Association of immune-related expression profile with sensitivity to chemotherapy in esophageal squamous cell carcinoma

Abstract

Neoadjuvant chemotherapy (NAC) followed by surgery is one of the standard therapeutic approaches in Japan for patients with locally advanced esophageal carcinoma. Recently, the JCOG1109 study revealed that NAC with docetaxel, cisplatin and 5-fluorouracil (5-FU) (DCF-NAC) is superior to NAC with cisplatin and 5-FU, and has now become the standard preoperative chemotherapy. Using a microarray system, we have previously investigated the expression profiles of endoscopic biopsy samples from patients with esophageal squamous cell carcinoma (ESCC) before DCF-NAC (preNAC) and identified 17 molecules as biomarkers predictive of a pathologically complete response to DCF-NAC. Here, we re-grouped our previous dataset based on the histopathological response grade with the addition of several microarray profiles and conducted a re-analysis using bioinformatic web tools including DAVID, GSEA, UALCAN, and CIBERSORTx. We identified 204 genes that were differentially expressed between the highly resistant and sensitive groups. Some of these differentially expressed genes (DEGs) were related to the immune response and showed higher expression in the sensitive group. UALCAN showed that high expression of 28 of the top 50 DEGs was associated with a favorable prognosis (p < 0.25), and that this reached a significant (p < 0.05) level for 18 of them, suggesting that patients with high expression of these genes might have benefited from chemotherapy and thus had a better outcome. In preNAC biopsy tissues from a DCF-sensitive case, we demonstrated the presence of cells expressing mRNA for CXCL9, one of the prognosis-related DEGs. Our results highlight the association of immune-related expression profile in preNAC ESCC with the DCF-NAC efficacy.

Keywords: biomarker; esophageal squamous cell carcinoma; immune-related expression profile; microarray; neoadjuvant chemotherapy.

FIGURE 1

Expression of representative immune‐related DEGs in the highly resistant and sensitive groups. (A) Probe numbers that showed differences in expression between the groups by expression microarray. Genes differentially expressed among the three groups were identified by one‐way ANOVA (p < 0.05) and post hoc testing correction (Tukey). The Benjamini and Hochberg correction was used for false‐positive reduction (multiple testing correction). (B) Normalized expressions of representative immune‐related DEGs in the three groups. Expressions were normalized and log₂‐transformed using GeneSpring. The normalized values are available from GEO under the accession number GSE225178. *, corrected p‐value <0.05.

FIGURE 2

Bioinformatic analysis of gene expression profiles by hierarchical clustering, DAVID, and GSEA. (A) Hierarchical clustering of the non‐tumor, highly resistant and sensitive samples using DEGs based on Euclidean distance. Here, 11 of the 12 highly resistant cases and all of the five non‐tumor samples were clustered on the same branch with less than half of the sensitive cases (11/27). (B) DAVID analysis using DEGs. Among the 204 DEGs including redundant and non‐coding genes, 172 unique genes were automatically selected and used by DAVID analysis. The data from the Functional Annotation Chart with the default setting were modified and are shown. The significance of enrichment for each term is expressed as FDR. Terms and their associated categories are listed according to their FDR. The number and percentage of DEGs related to the terms are indicated in the right bar graph. (C) GSEA with Hallmark gene sets (ver. 7.5) between the highly resistant and sensitive groups with microarray data. The top five upregulated gene sets in the sensitive group, along with the normalized enrichment scores, are shown according to their FDR values. Two of the gene sets, allograft_rejection and interferon_gamma_response, were significantly upregulated in the sensitive group (FDR < 0.05) and are shown as barcode plots and enrichment score. None of the gene sets was significantly upregulated in the highly resistant group.

FIGURE 3

CIBERSORTx estimation of immune cell infiltration. (A) The CIBERSORTx estimated absolute cell fractions of 22 leukocyte types in each sample using microarray data. (B) The total estimated absolute cell fractions in each sample were compared among the three groups. Totals of the estimated infiltrated leukocyte fractions were significantly increased in both the highly resistant and sensitive groups relative to the non‐tumor samples. (C) Estimated fractions of each leukocyte type were compared among the three groups. Immune cell types showing significant difference between fractions of the highly resistant and sensitive groups were indicated by red. The estimated M1 macrophage fraction showed the most significant difference between the highly resistant and sensitive groups. Significance was expressed using Steel–Dwass test. p‐values of less than 0.05 were considered significant and are indicated.

FIGURE 4

Expression of CXCL9 mRNA by microarray and RNAScope in situ hybridization. (A) Normalized values of CXCL9 mRNA expression by microarray analysis. Cases subjected to RNAScope in situ hybridization in (B) are indicated by bold numbers. (B) Distribution of CXCL9 mRNA in the selected sensitive (cases 11 and 33) and highly resistant (cases 20 and 25) cases were investigated by RNAScope in situ hybridization. Sensitive cases showed high expression of CXCL9 mRNA. (C) High‐power views of images in (B). Tumor islets in (C) are indicated in Figure S3.

FIGURE 5

Double staining of CXCL9 and cytokeratin or CD68 by RNAScope and immunohistochemistry in CXCL9‐expressing cases. (A, B) Double staining of CXCL9 mRNA (brown) and cytokeratin (red) using RNAScope and immunohistochemistry in the biopsy sample from Case 11. Yellow arrowheads indicate double‐positive cells. White arrowheads indicate cells positive for CXCL9 but negative for cytokeratin. (C–E). Double staining of CXCL9 mRNA (brown) and CD68 (red), which is a marker for macrophages, in the biopsy samples from cases 11 and 33. Yellow arrowheads indicate CXCL9‐expressing macrophages. White arrowheads indicate cells positive for CXCL9 but negative for CD68 in the biopsy samples from cases 11 and 33. Square area in (E) is shown as (F).