Molecular and cellular dynamics of squamous cell carcinomas across tissues

Abstract

Squamous cell carcinomas (SCCs), arising from the skin, head and neck, lungs, esophagus, and cervix, are collectively among the most common cancers and a frequent cause of cancer morbidity and mortality. Despite distinct stratified epithelial tissues of origin, converging evidence points toward shared biologic pathways across SCCs. With recent breakthroughs in molecular technologies have come novel SCC treatment paradigms, including immunotherapies and targeted therapy. This review compares commonalities and differences across SCCs from different anatomical sites, including risk factors and genetics, as well as cellular and molecular programs driving tumorigenesis. We review landmark discoveries of the "cancer stem cells" (CSCs) that initiate and propagate SCCs and their gene and translational regulation programs. This has led to an appreciation that interactions between CSCs and the immune system play key roles in invasion and therapeutic resistance. Here, we review the unifying principles of SCCs that have emerged from these exciting advances in our understanding of these epithelial cancers.

Keywords: cancer epigenetics; cancer genetics; cancer signaling; cancer stem cells; epithelial cancers; squamous cell carcinoma; translational regulation; tumor microenvironment.

Figure 1.

Common mutations in human SCCs versus all cancers. (A) Frequency of mutated genes per each SCC (skin, ESCC, lung, head and neck [HN], and cervical) as compared with all cancers. (Left) Frequent SCC mutations. (Right) Less frequent but still appreciable mutations in SCCs. Median is denoted with a line. Bars indicate range. (B) Most frequently mutated genes across SCCs. Note the low frequency of TP53 mutations in cervical SCC, explained by the high incidence of HPV, where TP53 is inactivated by the E6 viral oncoprotein. Additionally, there is a low frequency of CDKN2A mutations in this cancer, corresponding to a high frequency of RB inactivation by the E7 viral oncoprotein. (C) Mutations in RAS. (Left) A violin plot with the frequency of all SCCs plotted versus all cancers grouped together. (Right) Frequency of different RAS genes in each SCC versus all cancers. (A–C) Mutation data from Memorial Sloan Kettering Integrated Mutation Profiling of Actionable Cancer Targets (MSK-IMPACT). n_samples = 176 skin SCC, n_samples = 189 esophagus (ESCC), n_samples = 1482 lung SCC, n_samples = 788 head and neck (HN SCC), n_samples = 218 cervical SCC, n_samples = 84,225 samples from 72,067 patients (all cancers). Note that all mutations are included. This overestimates driver mutations in UV-induced high TMB cSCCs. For example, if variants of unknown significance (VUSs) are excluded, HRAS mutations are present in 7% rather than 11% of samples. In a more extreme case, ROS1 is mutated in 37% of cSCCs, but <1% have ROS1 driver mutations. Similarly, 5% KRAS mutation in lung SCC is higher than expected. About 10% of these are VUSs. Also, pathology is based on a single biopsy, which could result in contamination of adenosquamous cancer into the lung SCC cohort, further overestimating the frequency of KRAS mutations.

Figure 2.

TP63 and SOX2 amplifications in human SCCs. Data were compiled from publicly available TCGA data on cBioPortal squamous cell carcinoma samples (Cerami et al. 2012; Gao et al. 2013). n_samples = 1833 as follows: n_samples = 297 cervical, n_samples = 227 esophagus, n_samples = 632 head and neck, n_samples = 511 lung, n_samples = 151 skin, n_samples = 15 vaginal.

Figure 3.

A unified perspective on SCCs. (A) SCC anatomical sites. Shown are (clockwise) head and neck (mouth), skin, lung, cervical, and esophagus. (B) Hypothetical PCA accounting for genetics and the molecular and clinical behaviors of SCCs versus other cancers. This illustrates how collectively SCCs are distinct from other cancers. (C) Squamous cell carcinoma engages multiple molecular and cellular pathways for initiation, progression, and therapeutic evasion, as described in the text. Created with BioRender.com.