Effects of tobacco on the DNA of smokers and non-smokers affected by OSCC: systematic review and meta-analysis

Abstract Scientific evidence about genetic and molecular changes in oral squamous cell carcinoma (OSCC) among smokers and non-smokers is inconclusive. This systematic review and meta-analysis assessed the effects of tobacco on the DNA of individuals with OSCC based on protein mutations. Electronic searches were conducted on PubMed, Ovid, Web of Science, and Scopus to identify observational studies published up to January/2022. The Joanna Briggs Institute tool was used for the critical appraisal of studies. The certainty of the evidence was evaluated. Twenty-three studies assessing 4,060 individuals (2,967 smokers vs. 1,093 non-smokers) were included in this review. Fifteen groups of proteins/genes were investigated. Analysis of the quality of articles revealed low risk of bias in most studies. The certainty of the evidence was very low. The meta-analysis confirmed no significant difference between smokers and non-smokers with respect to damage to GSTM1 (OR: 0.60; 95%CI: 0.30–1.18), GSTT1 (OR: 1.18; 95%CI:0.49–2.83), hydrolase proteins (Ku70 and Ku80) (OR: 0.74; 95%CI: 0.18–3.05), and transferase proteins (GSTM1, GSTT1, GSTM3) (OR: 0.74; 95%CI: 0.47–1.18). Most of the studies included showed that smokers are more likely to exhibit genetic instability. However, the meta-analysis revealed that smokers do not necessarily have more genetic alterations in the DNA than non-smokers.


Introduction
3][4] OSCC manifests as an outcome of several biochemical, cellular, and clinical changes in the epithelium of the affected oral mucosa. 5he etiology of OSCC is multifactorial and the main risk factors are tobacco, alcohol, genetic predisposition, biological agents, systemic status, Declaration of Interests: The authors certify that they have no commercial or associative interest that represents a conflict of interest in connection with the manuscript.
carcinoma" OR "oral tumor" OR "mouth tumor" OR "oral tumour" OR "mouth tumour" AND smoking OR smoke OR smoker OR tabagism OR tobacco OR nicotine.

Study selection
The reference were managed using the EndNote X7.4 software (Clarivate Analytics, Toronto, Canada).Duplicates were removed upon identification.After duplicate removal, the titles/abstracts of the retrieved references were assessed by two independent reviewers (L.F.S. and K.S.S.V.).Percent inter-observer agreement was calculated. 18The references whose title/abstract seemed to meet the eligibility criteria were selected for full-text reading.Full text evaluation was also performed by the two reviewers independently.After assessment of the full texts, those that met the eligibility criteria were included in this systematic review and metaanalysis.Disagreements between reviewers were resolved by a third examiner (V.F.B.).

Data extraction
Data were extracted by one author (KSV), and double-checked by a second author (LFS).Disagreements were resolved by discussions, and if needed, another author (VFB) was consulted.The following items were extracted from the articles included in the study: name of author(s), year of publication, country where the study was conducted, study design, overall sample size, participants' sex, number of individuals who were smokers and non-smokers, gene/protein analyzed, method for gene/protein assessment, and main findings.If necessary, contact with authors was made to obtain additional information.

Appraisal of the methodological quality of the included studies
The Critical Appraisal Checklist for cross-sectional studies recommended by the Joanna Briggs Institute of the University of Adelaide was employed. 19The included articles were evaluated according to specific parameters.Two reviewers (L.F.S. and K.S.S.V.) independently evaluated the included studies.For each parameter, the included articles were rated as "low risk of bias", "high risk of bias", "unclear risk of bias", or "not applicable".Any discrepancy between reviewers was resolved by discussion.If necessary, a third examiner (V.F.B.) was consulted.

Synthesis of the results
To illustrate the conducted meta-analysis, a forest plot was provided.

Additional analyses
Four subgroup analyses were conducted: by protein, by gene, by group of proteins/genes assessed, and by method of assessment.In the analyses, dichotomous data (number of smokers with DNA damage among the total number of smokers evaluated and number of non-smokers with DNA damage among the total number of non-smokers evaluated) were used.Comparisons between smokers and non-smokers were carried out.The results are reported as odds ratio (OR) and confidence intervals (CI).Two p values were also reported, one from the chi-square test related to heterogeneity and one from the Z test related to the summary effect, all with the significance level set at p<0.05.

