9.8%), ROI 4 (−5.3%), ROI 5 (−3.4%) and ROI 7 (−23.1%). ConclusionsThe evaluated short stem prosthesis shows a remarkably high survival rate in elderly patients, accompanied by excellent clinical results. Load transfer measurements show a metaphyseal-diaphyseal pattern with a trend towards increased diaphyseal transfer over the period observed.

Sample size determination for prediction models via learning-type curves.

This article is concerned with sample size determination methodology for prediction models. We propose to combine the individual calculations via learning-type curves. We suggest two distinct ways of doing so, a deterministic skeleton of a learning curve and a Gaussian process centered upon its deterministic counterpart. We employ several learning algorithms for modeling the primary endpoint and distinct measures for trial efficacy. We find that the performance may vary with the sample size, but borrowing information across sample size universally improves the performance of such calculations. The Gaussian process-based learning curve appears more robust and statistically efficient, while computational efficiency is comparable. We suggest that anchoring against historical evidence when extrapolating sample sizes should be adopted when such data are available. The methods are illustrated on binary and survival endpoints.

Group sequential design for time-to-event outcome with non-proportional hazards using the concept of relative time utilizing two different Weibull distributions

A group sequential design allows investigators to sequentially monitor efficacy and safety as part of interim testing in phase III trials. Literature is well developed in the case of continuous and binary outcomes, however, in case of trials with a time-to-event outcome, popular methods of sample size calculation often assume proportional hazards. In situations where the proportional hazards assumption is inappropriate as indicated by historical data, these popular methods are very restrictive. In this paper, a novel simulation-based group sequential design is proposed for a two-arm randomized phase III clinical trial with a survival endpoint for the non-proportional hazards scenario. By assuming that the survival times for each treatment arm follow two different Weibull distributions, the proposed method utilizes the concept of Relative Time to calculate the efficacy and safety boundaries at selected interim testing points. The test statistic used to generate these boundaries is asymptotically normal, allowing p-value calculation at each boundary. Many design features specific to time-to-event data can be incorporated with ease. Additionally, the proposed method allows the flexibility of having the accelerated failure time model and the proportional hazards model as constrained special cases. Real life applications are discussed demonstrating the practicality of the proposed method.

Estimating the long-term survival of unresectable stage III non-small cell lung cancer based on cure model analysis

BackgroundThe predictors of long-term survival and appropriate surrogate endpoints in unresectable stage III non-small cell lung cancer (NSCLC) treated with radiotherapy remain unclear, especially in the immune therapy era. MethodsThis study retrospectively analyzed a prospective cohort of 822 patients treated at the Chinese National Cancer Center from 2013 to 2022. Cure fractions, surrogates for long-term survival, and associated factors were assessed using a mixture cure model, with validation against a matched Surveillance, Epidemiology, and End Results (SEER) dataset. Results27.3% of patients with unresectable stage III NSCLC can achieve long-term survival after treated by radiotherapy. 4-year PFS and 5-year OS, when 80% of patients were considered cured, showed significant correlations with cure rates based on background mortality-adjusted PFS and relative survival, with R-squared values exceeding 0.85. Independent predictors of long-term survival included non-squamous cell carcinoma (non-SCC) pathological type, N category, gross tumor volume, and treatment combination with immune checkpoint inhibitors (ICIs). ConclusionsRadiotherapy, especially when combined with ICIs, offers a potential cure for a proportion of patients with unresectable stage III NSCLC. Tumor burden and ICIs are key predictors of long-term survival. The study suggested 4-year PFS and 5-year OS as surrogate endpoints for cure and long-term survival assessment.

Outcomes from hematopoietic stem cell transplantation following treosulfan-based conditioning: A clinical and pharmacokinetic analysis.

The aims of this study are to report our experience with treosulfan-based conditioning regimens for patients with non-malignant hematologic conditions, correlating clinical outcomes at different time points post-transplant with treosulfan exposure (AUC). This study was a single-center observational study investigating overall survival (OS), disease-free survival (DFS), and event-free survival (EFS) end-points post-transplant. The consequences of treosulfan AUC with respect to toxicity, correction of underlying disease, and long-term chimerism were also explored using pharmacokinetic analysis. Forty-six patients received 49 transplants with treosulfan and fludarabine-based conditioning between 2005 and 2023. Twenty-four patients also received thiotepa. Donor chimerism was assessed on either whole blood or sorted cell lines at different time points post-transplant. Thirty-nine patients received treosulfan pharmacokinetic assessment to evaluate cumulative AUC, with five infants receiving real-time assessment to facilitate daily dose adjustment. OS, DFS, and EFS were 87%, 81%, and 69%, respectively. Median follow-up was 32.1 months (range 0.82-160 months) following transplant. Lower EFS was associated with patient age (<1 year; p = .057) and lower cumulative treosulfan dose (<42 g/m2; p = .003). Stable donor chimerism in B-cell, NK-cell, and granulocyte lineages at 1-year post-transplant were more prevalent in patients receiving thiotepa conditioning. Two infants required daily dose adjustment to treosulfan to avoid high AUC. Excellent clinical outcomes and stable chimerism were observed in this patient series. The addition of thiotepa conferred no significant toxicity and trended toward sustained ongoing donor engraftment. Correlating treosulfan AUC with long-term patient outcomes is required.

