Cancer specific up-regulated lactate genes associated with immunotherapy resistance in a pan-cancer analysis

Abstract

Background: Although the lactate pathway has been reported to lead to immune escape through the inhibition of effector T cells, the cancer-intrinsic lactate signature has not been identified, and the immunotherapeutic efficacy and potential mechanism of the lactate signature are still unclear.

Methods: We defined a pan-cancer up-lactate score by comparing malignant tissues and normal tissues in the TCGA cohort. The immunotherapeutic efficacy was evaluated in non-small cell lung cancer (NSCLC), metastatic renal cancer (mRCC), bladder cancer (BLCA) and melanoma cohorts. The cancer cell-intrinsic mechanism to immune checkpoint inhibitors (ICIs) resistance was measured using single cell sequencing (scRNA-seq) data. Pathway activation was evaluated in the TCGA cohort and CPTAC cohort with transcriptomics and proteomics. The co-occurrence of up-lactate signature and mTOR signaling was determined by spatial transcriptomics of the tissue samples. Immunotherapy resistance and pathway regulation were validated in the in-house NSCLC cohort.

Results: Patients with the high up-lactate scores had significantly short overall survival (OS) than those with the low up-lactate scores (p < 0.001) across multiple types of cancers. The up-regulated lactate signature exhibited higher expression in the malignant cells compared with stromal cells and immune cells in multiple scRNA-seq datasets. A high up-lactate score was associated with poor OS in NSCLC, mRCC, BLCA and melanoma patients who received anti-PD(L)1 antibody. The up-lactate score was higher in the responders of cancer cells, but not in immune cells and stromal cells compared with the non-responders (p < 0.05). Moreover, up-lactate score was positively correlated with mTOR signaling across multiple cancers. In patients with NSCLC who received anti-PD-1 antibody, higher up-lactate scores were associated with significantly shorter PFS compared to lower up-lactate scores (p < 0.001). Additionally, the up-lactate score was associated with cold tumor, and was positively correlated with mTOR signaling.

Conclusion: Collectively, we defined a pan-cancer up-lactate signature, which is a feature of malignant cells and is associated with ICIs resistance. This reveals a coherent program with prognostic and predictive value that may be therapeutically targeted.

Keywords: Cancer intrinsic feature; Immune checkpoint inhibitor; Immunosuppression; Lactate.

Fig. 1

Flow chart of this study.

Fig. 2

Landscape of lactate pathway in the pan-cancer and the association with the prognosis in TCGA cohort and CPTAC cohort. (A) Boxplots of up-lactate scores and down-lactate scores in multiple cancer types of the TCGA cohort. (B) The intergene Spearman's correlations among up-regulated genes and down-regulated genes of lactate pathway between cancer and normal samples in the TCGA cohort. (D–E) Unadjusted Kaplan-Meier curves showing survival by the quartile of the up-lactate or down-lactate score. P value was calculated by the log-rank test. P value was calculated by the log-rank test.

Fig. 3

Up-regulated lactate signature instead of down-regulated lactate signature was cancer cell origin. (A) Positive or inverse correlations between up-lactate scores or down-lactate scores and purity estimated by ABSOLUTE. (B–C) Positive or inverse correlations between up-lactate scores or down-lactate scores and mean variation allele frequency and total cancer specific mRNA. (D–G) The up-lactate and down-lactate scores in different cell types in the HNSCC dataset, CHOL dataset, NSCLC dataset, LIHC dataset. Note: Different cancer types were marked by different colors for (A–C).

Fig. 4

Up-regulated lactate signature was corrected with immune cold tumor. (A–B) Positive correlations between up-lactate scores and log transformed TMB and predicted neoepitope peptides. Different cancer types were marked by different colors. (C) Bubble plots illustrating the positive correlations between up-lactate scores and immune cells infiltration estimated by TIMER across different cancer types in the TCGA cohort. (D) Heatmap depicting the different signatures and phenotypes (e.g., angiogenesis, fibroblasts, pro-tumor immune infiltrate, anti-tumor immune infiltrate, EMT signature, proliferation, defined by Bagaev et al. [22]) in 1st-3rd quartile (Q1-Q3) groups of up-lactate score. (E) Scatter plots illustrating the positive or inverse correlations of up-lactate scores and immunity signatures defined previously [23]. (F–H) The cell counts of immune cells, stromal cells, and malignant cells in the 1st-3rd quartile (Q1-Q3) groups of up-lactate score in the HNSCC, SKCM, and NSCLC single-cell sequencing datasets. P values were calculated by Fisher exact test.

Fig. 5

Up-regulated lactate signature was associated with immunotherapy failure. (A–D) Kaplan-Meier curves showing survival by the 1st-3rd quartile (Q1-Q3) of up-lactate score in the mRCC, NSCLC, BLCA, and melanoma patients who received anti-PD1 regimens.

Fig. 6

Up-regulated lactate signature of malignant cells was associated with intrinsic immune resistance. (A–B) Bar plots representing the tumor mutational burden (Muts) in different groups stratified by the 1st-3rd quartile (Q1-Q3) of up-lactate scores in the BLCA immunotherapy cohort, with error bars to indicate s.d. (C) The frequency of PD-L1 expression (negative: IC0, moderate: IC1, strong positive: IC2) in different groups stratified by the 1st-3rd quartile (Q1-Q3) of up-lactate scores in the BLCA immunotherapy cohort. (D) The correlations between up-lactate scores and T cell exclusion scores in the malignant cells of the SKCM cohort. P values were compared by Spearman's correlation. (E) The comparisons of up-lactate scores between the immune response and immune resistance group in the malignant cells. P values were compared by Mann-Whitney test. (F–G) The comparisons of up-lactate scores between immune response group and immune resistance group in the different immune cells, and stromal cells. P values were compared by Mann-Whitney test.

Fig. 7

Up-lactate score was associated with immune resistance in in-house cohort. (A) Kaplan-Meier curves showing survival by the 1st-3rd quartile (Q1-Q3) of up-lactate score in NSCLC patients who received anti-PD1 regimens. (B) Bubble plots illustrating the positive correlations between up-lactate scores and immune cells infiltration estimated by CIBERSORT. (C) The correlations between up-lactate scores and mTOR signaling score. P values were compared by Spearman's correlation.