Assessment of the certainty of evidence
The Grading of Recommendations, Assessment, Development and Evaluations (Grade) was used as a tool for evaluation of the certainty of evidence.The Grade has two sections: the first is the certainty assessment with which publication bias, imprecision, indirectness, inconsistency, risk of bias, studies' design, and number of studies were evaluated.The second is the summary of findings with which the number of participants was evaluated.According to the assessment, the certainty of evidence could be rated high, moderate, low, or very low. 21The GRADEpro GDT was used. 22
The included studies showed wide variation in sample size (ranging from 27 to 680 individuals).The total number of individuals evaluated in the 23 included studies was 4,060.Of these, 2,967 (73.07%) were smokers and 1,093 (26.92%) were non-smokers.
Regarding the protein used for the identification of DNA damage, a high heterogeneity was observed among studies.Different methods of protein evaluation (PCR, IHC, and ELISA) were also employed.Table 1 shows the characteristics (including protein used and method of protein evaluation) and the results of the included studies.

Appraisal of the methodological quality of the included studies
Overall, the 23 included studies showed a low risk of bias for inclusion criteria of the sample, detailed description of sample characteristics and study setting, measurement of exposure in  For African-American subjects who smoked more than 24 PY, risk for oral cancer was significantly associated with the GSTM1 null polymorphism (OR: 5.4, 95%CI: 1.2 ± 24).No association was observed in African-Americans who were light-smokers (i.e.24 PY).A test for interaction between smoking and the GSTM1 genotype was not significant when the smoking-GSTM1 genotype interaction variable was introduced into the multiple logistic regression model for oral cancer risk.Significant associations were not observed between the GSTM3 genotype and oral cancer risk in African-Americans after stratification by smoking dose, although a trend was observed between the GSTM3 (B/B) genotype and oral cancer risk in the light-smoking African-American group (OR: 0.19, 95%CI: 0.03 ± 1.3).
Hsieh et al. 20 187 182 5 p53 PCR A specific pattern of mutation was observed in exons 5-9 of the p53 gene in OSCCs from smokers, alcohol users and BQ chewers.G:C to A:T transitions were the predominant mutations observed and associated with BQ and tobacco use.Seventeen of the 18 (94.44%)frameshift mutations including deletions and insertions occurred in smokers.Among them, 14 patients (82.35%) were also BQ chewers.In addition, most (20/22, 90.91%) G:C to T:A transversions occurred in smokers.All A:T to T:A and G:C mutations (n = 11) occurred in BQ chewers.All G:C to C:G transversions occurred in either smokers or BQ chewers.All of the mutations identified in patients with OSCCs in this series were somatic and not germ-line in origin, as DNA from normal tissue adjacent to p53-mutated tumor was negative for p53 mutations by both PCR-SSCP and DNA sequencing analysis.