Abstract IA006: Towards ctDNA-guided treatment of muscle-invasive bladder cancer

Abstract Cancer cells release cell-free DNA with tumor-specific molecular alterations into circulation, known as circulating tumor DNA (ctDNA). Numerous studies have underscored the biomarker potential of ctDNA in detecting minimal residual disease across various cancer types, including bladder cancer. The rapid clearance of ctDNA from circulation, typically under two hours, makes it an ideal candidate for real-time monitoring of tumor burden following surgical interventions and throughout oncological treatments. For ctDNA measurements to be effectively incorporated into clinical decision-making, it is crucial to conduct ctDNA-guided intervention trials that demonstrate tangible clinical benefits. Current trials focusing on muscle-invasive bladder cancer (MIBC) are supported by recent exploratory sub-analyses of completed clinical trials, which have shown that ctDNA can effectively guide treatment decisions in this context. Beyond guiding adjuvant treatments, ctDNA holds potential for directing bladder preservation strategies, potentially in combination with urinary tumor-derived DNA (utDNA) measurements. The need for treatment de-escalation, such as in neoadjuvant chemotherapy (NAC), is underscored by the observation that a significant proportion of patients undergo treatments that result in considerable adverse effects without achieving pathologic response. Traditional decision-making in cancer treatment, largely reliant on imaging-based criteria like the Response Evaluation Criteria in Solid Tumors (RECIST), often does not adequately reflect the ultimate clinical endpoint of overall survival. A ctDNA-based measure that considers ctDNA dynamics during treatment could serve as a superior tool or complement to current strategies. In conclusion, ctDNA has demonstrated significant promise as a biomarker in both retrospective and prospective studies for risk assessment in MIBC. The clinical value of ctDNA-guided treatment will soon be determined by ongoing clinical trials (TOMBOLA, IMvigor011, MODERN). In my presentation I will address the potential clinical impact of ctDNA analyses, the technological aspects involved, how urine analyses may supplement these findings, and future directions in the field. Citation Format: Lars Dyrskjøt. Towards ctDNA-guided treatment of muscle-invasive bladder cancer [abstract]. In: Proceedings of the AACR Special Conference on Bladder Cancer: Transforming the Field; 2024 May 17-20; Charlotte, NC. Philadelphia (PA): AACR; Clin Cancer Res 2024;30(10_Suppl):Abstract nr IA006.

Prediction of survival after neoadjuvant therapy in locally advanced rectal cancer - a retrospective analysis.

The aim of this retrospective analysis was to determine if the response to preoperative radio(chemo)therapy is predictive for survival among patients with locally advanced rectal cancer and may act as a potential surrogate endpoint for disease free survival and overall survival. Eight hundred seventy-eight patients from five centers were analyzed. There were 304 women and 574 men; the median age was 64.7 years. 77.6% and 22.4% of patients received neoadjuvant radiochemotherapy or short-course radiotherapy, resulting in a pathological complete response in 7.3%. T-downstaging and N-downstaging occurred in 50.5% and 37% of patients after neoadjuvant therapy. In patients with T-downstaging, the 10-year DFS and 10-year OS were 64.8% and 66.8% compared to 37.1% and 45.9% in patients without T-downstaging. N-downstaging resulted in 10-year DFS and 10-year OS in 56.2% and 62.5% compared to 47.3% and 52.3% without N-downstaging. Based on routinely evaluated clinical parameters, an absolute risk prediction calculator was generated for 5-year disease-free survival, and 5-year overall survival. T-downstaging and N-downstaging after neoadjuvant radiochemotherapy or short-course radiotherapy resulted in better DFS and OS compared to patients without response. Based on clinical parameters, 5-year DFS, and 5-year OS can be predicted using a prediction calculator.

