Glucose-6-phosphate dehydrogenase correlates with tumor immune activity and programmed death ligand-1 expression in Merkel cell carcinoma

Abstract

Background: Merkel cell carcinoma (MCC) is a rare and highly malignant skin cancer. Some cases have a good prognosis and spontaneous regression can occur. Reported prognostic markers, such as Merkel cell polyoma virus infection or programmed death ligand-1 (PD-L1) expression, remain insufficient for precisely estimating the vastly different patient outcomes. We performed RNA sequencing to evaluate the immune response and comprehensively estimate prognostic values of immunogenic factors in patients with MCC.

Methods: We collected 90 specimens from 71 patients and 53 blood serum samples from 21 patients with MCC at 10 facilities. The mRNA was extracted from formalin-fixed paraffin-embedded tissues. Next-generation sequencing, immunohistochemical staining and blood serum tests were performed.

Results: Next-generation sequencing results classified MCC samples into two types: the 'immune active type' was associated with better clinical outcomes than the 'cell division type'. Expression of the glucose-6-phosphate dehydrogenase (G6PD) gene was highly significantly upregulated in the 'cell division type'. Among 395 genes, G6PD expression correlated with the presence of lymph node or distant metastases during the disease course and significantly negatively correlated with PD-L1 expression. Immunohistochemical staining of G6PD also correlated with disease-specific survival and exhibited less heterogeneity compared with PD-L1 expression. G6PD activity could be measured by a blood serum test. The detection values significantly increased as the cancer stage progressed and significantly decreased after treatment.

Conclusions: G6PD expression was an immunohistochemically and serum-detectable prognostic marker that negatively correlated with immune activity and PD-L1 levels, and could be used to predict the immunotherapy response.

Keywords: immunotherapy; programmed cell death 1 receptor; skin neoplasms; tumor biomarkers.

Figure 1

Gene expression heatmap of all 41 samples of MCC. Samples are divided into two groups based on the expression of 395 genes. Group A comprised 23 samples and group B comprised 18 samples (A). Kaplan-Meier survival analysis showed a significantly poorer prognosis for patients in group A compared with patients in group B (p=0.035, log-rank test) (B). Gene expression levels in each group are presented as volcano plots. Vertical and horizontal broken lines represent threshold of log2 fold-change (−0.5 and +0.5) and p value (1.0×10^–5) (C). MCC, merkel cell carcinoma.

Figure 2

The top graph is the highest ranked and only one gene set had a p<0.05 in group A. The bottom three graphs are the top three ranked gene sets with smallest normalized enrichment scores (high relative expression in the group (B) with a p <0.01. Graphs display the enrichment score (y axis) versus the gene rank in an ordered dataset (x axis); genes with high relative expression in group A were given a low rank order value (leftmost tail), and genes with high relative expression in group B were given a high rank order value (rightmost tail) of the gene rank representation. Normalized enrichment scores, p values and FDR q values for each analysis are shown (A). Heatmap of analyzed genes constituting the top-ranked gene set for group A, ‘cell division’ (B). Heatmap of analyzed genes constituting the top-ranked gene set for group B, ‘T cell receptor signaling pathway’ (C). FDR, false discovery rate.

Figure 3

Merkel cell polyoma virus (MCPyV) infection, positive (n=24) versus negative (n=10) (A). Spontaneous regression after biopsy, positive (n=4) versus negative (n=29) (B). Lymph nodes or distant metastases during follow-up, positive (n=17) vs negative (n=17) (C). Programmed death ligand 1 (PD-L1) expression in tumor cells, high (n=18) vs low (n=10) (D). Red dots indicate significantly upregulated genes with p value <1.0×10^-5 in group A ‘cell division type’ and blue dots indicate significantly upregulated genes with p value <1.0×10^-5 in group B ‘immune active type’. The vertical and horizontal broken lines represent threshold of log2 fold-change (−0.5 and +0.5) and p value (0.05).

Figure 4

Kaplan-Meier survival curves comparing MCC with high and low G6PD expression classified on the basis of RNA expression (counts per million; CPM) using log-rank test (p=0.027, n=40) (A). Kaplan-Meier survival curves comparing MCC with high and low G6PD expression classified on the basis of immunohistochemistry using log-rank test (p=0.034, n=79) (B). Representative immunohistochemical staining in samples with high G6PD expression. This case showed spontaneous regression after biopsy and recurrent distant metastases after 10 months (C). Representative immunohistochemical staining samples with low G6PD expression. This case showed spontaneous regression after biopsy and no recurrence (D). Low PD-L1 and G6PD expression in a primary lesion of the same Case shown in figure 4F (E). Upregulated PD-L1 expression and stable low G6PD expression in the skin metastatic lesion of the same case shown in figure 4E (F).

Figure 5

Serum G6PD activity significantly increased as the tumor stage progressed in samples obtained from pretreated patients (n=19, p=0.0064, Kruskal-Wallis test) (A). Serum G6PD activity decreased after treatment including surgery, radiation, or both (n=8 paired, p=0.030, paired-t test) (B). Representative data of the G6PD activity change in cases responding to avelumab. Serum G6PD activity decreased during treatment (C). Representative data of G6PD activity changes in cases not responding to avelumab. Serum G6PD activity increased during administration of the ICI (D). CBDCA, carboplatin; ICI, immune checkpoint inhibitor; IMRT, intensity modulated radiation therapy; VP-16, etoposide; XRT, X-ray radiation therap.