T cell receptor repertoire profiling predicts the prognosis of HBV-associated hepatocellular carcinoma

Abstract

Tumor-infiltrating T cell repertoire has been demonstrated to be closely associated with anti-tumor immune response. However, the relationship between T cell repertoire in tumor tissue and prognosis has never been reported in Hepatocellular carcinoma (HCC). We performed the high-throughput T cell receptor (TCR) sequencing to systematically characterize the infiltrating T cell repertoires of tumor and matched adjacent normal tissues from 23 HBV-associated HCC patients. Significant differences on usage frequencies of some Vβ, Jβ, and Vβ-Jβ paired genes have been found between the 2 groups of tissue samples, but no significant difference of TCR repertoire diversity could be found. Interestingly, the similarity of TCR repertoires between paired samples or the TNM stage alone could not be helpful to evaluate the prognosis of patients very well, but their combination could serve as an efficient prognostic indicator that the patients with early stage and high similarity showed a better prognosis. This is the first attempt to assess the potential value of TCR repertoire in HCC prognosis, and our findings could serve as a complement for the characterization of TCR repertoire in HCC.

Keywords: TCR repertoire; hepatocellular carcinoma; high throughput sequencing; infiltrating T lymphocyte; prognosis.

Figure 1

The usage of Vβ, Jβ, and Vβ‐Jβ paired genes in tumor tissues and paired adjacent normal tissues. A, Heat map of the usage frequency of Vβ and Jβ genes. *P < .05 and **P < .01 by two‐tailed paired t‐tests. Several Vβ or Jβ usage frequencies were significantly higher (red star) or lower (black star) in tumor tissue than paired adjacent normal tissue. B, The Vβ‐Jβ paired genes with significantly difference on usage frequency between paired tissues. Fold change was calculated by the usage frequency in tumor tissue divided by the usage frequency in paired adjacent normal tissue. T, tumor tissue; NT, paired adjacent normal tissue

Figure 2

Comparison of the TCR repertoire diversity indexes between tumor tissues and paired adjacent normal tissues. A, U/T index. B, Clonality index. C, The percentage of highly expanded clones. D, The cumulative percentage of the most 100 frequent TCRβ sequences in each sample. Differences between groups were compared using Wilcoxon matched pairs test. The horizontal line means the median. T, tumor tissue; NT, paired adjacent normal tissue

Figure 3

Analysis of the relation between TCR repertoire and clinical data. A, Comparison of TCR repertoire diversity in tumor tissues among different groups. B, Comparison of TCR repertoire similarity of paired samples among different groups. TCR repertoire overlap between paired samples was calculated by overlap. C, Heat map associations of TCR repertoire similarity and clinical information in 23 patients. TNM stages (I, n = 8; II, n = 7; III, n = 8). Disease progress group (n = 8) and non‐progress group (n = 15). ns indicates not significant difference

Figure 4

Analysis of the relation between TCR repertoire similarity and clinical data. A, The PFS curves of the high similarity group (n = 12) and the low similarity group (n = 11). B, The PFS curves of the different TNM stage groups (I, n = 8; II, n = 7; III, n = 8). C, The PFS curves of the 1 and 2 prognostic score group (n = 12) and the 3 and 4 prognostic score group (n = 11). D, ROC analysis for the prognostic score and the AUC value was 0.842 relative to 0.750 for TNM stage