Novel non-linear models for clinical trial analysis with longitudinal data: A tutorial using SAS for both frequentist and Bayesian methods.

Longitudinal data from clinical trials are commonly analyzed using mixed models for repeated measures (MMRM) when the time variable is categorical or linear mixed-effects models (ie, random effects model) when the time variable is continuous. In these models, statistical inference is typically based on the absolute difference in the adjusted mean change (for categorical time) or the rate of change (for continuous time). Previously, we proposed a novel approach: modeling the percentage reduction in disease progression associated with the treatment relative to the placebo decline using proportional models. This concept of proportionality provides an innovative and flexible method for simultaneously modeling different cohorts, multivariate endpoints, and jointly modeling continuous and survival endpoints. Through simulated data, we demonstrate the implementation of these models using SAS procedures in both frequentist and Bayesian approaches. Additionally, we introduce a novel method for implementing MMRM models (ie, analysis of response profile) using the nlmixed procedure.

Need for consensus on primary end points and efficacy definitions in trials for adult acute lymphoblastic leukemia

AbstractThe lack of consensus on acceptable primary end points and definitions of response and survival in phase 2/3 efficacy studies for adult acute lymphoblastic leukemia has led to widely different clinical trial designs. Inconsistency in primary end point selection and lack of consensus on response, survival end points, and adequate follow-up time lead to difficulty in interpreting completed studies and developing future trials. The lack of consensus also runs the risk of integrating ineffective or unacceptably toxic regimens into clinical practice and future trials. Increasingly, studies integrating highly active, targeted agents into chemotherapy use short-term end points of response, measurable residual disease–negative response, and early event-free survival without confidence that these end points will translate into improved late patient outcomes. This article highlights the current consequences and dilemmas caused by this lack of consensus. The hope is to stimulate discussion and ultimately consensus to improve the interpretation and application of clinical trial results.

Characterization and expression of the microtubule associated protein tau gene isoforms and their impact on clinical outcomes in glioma patients

Gliomas are the most common tumors found in the central nervous system. They are classified based on their histopathological malignancy. Low-grade gliomas (LGG) can turn into glioblastomas (GBM), the most aggressive form of brain cancer with the worst prognosis. Previous studies have shown that microtubule-associated protein tau (MAPT) is increased in GBM and that reducing it can improve survival in animal models. On the other hand, elevated MAPT expression has been linked to increased survival in LGG, indicating its potential as a survival biomarker. However, previous studies have primarily focused on the canonical protein-coding transcripts and gene-level expression of MAPT, disregarding the diversity of its many isoforms. Many diseases in almost every tissue have been associated with a dysregulation of alternative splicing, either through an altered ratio of the expressed isoforms or by expressing aberrant transcripts with different alternative transcription start and termination sites. This study investigates the association of specific MAPT transcripts with survival in LGG and GBM using publicly available genomic datasets and performing statistical analyses. The results of the study indicated coding and several non-coding isoforms associated with overall survival (OS), disease-specific survival (DSS), and progression-free interval (PFI). After including the relevant predictors as covariates, it was observed that among the protein-coding isoforms, MAPT-205 and MAPT-201 were significantly associated with clinical survival endpoints, but with opposite trends. These findings highlight the importance of studying the role of MAPT isoforms in glioma cellular phenotypes, aggressiveness, and patient survival.

Health-related quality of life in trials with high rates of early censoring: Caution advised

Health-related quality-of-life (HRQoL) data are central to capturing the quality of patients’ life, while endpoints like overall survival (OS) focus on the quantity of life. When analyzing HRQoL data gathered from patients in a randomized trial, a key consideration is the completion rate – indicating the proportion of patients remaining in the trial and with completed questionnaires. When completion rates are disproportionately low in one treatment arm, one likely explanation is that patients who did not complete questionnaires suffered more from toxicities, negatively impacting their HRQoL. This is likely the case when low completion rates occur in the more toxic arm within a randomized trial. If the HRQoL analysis is run as a complete-case analysis – only considering patients without missing data – a decrement in HRQoL can be missed. Conversely, when completion rates are high, the HRQoL data are thought to be more reliable, and informative censoring is less likely. We describe why this reasoning can be inadequate. In trials where high and imbalanced rates of early censoring affect progression-free survival or OS endpoints, the completion rates only apply to the fraction of patients remaining in the trial. In those, HRQoL results should be considered with caution, and reasons for censoring in the primary time-to-event analyses should be explored before making definite conclusions about HRQoL. This is even more relevant in trials with non-inferiority design, where a benefit in HRQoL could be used as a justification to modify practice.