Continue Continuation
Chaves et al. 37 71 66 5 p53 PCR The presence of TP53 mutation was independent of tobacco consumption.There was a predominance of A-G (45%) substitutions in current smokers and drinkers, followed by A-C transversions in 20% of the cases.The G-T mutation characteristic of tobacco smoke was found in only 1 tumor.Non-smokers and non-alcohol users were found to have 42.8% of G-A mutations, 2 of them were located in CpG sites and 1 in non-CpG sites.The maximum values of Ku70 and Ku80 expressions in carcinomas from smokers reached 60% and 50%, respectively.In tumors from non-smokers, Ku70 positivity was observed in 87.5% of cases and Ku80 positivity in 93.8% of tumors.Ku70 and Ku80 expression values reached maxima of 40%.The comparison of Ku70 and Ku80 expressions in tumors from smokers and non-smokers demonstrated a highly significant result for Ku70 (p = 0.008).Significant correlations between Ku70 and Ku80 expression were found in carcinomas from non-smokers (p < 0.05).In tumors from smokers, these significant relationships were not preserved (p > 0.05).
Prior et al. 38 27 21 6 ND2 PCR For ND2 gene, nucleotide 4917 was a significant mutation hotspot (P 1/4 0.027) and thus a potential smoking-associated biomarker in oral SCC.All patients having a mutation were males and classed as smokers with the exception of patient 5 whose smoking status was not known.Seven different types of mutation were discovered in the region of the D-Loop between nt 8 and 429.Base substitutions were observed in 16 (53.3%)different patients, 15 of whom had a classified smoking status.Of these, the 10 male patients with mutations were all self-classified smokers whereas, conversely, 4 of the 5 females with mutations were self-classified as non-smokers.This association of sex (males) and smoking status was statistically significant (p = 0.003) for patients with mutations.OSCC patients with areca quid chewing (p=0.029),cigarette smoking (p=0.027), or alcohol drinking habits (p = 0.025) were prone to have a higher mean cytoplasmic hTERT labeling score in OSCC samples than OSCC patients without these oral habits.OSCC patients with all 3 oral habits (p = 0.005) or with at least one oral habit (p = 0.007) also had a significantly higher cytoplasmic hTERT LS than OSCC patients without any oral habit.(OR: 0.39; 95%CI: 0.18-0.86)and rs1381 (OR: 0.75; 95%CI: 0.63-0.89)significantly reduced the risk of oral cancer.rs2031920 and rs3813867 significantly reduced oral cancer risk among exclusive tobacco chewers, (OR = 0.13; 95%CI = 0.03-0.59)and a positive dose-response relationship was observed.Similar results were observed for rs13181 (OR = 0.76; 95% CI = 0.59-0.97).
Tsai et al. 23 213 159 54 CCND1 PCR The genotype distribution of the polymorphisms of CNND1 A870G was significantly different between oral cancer patients and controls with a smoking habit (p = 0.0006).The GG genotype frequency was still significantly lower (12.9%) in cancer patients with a smoking habit than in smoking controls (16.6%).
Mallick et al. 30 39 31 8 p53, BCL-XL IHC In patients with tobacco habits (chewers + smokers), increased p53 intensity (p = 0.063) was observed compared to those with no habits, although it did not reach statistical significance.The probability of treatment failure (hazard ratio) was 3.2 times higher in the unfavorable responders compared to that of favorable response group.Bcl-xL protein was significantly upregulated (p=0.048) in the unfavorable responders compared to the favorable responders.

Continue
a valid and reliable way, use of standard criteria for measurement of the condition, adequate identification of confounding factors, statement of strategies used to deal with confounding, and use of appropriate statistical analysis.All studies showed unclear risk of bias for measurement of outcome in a valid and reliable way, as illustrated in Figure 2.

Results of the individual studies
The role for NAT1 appears to be independent of smoking behavior. 32GSTM1 was found to have a protective role against OSCC, since a higher risk for OSCC was associated with GSTM1 null polymorphisms. 28,29,31,33,36,37,42Also, a protective factor was detected for GSTT1, [29][30][31]33,37 as well as for CYP1A1, 28,30,36 whose null polymorphisms were related Continuation Tsai et al. 24 230 168 62 hOGG1 PCR The genotype distribution of hOGG1 codon 326 polymorphism was significantly different between oral cancer and control groups who had a smoking habit (p = 0.0198), but was not significant (p = 0.8357) in non-smokers.Consistent with the findings, the C allele frequency was still significantly higher in patients with cancer with a smoking habit than in smoking controls.There was no such difference between the non-smoking groups Zavras et al. 25 239 194 45 ERCC5 PCR The use of areca nut products, a prevalent habit in southern and eastern Asia that has been linked with higher frequency of epithelial tumor, was significantly higher among subjects with cancer (77.8%) compared with control subjects (15%).The risk of OSCC is 2.2-fold higher (95%CI: 1.31-3.68;p = 0.002) in tobacco-betel quid chewers, which is one of the main factors for oral cancer and is a common practice in Northeast India.The association between mtDNA copy number and OSCC risk was evident among tobacco-betel quid chewers rather than tobacco-betel quid non-chewers; the interaction between mtDNA copy number and tobacco-betel quid was significant (P = 0.0005).Similar results were observed when cases and controls were classified as tobacco-betel quid chewers and non-chewers based on low and high mtDNA copy number: the tobacco-betel quid chewers with the low mtDNA copy number had a 3.54-fold increased risk of OSCC (95%CI: 1.59-7.87).Frequency of smokers among cases was significantly higher than in healthy controls.However, compared to oral cancer subjects, the proportion of nonsmokers was significantly higher among the healthy control group (p = 0.001).The result shows that smoking increases the risk of oral cancer by five times (OR 5.03; 95%CI: 2.91-8.69).
to a higher risk of OSCC. Park et al. 42 also described a protective factor for GSTM3, with no significant associations between GSTM3 genotype and oral cancer risk among African-Americans after stratification by smoking dose.Likewise, authors reported that the presence of p53 mutations was independent of tobacco consumption 39 and the difference between smokers and non-smokers was not statistically significant. 23,32allick et al. 32 reported that the Bcl-xL protein was significantly increased in patients who responded to unfavorable situations, compared to patients who responded to factors favorable to OSCC treated with radiotherapy.The XPA and XPD genetic factors were modified by smoking. 13XPC genes were also modified by smoking and had heterozygous genotypes associated with elevated risk for OSCC. 34,38PARP-1 variants and ERCC5, hOGG1, and hTERT mutations were significantly higher in patients with OSCC who were smokers. 8,24,26,27Ku70, Ku80, Exo1, and CCND1 were significantly lower in OSCC patients who were non-smokers. 15,41For the ND2 protein, all patients with mutation were classified as smokers. 40

