EORTC stomach cancer PD-L1 biomarker European initiative: the ASPIRE study protocol

• Published in: ESMO Gastrointestinal Oncology Vol. 5; p. 100071
• Main Authors: Petrillo, A., Oudijk, L., Sundar, R., Daumer, C., Casas, J., D’Haese, D., Mauer, M., van Grieken, N., Smyth, E.C., Moehler, M.
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.09.2024; Elsevier
• Subjects: combined positive score; immunotherapy; PD-L1; tumour area positivity; combined positive score; PD-L1; tumour area positivity; immunotherapy
• ISSN: 2949-8198; 2949-8198
• DOI: 10.1016/j.esmogo.2024.100071

The evaluation of programmed death-ligand 1 (PD-L1) expression and the methodology employed are central to identify suitable candidates for immunotherapy among patients with gastro-oesophageal cancer (GC). Yet, there are no comprehensive global studies comparing the various methods and antibodies utilized for assessing PD-L1 positivity in GC. The ASPIRE study, led by the European Organisation for Research and Treatment of Cancer Gastrointestinal Tract Group (EORTC GITCG) and the National University Health System, Singapore, seeks to standardize the assessment of PD-L1 expression in GC. By comparing various PD-L1 scoring systems and assays, the study aims to simplify and harmonize the quantification and qualification of PD-L1 expression. Ultimately, this effort aims to facilitate the translation of endpoints in companion diagnostic settings. Here, we report the protocol of the study. [Display omitted] Specifications tableSubject areaMedicine and DentistryMore specific subject areaGastro-oesophageal adenocarcinomaName of your trial in progressASPIREReagents/tools-Dako PD-L1 clones 28-8 and 22C3 pharmDx assay (Agilent Technologies, Santa Clara, CA)-PD-L1 clone SP263 (Ventana Medical Systems Inc, Tucson, AZ)Trial design1.Experimental description and trial designThe evaluation of programmed death-ligand 1 (PD-L1) and its methodology are central to identifying suitable candidates for immunotherapy in gastro-oesophageal cancer (GC). Yet, no studies are currently comparing PD-L1 assessment methods and immunohistochemistry (IHC) antibodies used in this context. Through this study, we seek to compare various PD-L1 scoring systems and assays, aiming to streamline and standardize the quantification and qualification of PD-L1 expression in GC.This study aims to collect ∼150 archival samples from GC through the European Organisation for Research and Treatment of Cancer (EORTC) network across Europe and with the help of international collaborators from Asia to ensure global diverse representation.Biopsy or surgical specimen formalin-fixed, paraffin-embedded blocks will be sent centrally to the laboratory at National University Health System, Singapore, where the PD-L1 staining will be carried out in a centralized and standardized manner using the PD-L1 IHC 22C3 pharmDx and PD-L1 28-8 pharmDx assays and Ventana PD-L1 (SP263).Subsequently, a group of 10-15 international gastrointestinal experts and clinically active pathologists will score the samples for PD-L1 using the combined positive score (CPS) and tumour area positivity (TAP). These scores will then be merged with clinical and pathological data.Agreement and interchangeability among the different scoring systems and assays will be assessed.2.Objectives and hypothesis of the project2.1Study objectives2.1.1Primary objective•To measure the interchangeability of the PD-L1 expression among IHC assays 22C3, SP263 and 28-8•To compare the scoring systems CPS and TAP in samples of GC2.2Endpoints2.2.1Primary endpoint•The agreement among PD-L1 expression status after dichotomizing the TAP and CPS scores at 1 and 5 cut-off2.2.2Secondary endpoints•The difference among the CPS and TAP scores of GC samples stained with PD-L1 22C3, SP263 and 28-8 antibodies•The concordance among PD-L1 expression status after dichotomizing the TAP and CPS scores at other cut-offs than 1 and 5•The cut-off that would result in the smallest difference among IHC assays•The inter-pathologist variability, quantified using the intraclass correlation coefficient (ICC) of CPS and TAP scores•The inter-pathologist variability, quantified using the Fleiss kappa (κ) statistic (FKS) on dichotomized scores at multiple cut-off levels2.2.3Exploratory endpoints•Co-expression of PD-L1 with other biomarkers of interest including Epidermal growth factor receptor 2 (HER2), Claudin 18.2 (CLDN18.2) and Fibroblastic growth factor receptor 2b (FGFR2b)•To estimate absolute performance among the IHC assays: differences in CPS and TAP scores of positive and negative control cell lines stained with PD-L1 22C3, SP263 and 28-8 antibodies3.Study population3.1Population sample size•Approximately 150 GC samples3.2Patient selection criteria•Adult (≥18 years old)•Histologically diagnosed with gastric or gastro-oesophageal junctional adenocarcinoma•Availability of tumour samples obtained within the past 3 years from biopsy or resection4.Research methodology and statistical plan4.1Sample size calculationThe number of GC samples needed for this study was estimated based on:•A lower boundary of 80% overall rate of agreement (ORA) below which two assays are considered significantly different•A two-sided type I error constraint alpha of 0.015 accounting