Abstract PO2-15-07: HER2DX genomic assay in triple-negative breast cancer (TNBC) patients treated with 12-weeks of neoadjuvant chemotherapy: a correlative analysis from WSG-ADAPT-TN phase II trial

Abstract HER2DX genomic assay in triple-negative breast cancer (TNBC) patients treated with 12-weeks of neoadjuvant chemotherapy: a correlative analysis from WSG-ADAPT-TN phase II trial Background: Biomarkers for de-escalation and escalation of systemic therapy in early-stage TNBC are needed. HER2DX is a standardized prognostic (risk-score) and predictive (pathological complete response [pCR]-score) assay based on clinical and gene expression-based data. Here we aimed to test the value of HER2DX assay in early TNBC. Methods: Standardized HER2DX was evaluated centrally using RNA from baseline FFPE tumor biopsies from the WSG-ADAPT-TN study (NCT01815242), a multicenter phase II trial that randomized 336 patients with stage I–III early TNBC to 12-weeks of nab-paclitaxel 125 mg/m2 plus gemcitabine 1000 mg/m2 d1,8 every 3 weeks (arm A) versus nab-paclitaxel 125 mg/m2 plus carboplatin AUC2 day 1,8 every 3 weeks (arm B). The primary aim was to test the ability of HER2DX pCR and risk-score models to predict pCR (ypT0/isN0) and survival endpoints, respectively, such as invasive disease-free survival (iDFS), distant disease-free survival (DDFS) and overall survival (OS). Secondary objectives were to assess the association of the i) HER2DX immune signature score and ii) percentage (%) of stromal tumor infiltrating lymphocytes (sTILs) with efficacy endpoints. Uni- and multi-variable logistic regression and Cox models tested the association of each variable with each endpoint. Results: HER2DX was evaluated in 126 (37.5%) baseline pre-treatment tumors. Mean age was 52.1 (range 26-76). cT1 represented 40.5% of cases and 72.2% were cN0. Overall, pCR rate was 34.1% (95%CI 26.1 – 43.2) and median follow-up was 5.0 years. Median sTILs at baseline was 31.6% (range 0-90) and moderately correlated with the IGG signature (coefficient=0.57). The % of HER2DX high-, medium- and low-pCR groups was 41.3%, 47.6% and 11.1%, respectively. HER2DX pCR score (as a continuous variable) was significantly associated with pCR in univariate (odds ratio [OR per 10-unit increase]=1.31, 1.06-1.64, p=0.016) and after adjusting by treatment arm (OR=1.31, 1.06-1.65, p=0.015). Overall, the pCR rates in HER2DX pCR-high, -medium and -low groups were 46.2%, 26.7% and 21.4%, respectively (high vs medium/low OR=2.48, 1.17-5.34, p=0.018). In arm B (n=51), the pCR rates in HER2DX pCR-high, -medium and -low groups were 47.8%, 25.0% and 0.0% respectively (high vs medium/low OR=3.36, 1.02-12.0, p=0.051). In terms of the risk-score, the % of HER2DX low-risk and high-risk groups was 60.3% and 39.7%, respectively. HER2DX risk score as a continuous variable was significantly associated with iDFS, DDFS and OS (p&amp;lt; 0.001 in all univariate analysis), and after adjusting by variables such as arm and pCR status (all p&amp;lt; 0.05). sTILs were not associated with any survival endpoint (all p &amp;gt;0.35). The HER2DX immune signature was significantly associated with iDFS (p=0.012), DDFS (p=0.022) and OS (p=0.012). Conclusion: The HER2DX genomic test in TNBC provides valuable insights into the response and survival following neoadjuvant taxane-based chemotherapy in the absence of immunotherapy. The development of a tailored genomic assay for TNBC is currently in progress. Citation Format: Oleg Gluz, Fara Brasó-Maristany, Ulrike Nitz, Laia Paré, Matthias Christgen, Guillermo Villacampa, Benedetta Conte, Sherko Kuemmel, Ronald Kates, Cornelia Kolberg-Liedtke, Eva-Maria Grischke, Helmut Forstbauer, Michael Braun, Mathias Warm, John Hackmann, Christoph Uleer, Bahriye Aktas, Claudia Schumacher, Ana Vivancos, Rachel Wuerstlein, Joel S. Parker, Monika Graeser, Charles M. Perou, Christine zu Eulenburg, Patricia Villagrasa, Hans-Heinrich Kreipe, Aleix Prat, Nadia Harbeck. HER2DX genomic assay in triple-negative breast cancer (TNBC) patients treated with 12-weeks of neoadjuvant chemotherapy: a correlative analysis from WSG-ADAPT-TN phase II trial [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO2-15-07.

