Background
Eligibility for afami-cel is determined by germline HLA-A*02 genotype and expression of MAGE-A4 in tumor cells as determined by immuno-histochemistry (IHC). Biomarker-based patient selection in a single arm clinical trial necessitates the investigation of the potential bias introduced by these biomarkers in regard to the clinical outcome of the study population compared to the overall patient population. A real word evidence study based on the data mining of a large database of Synovial Sarcoma (SyS) patients with available tumor samples and well documented clinical data (such as overall survival) was conducted with Centre Léon Bérard hospital (CLB) in Lyon, France, to assess the prognostic value of HLA-A*02 and MAGE-A4 expression in SyS. Here, we describe the validation of the MAGE-A4 IHC Clinical Trial Assay (CTA) and the development of an equivalent IHC assay used to determine MAGE-A4 expression in these archival samples.
Methods
A MAGE-A4 CTA was developed on the DAKO Link 48 platform and validated to screen patients for tumor expression of MAGE-A4. The positivity cutoff for enrollment in the phase I study (NCT03132922) and the pivotal phase II study (NCT04044768) was defined as ≥30% of tumor cells stained at ≥2+. A prognostic study protocol was designed and carried out at CLB with IRB approval and in accordance with the Declaration of Helsinki. Restrictions in international sample transfer required the development of an IHC assay at CLB matching the performance of the CTA. Assay transfer to CLB was optimized and validated on Ventana Benchmark Ultra, followed by concordance testing between the two assays. All MAGE-A4 scoring occurred at third party central laboratory by trained pathologists.
Results
CTA was validated at CellCarta (Belgium) by staining 30 SyS samples for prevalence of MAGE-A4. Precision and reproducibility were established by selecting and staining 5 SyS samples from prevalence with a range of MAGE-A4 expression. Assay transfer tested 6 samples, which met acceptance criteria for interlab comparison with the CLB assay by having equivalent staining pattern intensity for all samples. Concordance study composed of 5 SyS samples, passed with unchanged MAGE-A4 status between labs, despite some qualitative differences. Next, CLB intra- and inter-run variability assessed 6 specimens in triplicate. These samples were also scored by third party pathologists and considered valid and acceptable.
Conclusions
The CLB MAGE-A4 immunostaining conditions were successfully adjusted and bridged to reach concordance with the clinical trial assay. This allowed the accurate assessment of the prognostic value of MAGE-A4 in SyS.