Association of Cell Death Markers With Tumor Immune Cell Infiltrates After Chemo-Radiation in Cervical Cancer

Abstract

Irradiation induces distinct cellular responses such as apoptosis, necroptosis, iron-dependent cell death (a feature of ferroptosis), senescence, and mitotic catastrophe. Several of these outcomes are immunostimulatory and may represent a potential for immunogenic type of cell death (ICD) induced by radiotherapy triggering abscopal effects. The purpose of this study is to determine whether intra-tumoral ICD markers can serve as biomarkers for the prediction of patient's outcomes defined as the metastasis status and survival over a 5-year period. Thirty-eight patients with locally advanced cervical cancer, treated with neoadjuvant chemoradiotherapy using cisplatin were included in this study. Pre-treatment tumor biopsy and post-treatment hysterectomy samples were stained for cell death markers and danger associated molecular patterns (DAMPs): cleaved caspase-3 (apoptosis), phosphorylated mixed lineage kinase domain like pseudokinase (pMLKL; necroptosis), glutathione peroxidase 4 (GPX4; ferroptosis) and 4-hydroxy-2-noneal (4-HNE; ferroptosis), high mobility group box 1 (HMGB1) and calreticulin. Although these markers could not predict the patient's outcome in terms of relapse or survival, many significantly correlated with immune cell infiltration. For instance, inducing ferroptosis post-treatment seems to negatively impact immune cell recruitment. Measuring ICD markers could reflect the impact of treatment on the tumor microenvironment with regard to immune cell recruitment and infiltration.

One sentence summary: Cell death readouts during neoadjuvant chemoradiation in cervical cancer.

Keywords: biomarkers; cell death; cervical cancer; immunogenic cell death; tumor infiltrating leucocytes.

Figure 1

A scheme of the protocol to collect and stain tissue samples is shown. For each patient, a tissue block of the pre-treatment biopsy and post-treatment hysterectomy samples were obtained. Immunohistrochemistry markers were used for cell death and DAMPs (cleaved caspase-3 and pMLKL, GPX4 and 4-HNE, HMGB1 and calreticulin). A previous study assessed the tumor infiltrating leucocytes (CD3, CD4, CD8, CD20, CD68, Iba1, CD163, FoxP3, PD-L1 and IL-33). Patients were followed-up up to 5 years post-treatment, period in which survival and relapse were assessed.

Figure 2

Representative immunohistochemical (IHC) staining of pre-treatment biopsy and post-treatment hysterectomy samples for the different cell death markers (A) and DAMPs (B). Bright field images were taken with ZEISS Axio Scan.Z1 at a magnification of 20x. Scales bars are 100 µm as indicated on the images. One example of isotype control is also shown in this figure. For the pre-treatment biopsy, the percentages of positive cells per tissue slice are the following: cleaved caspase-3 (6,77%), pMLKL (53,12%), GPX4 (53,9%), 4-HNE (0,73%), calreticulin (7,24%), HMGB1 (66,37%). For the post-treatment hysterectomy, the percentages of positive cells per tissue slice are the following: cleaved caspase-3 (1,66%), pMLKL (31,42%), GPX4 (15,5%), 4-HNE (0,86%), calreticulin (52,10%), HMGB1 (34,53%).

Figure 3

Statistically significant correlation in the biopsy samples. Statistically significant correlations were found between cleaved caspase-3 and CD3 (A), CD8 (B), CD3 (C), CD8 (D), CD20 (E), CD163 (F); calreticulin and PD-L1 (G) and Iba1 (H); HMGB1 and Iba1 (I); cleaved caspase-3 and pMLK (J); HMGB1 and calreticulin (K); 4-HNE and cleaved caspase-3 (L). All p-values and correlation coefficients were evaluated with two-sided test of correlation (Genstat 64-bit Release 20.1) are indicated on the graphics which were generated with GraphPad (version 8).

Figure 4

Statistically significant correlation in the hysterectomy samples, treatment-induced and biopsy samples vs hysterectomy samples. Statistically significant correlations were found post-treatment between GPX4 and PD-L1 (A). Correlation in the treatment-induced analysis between PD-L1 and cleaved caspase-3 (B) and 4-HNE and PD-L1 (C). Correlations between the biopsy samples vs hysterectomy samples in all patients are significant for pre-calreticulin and post-calreticulin (D); pre-calreticulin and post-HMGB1 (E); pre-GPX4 and post-FoxP3 (F); pre-HMGB1 and post-HMGB1 (G) and post-4-HNE and post-IL-33 (H). All p-values and correlation coefficients were evaluated with two-sided test of correlation (Genstat 64-bit Release 20.1) are indicated on the graphics which were generated with GraphPad (version 8).

Figure 5

Statistically significant correlations in the post-treatment samples and treatment-induced analysis for patients with an incomplete response. Statistically significant correlations were found between pMLKL and Iba1 in the post-treatment samples (A). Correlations between the treatment-induced levels of pMLKL and CD3 (B), GPX4 and CD3 (C); 4-HNE and CD20 (D) and FoxP3 (E) and pMLKL and 4-HNE (F) were also statistically significant. All p-values and correlation coefficients were evaluated with two-sided test of correlation (Genstat 64-bit Release 20.1) are indicated on the graphics which were generated with GraphPad (version 8).