Abstract PO1-03-03: STEEP criteria v2.0 validation: A multi-trial analysis using GEICAM and TRIO adjuvant trials to evaluate surrogate endpoints for overall survival

Abstract Background: Standardized definitions of efficacy endpoints are needed to improve clinical trial designs, data reporting and interpretation of results, and to allow cross-trial comparisons. STEEP (Standardized Definitions for Efficacy Endpoints) criteria were proposed in 2007 by an expert panel from the National Cancer Institute (NCI) to provide common definitions for clinical trials in early breast cancer (BC). Invasive disease-free survival (IDFS) was selected as a candidate surrogate endpoint for overall survival (OS) in adjuvant trials and has been widely used in the last 15 years. Advances in BC systemic treatment led to review these criteria (STEEP v2.0), with the definition of a new modified endpoint: invasive BC-free survival (IBCFS). IBCFS excludes second primary non-breast invasive cancers from IDFS which may better detect efficacy effects when the toxicity profile of the intervention is well-known and the risk of second primary cancer is small. Methods: We proposed a multi-trial analysis of 8 adjuvant chemotherapy (CT) trials from GEICAM and TRIO to evaluate the different surrogate endpoints defined in the STEEP v2.0. Trial-level surrogacy was assessed through correlation between the ln hazard ratio (HR) of OS and the ln HR of the different studied endpoints across trials. The correlation was estimated by the weighted, by trial size, least squares regression (WLS) line and evaluated comparing the R2 values between OS and the corresponding endpoint tested. Primary analyses evaluated the correlation of OS with IDFS/IBCFS. Secondary analyses evaluated the correlation of OS with other STEEP v2.0 endpoints (distant disease-free survival [DDFS], distant relapse-free survival [DRFS], recurrence-free survival [RFS], recurrence-free interval [RFI] and distant recurrence-free interval [DRFI]) and explored the surrogacy between OS and all endpoints in specific groups of patients (pts) according to their risk of relapse: hormone receptors positive (HR+)/HER2-negative (HER2-) pts with intermediate/high and high-risk and triple negative (TN) pts. Results: A total of 14,473 pts were included. Median age was 50 years (range, 20–82). All were female, and 51.8% were postmenopausal. With a median follow-up of 115 months (range, 0.0–188) we found 2,504 OS events, 4,146 IDFS events (60.9% being distant recurrences) and 3,810 IBCFS events (62.4% being distant recurrences). For each trial, HR of OS, IDFS and IBCFS indicated same conclusions. If HR&amp;lt; 1 for OS, it was also for IDFS and IBCFS and the same was true when HR &amp;gt;1. R2 results of the weighted linear correlations between the ln HR of OS versus (vs) IDFS and OS vs IBCFS were 0.86 and 0.89, respectively. Confidence intervals (CI) at 95% were both overlapping. All other efficacy endpoints showed strong correlation with OS in the overall analysis. R2 values for HR+/HER2- intermediate/high and high-risk pts were 0.89 (OS vs IDFS) and 0.84 (OS vs IBCFS). For TN pts the correlations were weaker, being 0.63 (OS vs IDFS) and 0.73 (OS vs IBCFS). All WLS R2 values are described in the table. Conclusions: All tested STEEP v2.0 endpoints were strongly correlated with OS (R2 close to 1), meaning that their use as surrogate endpoints for adjuvant CT trials is appropriate. Notably, no differences were found between them in this multi-trial analysis. Therefore, none of them is more appropriate than other to discriminate the value of a new intervention vs comparator in adjuvant CT studies with mature data and long-term follow-up. Table: Citation Format: Sara López-Tarruella, Marina Pollan, Ander Urriticoechea, Eva Carrasco, Gonzalo Spera, Miguel Martín, Begoña Bermejo, Linnea Chap, Manuel Ruíz - Borrego, John Crown, Jose Angel García-Sáenz, Arlene Chan, Angel Guerrero, Valerie Bee, Lourdes Calvo, Rodrigo Fresco, Andrea Blasco, Andrés Hernando, Dennis Slamon. STEEP criteria v2.0 validation: A multi-trial analysis using GEICAM and TRIO adjuvant trials to evaluate surrogate endpoints for overall survival [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO1-03-03.

