Machine Learning-assisted immunophenotyping of peripheral blood identifies innate immune cells as best predictor of response to induction chemo-immunotherapy in head and neck squamous cell carcinoma - knowledge obtained from the CheckRad-CD8 trial

Abstract

Purpose: Individual prediction of treatment response is crucial for personalized treatment in multimodal approaches against head-and-neck squamous cell carcinoma (HNSCC). So far, no reliable predictive parameters for treatment schemes containing immunotherapy have been identified. This study aims to predict treatment response to induction chemo-immunotherapy based on the peripheral blood immune status in patients with locally advanced HNSCC.

Methods: The peripheral blood immune phenotype was assessed in whole blood samples in patients treated in the phase II CheckRad-CD8 trial as part of the pre-planned translational research program. Blood samples were analyzed by multicolor flow cytometry before (T1) and after (T2) induction chemo-immunotherapy with cisplatin/docetaxel/durvalumab/tremelimumab. Machine Learning techniques were used to predict pathological complete response (pCR) after induction therapy.

Results: The tested classifier methods (LDA, SVM, LR, RF, DT, and XGBoost) allowed a distinct prediction of pCR. Highest accuracy was achieved with a low number of features represented as principal components. Immune parameters obtained from the absolute difference (lT2-T1l) allowed the best prediction of pCR. In general, less than 30 parameters and at most 10 principal components were needed for highly accurate predictions. Across several datasets, cells of the innate immune system such as polymorphonuclear cells, monocytes, and plasmacytoid dendritic cells are most prominent.

Conclusions: Our analyses imply that alterations of the innate immune cell distribution in the peripheral blood following induction chemo-immuno-therapy is highly predictive for pCR in HNSCC.

Keywords: Chemotherapy; HNSCC; Immune phenotyping; Immunotherapy; Induction therapy.

Fig. 1

Study approach. CheckRad-CD8 trial patients underwent immune phenotyping of the peripheral blood before and after chemoimmunotherapy induction yielding data at time point T1 and T2, respectively. Either the data at T1 or T2 alone, or their combination using their (absolute) difference in cell count values were used after preprocessing in machine learning-based classifiers to predict the pathological complete responder event (pCR). We additionally evaluated ensemble solutions to achieve a higher predictive power by combining several “weaker” classifiers.

Fig. 2

Low dimensional feature representations allow accurate pCR prediction. A) Preprocessing pipeline for each dataset generated in Fig. 1. After univariate feature selection, principal component analysis was performed. The newly transformed vectors are used for Machine Learning. B) Classification accuracy heatmap of used features from univariate feature selection against used principal components for |T2-T1|. Red circle indicates maximum accuracy, yellow circle high accuracy with only a single principal component. C) – E) same as panel B) but for T1, T2 and T2-T1, respectively.

Fig. 3

Classifier comparison. The graph shows the relation of incorporated biomarkers (x-axis) and the yielded mean accuracy (y-axis) for a given classifier in relation to before the chemoimmunotherapy induction (left) or after (right). It is worth noting that the overall prediction accuracy is higher after the induction therapy than before.

Fig. 4

Identified peripheral blood immune biomarkers with predictive value. Overlap of biomarkers with high predictive power across the investigated datasets. Notably, especially cells of the innate immune system cells are highly represented.