<p>MVD gene score development in preclinical study. <b>A,</b> Six MVD-related genes. <b>B,</b> Expression levels of the six genes in the vascular endothelial cell cluster and other clusters using the scRNA-seq dataset of Hepa 1-6 tumors. <b>C,</b> Relationship between the MVD gene score and MVD(IHC) in treatment-naïve tumors of 12 mouse syngeneic tumor models. <i>r</i> = 0.77 (<i>P</i> = 0.0032) by Pearson correlation coefficient. <b>D,</b> Relationship between the MVD gene score and MVD(IHC) in human samples. <i>r</i> = 0.57 (<i>P</i> = 3.3 × 10<sup>−5</sup>) by Pearson correlation coefficient. MVD(IHC) and the MVD gene score computed from RNA-seq data were compared using commercially available FFPE samples of human tumors. The colors at each point represent the tumor type in the sample, as shown in the legend.</p>
ARTICLE ABSTRACT
Combination therapy with antiangiogenic drugs and immune checkpoint inhibitors has shown enhanced clinical activity and has been approved for the treatment of multiple tumor types. Despite extensive research, predictive biomarkers for combination therapy remain poorly understood. Microvessel density (MVD), a surrogate marker for aberrant angiogenesis measured by IHC, has been associated with response to monotherapy with antiangiogenic inhibitors. However, obtaining tumor tissue with a sufficient mass for IHC analysis is not always practical, and IHC-based MVD measurements are unavailable in large public datasets. In this study, we developed an MVD gene score based on RNA sequencing data that reflects MVD by using RNA sequencing and MVD measured by IHC in 12 mouse syngeneic tumor models. We explored the relationship between the MVD gene score and a gene signature, predicting the response to anti–PD-1 therapy in mouse and human tumor datasets. The MVD gene score correlated with the antitumor activity of lenvatinib, a multiple tyrosine kinase inhibitor mainly targeting VEGFRs and FGFRs, in mouse tumor models, and MVD measured by IHC in commercially available human formalin-fixed, paraffin-embedded tumor samples. Tumor types in The Cancer Genome Atlas were classified into four subgroups based on the MVD gene score and T cell–inflamed gene expression profile, which were correlated with clinical indications for treatment. In conclusion, the newly developed MVD gene score enables the estimation of MVD in large public datasets in which IHC data are unavailable and has potential clinical utility together with the T cell–inflamed gene expression profile to characterize tumors of patients for precision medicine.
A novel gene signature for MVD was developed. This MVD gene score enables the estimation of MVD, reflecting the sensitivity to antiangiogenic inhibitors, in transcriptomic datasets. We demonstrated the utility of the MVD gene score together with a T cell–inflamed gene signature for potential future use as a clinical biomarker.