Abstract PO2-01-14: Development of Artificial Intelligence-based Machine Learning Models for Predicting Survival In Hormone-Receptor-Positive/HER2-Negative Early Breast Cancer undergoing Neoadjuvant Chemotherapy

Abstract Background. In hormone-receptor positive (HoR)/HER2 negative early breast cancer (BC) multiple efforts have been made to predict disease recurrence and survival. Machine learning techniques have been used but few studies have looked into their applicability in predicting survival on the basis of clinico-pathological characteristics. The aim of this study was to evaluate a random survival forest model to predict the prognosis of this specific BC subtype. Methods. In this multicenter, retrospective study, patients who fulfilled the following inclusion criteria were included: (1) diagnosis of pathologically confirmed HoR-positive/HER2-negative invasive BC; (2) early or locally advanced, stage I-II-III (3) patients receiving neoadjuvant anthracycline and/or taxane-based chemotherapy, concurrently or sequentially; (4) patients undergoing surgery for primary BC. Survival endpoints were disease-free survival (DFS) and overall survival (OS). A random survival forest algorithm was used to develop the predictive model. Ten-fold cross-validation was performed. The C-index and the continuous rank probability score (CRPS) were used to evaluate the discrimination of the predictive model, and a ROC curve was used to evaluate model precision. A cut point analysis based on maximally selected rank statistics was conducted to evaluate the best cut-off in the out-of-bag (OOB) mortality that could maximize DFS and OS prediction. Variable importance was assessed using Breiman-Cutler permutation importance. Results. Overall, 572 patients with HoR-positive/HER2-negative early BC were included. At univariate analysis age, T stage, N stage, grading, ER, PR and Ki67 were found to be significantly associated with DFS and ER, PgR, HER2, pCR ypT and ypN retained statistical significance at multivariate analysis. ER, pCR, pathological T and N stage were found to be significantly associated with OS at univariate and multivariate analysis. The following variables were included in the final model: menopausal status, age, histology, grade, clinical T/N, ER/PgR, Ki67 and HER2, pathological complete response (pCR), ypT and ypN. For DFS, the cross-validated C-index was 0.68 (95% CI 0.63-0.73) and the OOB CRPS was 0.15, with a OOB performance error of 0.31. The AUC calculated at 60 months was 0.91 (95% CI 0.88-0.94). Out-of bag RSF-based risk scores for individual patients were calculated and an optimal cut-off of 22.58 was identified. The HR between high risk and low risk group was 3.08 (95% CI 2.16-4.39, p&amp;lt; 0.001). For OS, the cross-validated C-index was 0.66 (95% CI 0.60-0.71) and the OOB CRPS was 0.17, with a OOB performance error of 0.35. The AUC calculated at 60 months 0.91 (95% CI 0.88-0.95). An optimal cut-off of 21.54 was identified and the HR between low risk and high risk group was 2.50 (95% CI 1.73-3.61, p&amp;lt; 0.001). For DFS, the most important variables were cN (9.00), ypN (7.35), ER (7.21), ypT (5.38), PgR (4.11), pCR (4.01) and age (2.82) while for OS the most important variables were cN (7.88), ER (5.37), PgR (3.61), age (3.37), pCR (2.80), ypT (2.64) and ypN (2.60). Discussion. We analyzed the performance of RSF in the prediction of DFS and OS based on the contribution of clinico-pathological features commonly available at the baseline of the NCT and post-surgery. By selecting patients with HoR-positive HER2-negative disease, we were able to show which clinico-pathological features within a ML model have greater predictive importance. The model could be integrated with image-based tumor and radiology features to improve its predictive accuracy. This retrospective multicenter study suggests that the combination of easily accessible clinico-pathological features within a ML model may reliably predict DFS and OS in the context of HoR-positive/HER2-negative BCs. Citation Format: Luca Mastrantoni, Giovanna Garufi, Noemi Maliziola, Elena Di Monte, Giorgia Arcuri, Valentina Frescura, Angelachiara Rotondi, Giulia Giordano, Luisa Carbognin, Alessandra Fabi, Ida Paris, Gianluca Franceschini, Armando Orlandi, Antonella Palazzo, Giovanni Scambia, Giampaolo Tortora, Emilio Bria. Development of Artificial Intelligence-based Machine Learning Models for Predicting Survival In Hormone-Receptor-Positive/HER2-Negative Early Breast Cancer undergoing Neoadjuvant Chemotherapy [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO2-01-14.