Synthesis of results and additional analysis
Three studies reporting dichotomous data regarding GSTM1 were incorporated into one subgroup for analysis. 29,37,42No significant difference was observed between smokers and non-smokers with respect to damage to GSTM1 (OR: 0.60; 95%CI: 0.30-1.18;I 2: 0%).Two studies 29,37 reporting dichotomous data regarding glutathione S-transferase theta 1 (GSTT1) were incorporated into the second subgroup.No significant difference was observed between smokers and non-smokers with respect to damage to GSTT1 (OR: 1.18; 95%CI: 0.49-2.83;I 2: 0%).Dichotomous data regarding hydrolase proteins (Ku70 and Ku80) were incorporated into the third subgroup. 41No significant difference was observed between smokers and non-smokers with respect to damage to hydrolase proteins (OR:0.74;95%CI: 0.18-3.05;I 2 : 0%).Dichotomous data with respect to transferase proteins (GSTM1, GSTT1, and GSTM3) were incorporated into the last subgroup. 37,38o significant difference was observed between smokers and non-smokers with respect to damage to transferase proteins (OR: 0.70; 95%CI: 0.42-1.12;I 2: 0%). Figure 3 shows the subgroup analyses.

Assessment of the certainty of evidence
The certainty of evidence was very low.Table 2 shows the complete information on evaluation of certainty of evidence.

Discussion
Summary OSCC follows a multifactorial and dynamic course, with numerous changes contributing to the development of the disease.This systematic review and meta-analysis investigated the effects of tobacco on DNA of individuals with OSCC, comparing smokers and non-smokers.To our knowledge, this

Uncertain risk of bias Low risk of bias
Was appropriate statistical analysis used?
Were the outcomes measured in a valid and reliable way?
Were strategies to deal with confounding factors stated?
Were confounding factors identified?
Were objective, standard criteria used for measurement of the condition?
Was the exposure measured in a valid and reliable way?
Were the study subjects and the setting described in detail?
Were the criteria for inclusion in the sample clearly defined?
is the first comprehensive analysis about the effect of tobacco on DNA of smokers and non-smokers with OSCC.DNA damage response is a complex signaling network involving cell cycle checkpoints as well as DNA damage and repair pathways. 43Herein, 14 molecular changes in gene/protein groups and altered genes/proteins, such as tumor suppressor, antiapoptotic, cyclin, monooxigenase, glycosidase, enzyme binding, transferase, DNA binding, hydrolase, helicase, ribonucleoprotein, exonuclease, endonuclease, and translocase were examined.Meta-analysis for GSTM1 and GSTT1, as well as for transferase and hydrolase groups, showed no significant difference between smokers and non-smokers regarding the damage/polymorphism of these proteins.Although meta-analysis was impossible for tumor suppressor and anti-apoptotic genes, changes in these two groups were increased in OSCC smokers, as reported by Hsieh et al. 23 and Mallick et al. 32 , respectively.As to the other protein groups, the number of studies was insufficient to allow solid conclusions.

