Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2022 Nov 15;14(22):5605.
doi: 10.3390/cancers14225605.

Genomic and Transcriptomic Predictors of Response to Immune Checkpoint Inhibitors in Melanoma Patients: A Machine Learning Approach

Yaman B Ahmed  1 , Ayah N Al-Bzour  1 , Obada E Ababneh  1 , Hassan M Abushukair  1 , Anwaar Saeed  2   3
Affiliations

Genomic and Transcriptomic Predictors of Response to Immune Checkpoint Inhibitors in Melanoma Patients: A Machine Learning Approach

Yaman B Ahmed et al. Cancers (Basel). .

Abstract

Immune checkpoint inhibitors (ICIs) became one of the most revolutionary cancer treatments, especially in melanoma. While they have been proven to prolong survival with lesser side effects compared to chemotherapy, the accurate prediction of response remains to be an unmet gap. Thus, we aim to identify accurate clinical and transcriptomic biomarkers for ICI response in melanoma. We also provide mechanistic insight into how high-performing markers impose their effect on the tumor microenvironment (TME). Clinical and transcriptomic data were retrieved from melanoma studies administering ICIs from cBioportal and GEO databases. Four machine learning models were developed using random-forest classification (RFC) entailing clinical and genomic features (RFC7), differentially expressed genes (DEGs, RFC-Seq), survival-related DEGs (RFC-Surv) and a combination model. The xCELL algorithm was used to investigate the TME. A total of 212 ICI-treated melanoma patients were identified. All models achieved a high area under the curve (AUC) and bootstrap estimate (RFC7: 0.71, 0.74; RFC-Seq: 0.87, 0.75; RFC-Surv: 0.76, 0.76, respectively). Tumor mutation burden, GSTA3, and VNN2 were the highest contributing features. Tumor infiltration analyses revealed a direct correlation between upregulated genes and CD8+, CD4+ T cells, and B cells and inversely correlated with myeloid-derived suppressor cells. Our findings confirmed the accuracy of several genomic, clinical, and transcriptomic-based RFC models, that could further support the use of TMB in predicting response to ICIs. Novel genes (GSTA3 and VNN2) were identified through RFC-seq and RFC-surv models that could serve as genomic biomarkers after robust validation.

Keywords: immune checkpoint inhibitors; machine learning; melanoma; tumor mutational burden.

PubMed Disclaimer

Conflict of interest statement

A.S. reports research grants (to institution) from AstraZeneca, Bristol Myers Squibb, Merck, Clovis, Exelixis, Actuate therapeutics, Incyte Corporation, Daiichi Sankyo, Five prime therapeutics, Amgen, Innovent biologics, Dragonfly therapeutics, KAHR medical, Biontech, and advisory board fees from AstraZeneca, Bristol Myers Squibb, Exelixis, Pfizer, and Daiichi Sankyo. The remaining authors have no relevant financial interests to disclose.

