In search of predictive biomarkers in cancer immunotherapy – The importance of analytical validation

Datum: 24 april 2019

Locatie: Campus Drie Eiken, Gebouw O, Aula O5 - Universiteitsplein 1 - 2610 Antwerpen (Wilrijk)

Tijdstip: 17 - 19 uur

Promovendus: Kelly Schats

Promotor: Ingrid De Meester - Marc Kockx

Korte beschrijving: Doctoraatsverdediging Kelly Schats - Departement Farmaceutische Wetenschappen


During the course of the past years new insights in the importance of the immune system in the fight against cancer led to the new area of immunotherapy in the cancer treatment, with positive results in different tumor indications. Unfortunately, immunotherapy does not show positive effects in all patients/tumor entities. This highlights the need for predictive biomarkers allowing patient selection per treatment, which serves both the patient (no unnecessary exposure to drug toxicity) and society (high costs).

The current doctoral thesis highlights the importance of the analytical validation of biomarker tests in both immunohistochemistry (IHC) and mRNA analysis based assays. The programmed cell death ligand 1 (PD-L1) IHC assays and the PanCancer Immune profiling (PCI) panel from Nanostring are studied as an example. Moreover, the challenges for both approaches (respectively discrepancies in staining patterns and different normalization methods) are investigated in detail.

Robust and reproducible results are obtained for 6 different PD-L1 assays as well as for the validation of the PCI panel (Nanostring).

Furthermore, comparison of the 2 laboratory developed test (LDT) PD-L1 IHC assays as well as comparison of the 4 commercial PD-L1 IHC assays on tonsil and tumor tissue showed discrepancies in the PD-L1 staining pattern, mainly for the SP142 assay. Focusing on the commercial PD-L1 IHC assays, the detection system and the epitope of the primary antibodies were found as potential parameters influencing the staining pattern of the PD-L1 IHC assays. Despite the fact that further research is needed, the current results already show that the diagnostic labs should be aware of the impact of the detection system on the performance of the IHC assay. This is of importance when they develop LDT assay to be used patient selection instead of the recommended PD-L1 assays.

Two different published gene signatures were applied on 2 pilot studies in melanoma patients treated with immunotherapy to illustrate the challenges of interpretation and the influence of the different normalization methods. Further evaluation of the genes as single marker revealed some promising genes, irrespective of the applied normalization method, which should be investigated in more detail.

As second part of the multi-analyte testing, protein expression was evaluated using IHC. The most promising results were observed in the Trimixipi study, where the highest AUC values could be observed for the four immune cell characterization markers (CD8, CD20, CD3/FOXP3) together with PD-L1.