Abstract PO1-19-10: Durvalumab + datopotamab deruxtecan in patients with PD-L1 positive advanced/metastatic triple-negative breast cancer: Arm 8 of the phase 1b/2, open label, platform BEGONIA study

Abstract Background: Patients (pts) with advanced/metastatic triple-negative breast cancer (a/mTNBC) have limited treatment options and poor prognosis. Immune checkpoint inhibitors combined with chemotherapy is the standard of care for pts with PD-L1 positive tumors; still, most pts progress within a year, highlighting a need for new treatments. BEGONIA is an ongoing Simon 2-stage, multicenter, multi-arm platform study evaluating the safety and efficacy of durvalumab (D), an anti–PD-L1 monoclonal antibody, with or without paclitaxel, in combination with novel oncology therapies as first-line treatment for a/mTNBC (NCT03742102). Datopotamab deruxtecan (Dato-DXd) is an antibody-drug conjugate (ADC) consisting of a humanized anti-trophoblast cell surface antigen 2 (TROP2) IgG1 monoclonal antibody linked to a topoisomerase I inhibitor payload via a plasma stable, tumor selective, tetrapeptide-based cleavable linker. TROP2 is broadly expressed on breast and other epithelial tumors and Dato-DXd monotherapy showed antitumor activity in heavily pretreated mTNBC pts (Bardia SABCS 2022 P6-10-03). Pts treated with Dato-DXd+D in Arm 7 of BEGONIA showed a 74% objective response rate (ORR) with median 7.2 months follow-up (Schmid SABCS 2022 PD11-09). Although responses were observed in both PD-L1–high and –low-expressing tumors, most pts (87%) had PD-L1–low-expressing tumors. In Arm 8 of BEGONIA, pts determined to have PD-L1 positive tumors by pre-existing/local testing will be recruited and treated with Dato-DXd+D to evaluate efficacy and safety in a PD-L1–high population. Methods: Eligible female pts are aged ≥18 years with untreated unresectable, locally advanced or mTNBC; ≥6 months between completion of treatment for earlier-stage breast cancer and recurrence of distant disease; ≥12 months since prior taxane therapy; ECOG PS 0/1; adequate organ function; and ≥1 nonirradiated measurable lesion. For inclusion in Arm 8, a PD-L1 positive tumor as determined by local immunohistochemistry (IHC) testing is required (either pre-existing or obtained during prescreening). Exclusion criteria for Arm 8 are clinically significant corneal disease; history or suspicion of interstitial lung disease/pneumonitis; underlying pulmonary disorder; prior exposure to immune-mediated therapy; or prior treatment with an ADC containing a topoisomerase I inhibitor. Arm 8 will evaluate Dato-DXd (6 mg/kg) + D (1120 mg) given intravenously every 3 weeks until disease progression or unacceptable toxicity. Tumors will be assessed per RECIST v1.1 every 6 weeks for 48 weeks, then every 12 weeks thereafter. A safety run-in will not occur for Arm 8, as Dato-DXd+D was evaluated in Arm 7 and was tolerable with no DLTs reported. Part 1 of Arm 8 will enroll 30 pts. The primary endpoint of Part 1 is safety and tolerability. A futility analysis will be performed with an option to expand Arm 8 to Part 2, where an additional 27 pts will be enrolled if expansion criteria are met. The primary endpoint for Part 2 is ORR. Secondary endpoints include ORR (Part 1 only), testing for antidrug antibodies and treatment pharmacokinetics (Part 1 only), duration of response (DoR), progression-free survival (PFS), and overall survival (OS). ORR will be summarized with descriptive statistics and 95% Clopper-Pearson confidence intervals. Kaplan-Meier analysis will be used for DoR, PFS, and OS. All tumor response and survival endpoints will be based on investigator assessments. Additional PD-L1 testing with the VENTANA PD-L1 (SP263) Assay will be performed on pretreatment tumor samples; high expression is defined as ≥10% of the tumor area populated by PD-L1–expressing tumor or immune cells. TROP2 expression will be assessed by IHC. Enrollment is ongoing. Funding: AstraZeneca/Daiichi Sankyo Citation Format: Peter Schmid, Cynthia Ma, Robert Huisden, Ross Stewart, Katrin Heider, Petra Vuković, Neelima Denduluri. Durvalumab + datopotamab deruxtecan in patients with PD-L1 positive advanced/metastatic triple-negative breast cancer: Arm 8 of the phase 1b/2, open label, platform BEGONIA study [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO1-19-10.