Transferase proteins
Transferase proteins belong to a class of enzymes that transfer a specific functional group from the donor molecule and catalyze numerous biological reactions of critical importance for a living system.In this study, four proteins from this group were analyzed, i.e., GSTM1, GSTM3, GSTT1, and NAT1.Glutathione S-transferases (GSTs) are an important group of these enzymes, which detoxify both endogenous compounds and foreign chemicals such as pharmaceuticals and environmental pollutants. 44he presence of GSTT1 and GSTM1 is essential for carcinogenic detoxification. 33Some authors have shown that null GSTM1 and GSTT1 genotypes were likely to be associated with a higher risk of different types of cancers such as hepatocellular and thyroid malignancies. 45,48The increased risk factor of null GSTM1 in OSCC is higher than that of null GSTT1, as revealed by the findings presented herein.In this regard, the GSTM1 enzyme possibly plays an important role inside the mitochondrial matrix as an mtDNA-protecting factor for damage caused by reactive oxygen species. 33GSTM1 and GSTT1 polymorphisms, as well as detoxification enzymes have been identified in individuals with OSCC, but are not believed to be risk factors. 47For instance, Kietthubthew et al. 37 reported that the frequencies of null GSTM1 and GSTT1 in their non-cancer sample were 30.2% and 47.2%, respectively.On the other hand, the results showed that individuals with a susceptible version of the GSTM1 genotype (null genotype) had a 2.6 times higher risk of OSCC, regardless of exposure to environmental hazards such as tobacco.However, Minina et al. 48suggested that the GSTM1 null genotype increased the frequency of chromosomal damage in smoking patients with lung cancer.In addition, Park et al. 42 showed that the risk of oral cancer was significantly associated with GSTM1 null polymorphism among African American individuals who had smoked heavily for more than 24 years.
Polymorphisms of N-acetyltransferase-1 and -2 (NAT-1/2), another type of transferase responsible for the metabolism of tobacco carcinogens, have been investigated for a potential role in oral carcinogenesis.Nevertheless, no correlation was found indicating that they do not themselves contribute to the carcinogenic process. 49Unfortunately, there are discrepancies among studies associating these polymorphisms with OSCC, possibly related to demographics, as observed for GSTM1 associated with oral cancer in Asians, but not in Caucasians. 50Based on a hypothesized role for NAT1 in modulating the effects of carcinogens present in tobacco smoke, Katoh et al. 35 investigated a combined role for smoking and the NAT1 genotype.The authors suggested that individuals with NAT1*10 alleles were at higher risk for OSCC, but that smoking history did not play a role in this genetic relationship.Smoking behavior in cases or controls (either smoker or non-smoker index) was not associated with any NAT genotype.
The role of NAT1 appears to be independent of smoking behavior.

Hydrolase proteins
Hydrolase proteins are an enzyme system that catalyzes hydrolysis reactions.In the present study, Ku represented the protein associated with this group.It is now well established that, while not essential for individual life in the short term, Ku function is critical for the maintenance of genomic integrity and for proper cellular and organismal development. 51Ku70 and Ku80 regulate subunits of the DNA-dependent protein kinase, a crucial enzyme involved in the repair of double-strand breaks in DNA.Along this line, Korabiowska et al. 41 investigated the role of the Ku70 and Ku80 genes in the progression of OSCC.Among their findings, Ku70 expression correlated very strongly with smoking habits.The authors demonstrated that dysregulation of the Ku70 and Ku80 axis may be influenced by tobacco. 41mor suppressor and antiapoptotic proteins Two groups deserve recognition in this study, even though no meta-analysis was possible.Tumor suppressor and antiapoptotic proteins have also been highlighted in the literature when cancer is involved. 52he inactivation of tumor suppressor genes result in a phenotype only if both copies of the gene are lost.In the carcinogenic process, inactivation of one copy of a tumor suppressor gene must usually be followed by loss of the remaining copy of the gene and by the emergence of the tumor phenotype. 53The importance of the p53 tumor suppressor gene in the process of carcinogenesis has been well established in the current literature. 23,54Mutation of P53 has been reported in over 80% of all cancers, 55 with a higher incidence in tobacco-related cancers.Mallick et al. 32 reported an increased intensity of p53 among patients with tobacco habits compared to non-smokers.Notably, tobacco carcinogens played an important role in p53 mutations in Taiwanese patients with OSCCs. 23ntiapoptotic proteins were represented by BCL-2.The BCL-2 family may be understood as a tripartite apoptosis control system comprising one set of anti-apoptotic proteins and two sets of pro-apoptotic proteins, which interact to determine whether cells will live or die in many pathophysiological states. 56Overexpression of BCL-2 was originally described in leukemia and in B-cell non-Hodgkin's lymphoma.BCL-2 overexpression is, in most cases, the consequence of a t (14,18)  translocation that has its break-point close to the BCL-2 gene.However, BCL-2 overexpression is, by itself, insufficient for malignant transformation, but may provide a predisposition to the development of B-cell lymphomas. 57

