Squamous cell carcinomas (SCCs) originate from the epithelial tissues of the aerodigestive or genitourinary tracts. They often occur in the head and neck SCC (HNSC), esophageal SCC (ESCC), lung SCC (LUSC), and cervical SCC (CSCC), sharing common histological features and some risk factors. Among risk factors, tobacco smoking is an established risk factor for SCC, which can alter biological carcinogenesis pathways to promote cancer progression (Schuller, 2019; Sabbula and Anjum, 2020). Cigarette smoke contains many carcinogens that can cause genomic alteration and break immunologic homeostasis (Li et al., 2018). Many studies have reported that smoking cessation can eliminate the physiological driving force of cancer development and improve the prognosis of SCC patients (Dobson Amato et al., 2015; Yavorski and Blanck, 2016; Yang et al., 2020). However, the underlying mechanism of smoking-induced SCC remains unclear in smoking-related SCC.

Homologous recombination repair (HRR) is an important repair method for DNA double strand damage (Mackenroth and Alani, 2020). Cancer cells with homologous recombination deficiency (HRD) are sensitive to platinum drugs and poly-ADP-ribose polymerase (PARP) inhibitors. At present, BRCA1/2 mutation is the most comprehensive HRD biomarker (Hoppe et al., 2018). However, Turner and Ashworth proposed the concept of “BRCAness” to describe HRD without BRCA mutation but with a phenotype similar to BRCA mutation (Turner et al., 2004; Lord and Ashworth, 2016). Therefore, better biomarkers should be identified to screen a more effective platinum/PARP population. HRD scores, which are comprehensively calculated based on loss of heterozygosis (LOH), telomeric-allelic imbalance (TAI), and large-scale state transitions (LST), are considered as biomarkers of genomic instability with mutation (Takaya et al., 2020), which were applied in drug efficacy and tumor susceptibility evaluation (do Canto et al., 2019; Min et al., 2020).

In the present, we not only estimated the association between smoking cessation and smoking-related SCC prognosis but also investigated its potential mechanism. Importantly, we constructed a risk score model combining smoking and HRD, which could effectively screen suitable populations and improve prognostic prediction, especially for smoking-related SCC.

Tobacco smoking has been confirmed to be a critical risk factor for SCC development, and smoking cessation can increase overall survival after diagnosis (Dobson Amato et al., 2015; Yavorski and Blanck, 2016). To further understand the benefits of quitting smoking for patients’ prognosis, we selected smoking-related SCC patients from the TCGA database.

By plotting K-M curves, we found that current smokers had a worse prognosis than reformed smokers in most SCC cohorts, which was consistent with other studies (Li et al., 2020; Yang et al., 2020). Then, we found smoking-related genes and performed function analysis in LUSC, pointing out that HR was an important pathway of smoking to aggravate cancer. It is known that DNA double-strand breaks induced by cigarette smoke can be repaired mainly through HR repair (Helleday, 2003; Albino et al., 2004). Many genes involved in HR repair are significantly associated with smoking to influence the risk of lung cancer (Ryk et al., 2006; Nogueira et al., 2010). Hammouz et al. (2020) also constructed a gene co-expression network and found that smoking significantly affected HR to induce lung adenocarcinoma. In addition, many scholars have used HRD scores in cancer studies. Takaya et al. (2020) analyzed the association between HRD score and high-grade serous ovarian carcinoma and suggested that patients could be classified into different prognostic subtypes for personalized treatment. Kraya et al. (2019) investigated the molecular features of BRCA1/2 alterations in breast cancer patients and found that HRD scores and hormone receptor subtypes could predict the immunogenicity of BRCA1/2 breast cancer, and provided the basis for formulating the best immunotherapy strategy. In the present study, we also found that high HRD scores could increase survival time, suggesting that the cross-talk between smoking cessation and high HRD scores might help treatment and improve overall survival.

After the univariate and multivariate Cox regression analysis, the risk score model was established using five smoking- and HRD-related genes in the LUSC cohort, including LMBRD1, MAFK, MESDC1, KLHL15, and E2F4. LMBR1 domain containing 1 (LMBRD1) encodes a lysosomal membrane protein that may be involved in the transport and metabolism of cobalamin. Mutations of LMBRD1 are also associated with vitamin B12 metabolism disorder (Fettelschoss et al., 2017). MAF bZIP transcription factor K (MAFK), an important transcription factor of the MAF family, is associated with epithelial–mesenchymal transition and malignant progression in different cancers (Wang et al., 2015; Okita et al., 2017). Mesoderm development candidate 1 (MESDC1), also known as TLNRD1, is confirmed as an oncogenic function in bladder cancer and hepatocellular carcinoma (Tatarano et al., 2012; Wu et al., 2017). The kelch-like family member 15 (KLHL15) can encode a member of the kelch-like family of proteins and is involved in protein ubiquitination and cytoskeletal organization (Ferretti et al., 2016; Zhou et al., 2019), but few studies focus on cancers. E2F transcription factor 4 (E2F4), a member of the E2F family of transcription factors, plays an important role in inhibiting proliferation-associated genes, and its gene mutation and increased expression are related to different cancers (Gong et al., 2020; Zhuang et al., 2020). In the future, these genes should be deeply investigated roles in SCC. Afterward, we performed K-M and survival-ROC analysis to estimate the predictive value of the risk score model, suggesting that the model was a predictor with good sensitivity and specificity in LUSC patients.

To better predict the prognosis of each patient, we constructed a nomogram integrating risk scores, smoking status, and clinical information. As a practical tool to improve predictive accuracy, these methods have been applied in many studies on different cancers (Sun et al., 2020; Yang et al., 2020). The nomogram with different aspects of markers, including risk scores, smoking status, age, gender, and tumor stage, had high clinical application value, which might be a promising way to change clinical management (Birkhahn et al., 2007).

Lastly, we validated the risk score model in other SCC cohorts from TCGA and GEO databases, indicating that the application of risk scores could be extrapolated to CSCC and HNSC patients.

The present study is of great significance in theory and application. First, we demonstrated that keeping smoking after diagnosis could increase the risk of SCC death, suggesting that smoking cessation should be a part of cancer treatment. Second, the risk score model was constructed using smoking- and HRD-related genes, which could make up for the deficiency of HRD based on BRCA (Turner et al., 2004; Lord and Ashworth, 2016), more effective in screening the population suitable for targeted therapy. In the future, we should further verify the application of the risk score model in different SCC cohorts, especially in Asian or Chinese populations.

The present study provided comprehensive insights into the association among smoking status, HRD, and prognosis in SCC. Moreover, a risk score model integrating smoking and HRD was constructed to serve as the potential predictive biomarker and add effectiveness in screening the suitable population for targeted therapy.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

CC, XC, and SL conceived and designed the study, performed the bioinformatic analysis, and wrote the manuscript. HC, MC, LB, and HJ contributed to the revision of the manuscript draft. All authors read and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.