Abstract PO4-18-05: Safety and Feasibility of Administration of an Oral Cannabis Preparation in The Preoperative Period in Breast and Oral Cavity Cancer

Abstract Introduction- Cannabis is world’s oldest cultivated medicinal plant. It has potential applications in oncology, in reducing chronic pain, stimulating appetite, alleviating nausea/vomiting, improving overall well-being as well as its anti-cancer properties. We conducted a phase-1 dose escalation study to determine safety of pre-operative oral cannabis in breast and oral cavity squamous cell carcinoma (OC-SCC) with an intent to explore its anticancer potential in the “pre-operative” window. The primary objective was to determine maximum tolerated dose (MTD) and establish dose limiting toxicity (DLT). Secondary objectives were pharmacokinetic profiling and transcriptomic analysis of tumour tissue. Methodology- Primary objective was to determine maximum tolerated dose (MTD) and establish dose limiting toxicity (DLT). Secondary objectives were pharmacokinetic profiling and transcriptomic analysis of tumour tissue. Non-metastatic breast and OC-SCC patients planned for curative surgery were consented after thorough medical and mini-psychiatric evaluation with Brief Psychiatry Rating Scale (BPRS). The investigational product (IP) comprised of a capsule containing 2.5mg of tetrahydrocannabinol (THC) and 2.5mg cannabidiol (CBD) in 100 mg monocrotolyn oil extract of dried leaves of C. sativa. The capsules were manufactured in an Ayurvedic GMP facility. The classical 3+3, phase-1 design used modified Fibonacci sequence for dose escalation, starting with 5mg THC+5mg CBD. Patients received IP once a day for 5 days after breakfast with hemodynamic and psychiatric monitoring and underwent the planned surgery on day-6. Pharmacokinetic samples were collected at predefined time-points and plasma levels of THC, 11-OH-THC, 11-COOH-THC and CBD were determined using a validated LC-MS/MS method. Tumour tissue was collected before 1st dose of cannabis and during surgery for biomarker analysis. Results- A total of 12 patients were enrolled (6-breast, 3-buccal mucosa, and 3-tongue cancers). First 3 patients completed the study without DLT at 5+5mg dose. A DLT was observed at dose level-2 (10+10mg) and therefore additional 3 patients were enrolled at the same dose. Two patients out of six at this dose had DLT (anxiety- grade-3 and somnolence- grade-2) necessitating dose de-escalation in the next cohort. At 7.5+7.5mg dose, all 3 patients tolerated the IP(MTD). Ten patients completed the study protocol. Common adverse effects were headache (2/12, grade-1), heavy head (8/12, grade-1), hypotension (2/12, grade-1 and 3), epigastric discomfort (1/12, grade-1), diarrhoea (1/12, grade-1), hyponatremia (1/12, grade-1), hypertension (1/12, grade-2), anxiety (1/12, grade-2). Pharmacokinetics of both CBD and THC was less than dose proportional. At the MTD (7.5+7.5), day-1 and day-5 AUC0-24 [median (range)] for CBD were 21.75 (12.44–41.44) ng/ml−1*h and 24.85 (13.72–35.93) ng/ml−1*h respectively. Similarly, for THC, day 1 and day 5 AUC0-24 were 11.79 (8.06–70.58) ng/ml−1*h and 68.98 (1.78 – 136.17) ng/ml−1*h, respectively. Significant accumulation of both CBD and THC was observed across all doses from day 1 to 5. A high variability in pharmacokinetics was observed for both constituents (CV Cmax = 101.26% and 117.48%; CV AUC0-24 = 102.50% and 112.37% for CBD and THC respectively. The median half-life of CBD and THC was 2.2 (1.6–10.7) and 1.9 (1.6–2.4) h respectively. There was no correlation observed between drug exposure and toxicity. There was no significant change in BPRS score before and after IP. Conclusion- We report the first phase-1 study of cannabis in breast and OC-SCC cancer in pre-operative setting for its anti-cancer potential. Biomarker analysis is awaited. MTD identified will be further explored in phase 2/3 clinical trials in breast, oral cavity, lung and pancreatic cancer with survival endpoints. Citation Format: Rajendra Badwe, Shalaka Joshi, Shiva Thiagarajan, Vikram Gota, Vaibhav Vanmali, Pallavi Daphale, Jayita Deodhar, Rohan Chaubal, Rohini Hawaldar, Jaya Chitra Aadhi, Saurabh Kumar Gupta, Delzaad Deolaliwala, Anil Bhansali, Siddhant Mistry, Satyajeet Singh, Nita Nair, Vani Parmar, Sudeep Gupta. Safety and Feasibility of Administration of an Oral Cannabis Preparation in The Preoperative Period in Breast and Oral Cavity Cancer [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO4-18-05.