Limitations
The inherent limitations of a systematic review and meta-analysis should be considered here.First, due to the heterogeneity of genes/proteins, the comparison of a significant number of studies on the same molecule was unfeasible.Moreover, some studies did not show a relationship between smoking and gene/protein changes in their results and were thus unfit for inclusion in this systematic review and meta-analysis and for the analysis of some protein groups.

Conclusions
In summary, the articles included in the present systematic review and meta-analysis diverged in relation to the role of tobacco in genetic changes that predispose to OSCC.While in some studies smoking history has not been shown to play a differential role in carcinogenesis, 35,37,39 the vast majority confirm that smokers are more likely to have DNA alteration -mainly associated with genetic polymorphisms. 13,24,25,29,40,42Therefore, our study demonstrates that major changes in genes or proteins do not necessarily occur in smoking patients.Indeed, the role of tobacco in carcinogenesis is well known.As far as we know, there are nearly 60 carcinogenic compounds in tobacco smoke.However, great genetic changes in non-smoking patients were a common finding in some studies, 35 while other studies found similar patterns of genetic alterations between smokers and non-smokers, 41 suggesting that the genetic alteration evaluated was not related to smoking habit.It is possible that the referred genes do not play a relevant role in tobacco-related carcinogenesis, but are relevant to the carcinogenesis process as a whole.Thus, further studies are needed to understand OSCC pathways in smokers and non-smokers.

Figure 1 .
Figure 1.Flowchart showing the results of the search process.

Table 1 .
Articles included in this systematic review.
The crude OR of the stratification with either harboring variant XPA genotype (A/G or G/G, "variant") or with smoking habit was 3.52 (95%CI: 1.26-9.84),andthecrude OR of those with both harboring variant XPA genotype and smoking habit was increased to 47.7 (95%CI: 15.48-147.01).By the same analyses strategy, the same trend was observed and the joint effect of XPD genotype and smoking habit on oral cancer were also significant.The "common" group with putative low-risk XPD A/A genotype and without smoking habit was used as reference.The crude OR of the stratification with either harboring variant XPD genotype (A/C or C/C, "variant") or with smoking habit was 28.48(95%CI: 13.93-58.23),and the crude OR of those with both harboring variant XPD genotype and smoking habit was 26.33 (95%CI: 7.87-88.04).The crude ORs of the stratification with one of the three factors, variant XPA (A/G or G/ G), variant XPD (A/C or C/C) genotype, or smoking habit, was 3.59 (95%CI: 1.27-10.19),and the crude ORs of the stratification with two or all of the three factors were significantly increased to 24.05 (95%CI: 8.38-68.95).The 7-fold synergistic increase from 3.59 to 24.05 suggested that genetic factors (XPA and XPD), modified by the environmental factor (smoking), may also contribute to oral cancer risk.

Table 2 .
Assessment of the certainty of the evidence.The certainty of the evidence has been downgraded by one level.The studies did not take the characteristics of smokers and non-smokers into account in the analyses; b The certainty of the evidence has been downgraded by one level.The number of individuals is lower than the optimal information size.
a c Studies incorporated with non-significant results.