Figures

Figure 1
Figure 1
Workflow of the study. Four RFC models were built, a clinical and genomic model (RFC7) based on 3 cBioPortal cohorts, transcriptomic model (RFC-Seq) based on RNA-Seq in GSE91061, RFC-Surv based on survival-related genes and RFC16 based on genes in RFC-Seq and RFC-Surv models. DEGs of GSE91061 based on ICI response were further used to understand TME, GO, and to build the RFC-Seq model. The intersected genes between the top 100 immune-related genes identified via SVM-RFE and survival-related genes from SKCM-TCGA were used to identify prognostic role of these genes using Cox proportional hazard model and to build RFC-Surv model to predict response. Based on genes used in RFC-Seq and RFC-Surv, RFC16 was built to predict response to ICI. ICI: immune checkpoint inhibitors, DEGs: differential expressed genes, SVM-RFE: Support Vector Machine—Recursive Feature Elimination, SKCM-TCGA: The Cancer Genome Atlas- skin cutaneous melanoma, AUC: area under the curve.
Figure 2
Figure 2
ICI prediction models: (A) Feature contribution of RFC7 model showing TMB as the strongest predictor of response. (B) Feature contribution of RFC-Seq model showing GSTA3 as the strongest predictor of response. (C) Feature contribution of RFC-Surv model showing VNN2 as the strongest predictor of response. (D) Venn diagram showing the intersection between TCGA−SKCM Survival−related Genes and SVM−RFE Top 100 DEGs ICI−predictors. (E) The ROC curve of the top associated features from RFC7, RFC-Seq and RFC-Surv models. (F) The ROC curve of RFC7, RFC-Seq and RFC-Surv models.
Figure 3
Figure 3
DEGs and the landscape of TME: (A) Volcano plot showing the DEGs based on response to ICI in GSE91061 as blue dots denote upregulated genes (log2FC > 0.5) while red dots denote downregulated genes (log2FC < −0.5). (B) Dendrogram of the DEGs and their association with the TME.
Figure 3
Figure 3
DEGs and the landscape of TME: (A) Volcano plot showing the DEGs based on response to ICI in GSE91061 as blue dots denote upregulated genes (log2FC > 0.5) while red dots denote downregulated genes (log2FC < −0.5). (B) Dendrogram of the DEGs and their association with the TME.
Figure 4
Figure 4
Heatmap of top DEGs and TME based on response to ICI. Top rows show the top 4 upregulated (IGHD, SIGLEC8, CLDN14, and CLU) and the top 4 downregulated (ANXA3, GSTA3, EHF, and CORO2B) genes based on SVM-RFE in the responders and non-responder patients’ groups. Rest of the rows show the landscape of TME.
Figure 5
Figure 5
Functional enrichment analysis of the 1066 DEGs: (A) GO for the upregulated DEGs. (B) GO for the downregulated DEGs.
Figure 6
Figure 6
Overall survival value of survival-related genes and RFC16 model: (A) Kaplan–Meier plots of overall survival related-genes (CCL5, IL4I1, VNN2, PARP15, LCK, ZNF831, CD86, and FCRRL6). All the genes were significant based on best cutoff value. (B) Cox proportional hazard model of the 8 genes. None of the genes showed survival value in the multivariate cox proportional hazard model. (C) Feature contribution of RFC16 model.
Figure 6
Figure 6
Overall survival value of survival-related genes and RFC16 model: (A) Kaplan–Meier plots of overall survival related-genes (CCL5, IL4I1, VNN2, PARP15, LCK, ZNF831, CD86, and FCRRL6). All the genes were significant based on best cutoff value. (B) Cox proportional hazard model of the 8 genes. None of the genes showed survival value in the multivariate cox proportional hazard model. (C) Feature contribution of RFC16 model.
Figure 7
Figure 7
The expression of survival-related genes between responders and non-responders.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

References

    1. Madden K., Kasler M.K. Immune Checkpoint Inhibitors in Lung Cancer and Melanoma. Semin. Oncol. Nurs. 2019;35:683432. doi: 10.1016/j.soncn.2019.08.011. - DOI - PubMed
    1. Olbryt M., Rajczykowski M., Widłak W. Biological Factors behind Melanoma Response to Immune Checkpoint Inhibitors. Int. J. Mol. Sci. 2020;21:4071. doi: 10.3390/ijms21114071. - DOI - PMC - PubMed
    1. Gajic Z.Z., Deshpande A., Legut M., Imieliński M., Sanjana N.E. Recurrent somatic mutations as predictors of immunotherapy response. Nat. Commun. 2022;13:3938. doi: 10.1038/s41467-022-31055-3. - DOI - PMC - PubMed
    1. Taube J.M., Klein A., Brahmer J.R., Xu H., Pan X., Kim J.H., Chen L., Pardoll D.M., Topalian S.L., Anders R.A. Association of PD-1, PD-1 ligands, and other features of the tumor immune microenvironment with response to anti-PD-1 therapy. Clin. Cancer Res. 2014;20:5064–5074. doi: 10.1158/1078-0432.CCR-13-3271. - DOI - PMC - PubMed
    1. Binnewies M., Roberts E.W., Kersten K., Chan V., Fearon D.F., Merad M., Coussens L.M., Gabrilovich D.I., Ostrand-Rosenberg S., Hedrick C.C., et al. Understanding the tumor immune microenvironment (TIME) for effective therapy. Nat. Med. 2018;24:541–550. doi: 10.1038/s41591-018-0014-x. - DOI - PMC - PubMed