for the three comparisons (among the three assays)•A two-sided type II error constraint beta of 0.20A sample size of n = 127 is needed for the lower boundary of a one-sided 95% confidence interval (CI) around a true rounded-off ORA of 90% to remain above 80%. Taking a margin of 10% resulted in an estimated sample size of n = 140.4.2Assessment of relative performance4.2.1Primary endpointsThe binary primary endpoint will be assessed through three 2 × 2 contingency tables depicting the number of slides that agree (n++ and n−−) or do not agree (n+− and n−+) between two candidate IHC assays. For each contingency table, the sum of all counts in the table (n) corresponds to the number of slides for which both assays have produced a valid PD-L1 expression status. The association between the qualitative results of two IHC assays will be presented as the ORA:ORA=n+++n−−nThe ORA between two IHC assays will always be accompanied by the 95% CI calculated by Wilson.4.2.2Secondary endpointsThe continuous secondary endpoint on PD-L1 expression will be assessed through the within-slide differences among the mean CPS and TAP scores of the IHC assays and their statistical significance. Three assays will have three such inter-assay differences, which will be calculated.The concordance among PD-L1 expression status after dichotomizing TAP and CPS scores at other cut-offs than 1 and 5 will be calculated as explained in 4.2.1.The best cut-off, i.e. the cut-off that results in the smallest difference among IHC assays, will be established as the cut-off that maximizes the sum of the positive (PA) and negative agreement (NA), where PA and NA are the symmetric counterparts of the positive percent agreement and negative percent agreement as defined by the Food and Drug Administration.The inter-pathologist variability will be quantified by way of the ICC when dealing with continuous CPS and TAP scores and with the FKS when dealing with dichotomized PD-L1 statuses. For each combination of scoring system (TAP and CPS) and PD-L1 IHC assay (22C3, SP263 and 28-8), the continuous or dichotomized scores will be summarized allowing to assess whether the pathologist bias is comparable among scoring systems and IHC assays. The relation between the cut-off value and the amount of inter-pathologist variation will be demonstrated for the dichotomized scores.4.2.3Exploratory endpointsThe co-expression of PD-L1 with other biomarkers of interest, including HER2, CLDN18.2 and FGFR2b, will be tabulated. Co-expression will be tested using multiple IHC with a panel of various markers, including, but not limited to, CD45, CD8, CD4, CD68, FGFR, CLDN18.2, CEACAM5, TROP2, HER2, EBER, PanCK and SMA.To estimate the feasibility to measure the absolute performance among the IHC assays: differences in CPS and TAP scores of positive and negative control cell lines stained with PD-L1 22C3, SP263 and 28-8 antibodies will also be compared. If sufficient data are available, PA and NA will be calculated and displayed alongside confusion matrices, providing information on whether the higher degree of staining of some of the IHC assays correspond to a higher PA or a lower NA.4.3Clinical data collectionThe following data will be collected at study entry, including but not limited to:•Demographic data: age and sex;•Basic relevant laboratory assessment, if available;•Primary disease information: date of diagnosis, primary tumour location (gastric versus gastro-oesophageal junction), histology, staging according to TNM (tumour–node–metastasis) based on American Joint Committee on Cancer latest recommendations;•Results of locally carried out tumour biomarker analyses, if available (e.g. HER2, microsatellite status, others according to local procedures, such as CLDN18.2 and FGFR2b);•Type of treatment received and setting, if available, including specifying if they included immunotherapy; and•Type of tumour specimen (from primary tumour versus metastatic sites; biopsy versus resection). Sites will be encouraged to prioritize biopsies from primary tumours as the predominant source of samples. However, samples from surgical resection sites and/or metastatic tumour biopsies will also be included to keep the study design pragmatic and real world. Minimally 30% of samples will be from primary tumour endoscopic biopsies and not more than 10% of samples from metastatic sites.Trial registrationEORTC-RP-2321EthicsThe study is conducted in adherence to the latest Declaration of Helsinki and Ethical Guidelines for Clinical Studies. The study protocol has received approval from the independent ethical review committee of EORTC and has been submitted to the local ethical committees. Before enrolment, written informed consent will be obtained from all patients. In specific sites, patients who have passed away may have sample access with a waiver of consent after local ethics approval.Value of the trial in progress•The ASPIRE study is the first global study evaluating the interchangeability of PD-L1 scoring systems and assays in GC.•The study will provide data to harmonize the quantification and qualification of PD-L1 expression in GC.•The results of the trial will help better select patients who derive benefit when treated with immunotherapy. •The ASPIRE study is the first global