Introduction

The incidence of oral squamous cell carcinoma (OSCC) has increased in the past few decades in many countries. Patients are usually diagnosed between their sixth and seventh decade of life, after several years of tobacco and alcohol consumption. However, an increasing number of individuals have been diagnosed with OSCC under the age of 45 years [1, 2].

In literature, these subjects who precociously develop OSCC are referred to as “young patients.” Young individuals represent about 6% of OSCC patients [2]. In a significant proportion, especially among females, the absence of traditional risk factors, smoking and drinking, has been noticed. Additionally, the period for such carcinogens to exert a detrimental effect on these individuals is relatively short [2, 3].

Molecular features and environmental factors related to the earliest establishment of OSCC remain to be unveiled. Particularly, the role of tobacco and alcohol consumption in the development of OSCC at young ages has been investigated. While some studies demonstrate a small proportion of young OSCC patients who are tobacco smokers and alcohol users compared to older OSCC individuals [4,5,6,7], other works did not find remarkable differences in these habits between the two age groups [8,9,10].

The elucidation of such matter is crucial to appraise the real role of tobacco and alcohol in oral carcinogenesis at early ages. It also aids to explore the identification of other possible etiological agents for OSCC. Thereby, individualized preventive measures according to the main risk factors in each group can be applied. The purpose of this systematic review (SR) and meta-analysis was to compare the proportion of tobacco and alcohol consumption between young (up to 45 years of age) and older (over 45 years of age) OSCC patients.

Methods

Eligibility criteria

Inclusion criteria

Primary studies on OSCC comprising lip, anterior two-thirds of the tongue, and rest of mouth (gingiva, palate, floor of mouth, buccal mucosa, alveolar ridge) were included. To be eligible, original research papers needed reporting of data from two age groups, divided into a young OSCC group (individuals up to 45 years old), and a comparison group of older OSCC patients (over 45 years old). Due to the lack of standardization in the literature regarding the term “young,” a 5-year overlap among studies was accepted; however, this 5-year overlap should not be present within individual studies. Additionally, studied needed to report alcoholic beverage consumption and use of tobacco in any forms (e.g., smoked, chewed, sniffing).

Exclusion criteria

Studies were excluded when (1) it was not possible to have data separately for young and older age groups; (2) did not report tobacco and alcohol consumption in young and older patients; (3) the sample had no oral squamous cell carcinoma as a separate group; (4) reviews, case reports, protocols, personal opinions, letters, posters, conference abstracts, in vitro and in vivo animal studies.

Information sources and search strategies

Search strategies were conducted using terms for “risk factors,” “tobacco,” “alcohol,” “young patients,” and “OSCC.” Appropriate word combinations were developed for each of the following bibliographic databases: PubMed. EMBASE, Scopus, LILACS, and Web of Science. A nonpeer-reviewed literature search was also performed on Google Scholar, OpenGrey, and ProQuest Dissertations & Theses Global (Appendix 1). Additionally, manuscripts were hand-searched by checking the list of references of the included studies. The references were managed using the EndNote X7 software (Thomson Reuters, New York, NY, USA). The bibliographic search was updated on July 9, 2021.

Study selection

A two-phase process was followed to select studies. In phase 1, titles and abstracts of identified records were independently screened by three reviewers (E.A.B., R.G., E.R.C.R.); studies that did not meet the inclusion criteria were excluded. In phase 2, the same investigators applied these criteria to the full text of the manuscripts. Any inconsistencies were resolved by consulting experts (E.G., S.W.) to finally decide whether the article should be included in the SR.

Data collection process

Data collection was also independently performed by the three reviewers and then cross-checked. For each included study, data regarding study design, country of first author, and year of publication; patients’ age; sample size; tumor site; and prevalence of alcohol and tobacco consumption were collected. Information about the number of individuals who reported tobacco and alcohol consumption or no habits was compiled in the software Excel version 16.27.

Risk of bias assessment

Risk of bias (RoB) was independently assessed by three reviewers (E.A.B., R.G., E.R.C.R.) using the Joanna Briggs Institute Critical Appraisal Checklist for Studies Reporting Prevalence Data [13]. Scoring decisions were agreed upon by all reviewers before the critical appraisal. RoB was categorized as “high” when the study reached up to 49% score “yes”; “moderate” when the study reached 50 to 69% score “yes”; and “low” when the study reached more than 70% score “yes.” A weighted bar plot and a risk of bias graph were generated with software (Review Manager 5.4; the Cochrane Collaboration, 2021).

Summary measures

The comparison in the proportion of young and older OSCC patients who reported alcohol consumption and tobacco smoking was considered as the main outcome.

Synthesis of results

Data on the frequency of individuals exposed to tobacco or alcohol in both young and older patient groups collected from articles were gathered in a Microsoft® Excel (Microsoft Corporation, version 16.42) table, and then transferred to the R software Statistics version 4.0.2 (The R Foundation) (α = 0.05) to conduct the meta-analyses. The “meta” package was applied; raw data were converted by using the “logit transformation.” To estimate confidence intervals for individual studies, the Clopper-Pearson interval was applied. Regarding statistical heterogeneity analysis, the following parameters were calculated: Cochran Q (χ2), I-squared (I2), Tau-squared (τ2), and the prediction interval. Furthermore, the τ2 was assessed through the restricted maximum likelihood method.

The distribution of true effect sizes was expected across included studies; therefore, the random effect model was applied [14]. Studies were grouped according to the reported habits. The first meta-analysis included data from studies reporting the proportion of tobacco smokers. The second meta-analysis included data from studies informing the proportion of individuals reporting alcohol consumption. The third meta-analysis assessed the proportion of individuals who reported the concurrent habits of tobacco and alcohol consumption. Finally, a fourth meta-analysis assessed the proportion of individuals who reported no habits. These analyses were conducted for both age groups. From each of the meta-analyses, a forest plot was generated to illustrate the combined results with the software R Statistics version 4.0.2 (The R Foundation).

Certainty of evidence assessment

The Grading of Recommendations Assessment, Development and Evaluation (GRADE) evidence profile was used to verify the overall quality of evidence using the online software version (GRADEpro GDT) [14].

Results

In phase 1 of study selection, the search across electronic databases identified 6675 studies. After removing duplicates, 4808 articles remained. In addition, 107 studies were identified from nonpeer-reviewed literature. Sixteen articles were included after searching reference lists. After a comprehensive evaluation of the titles and abstracts, the reviewers considered 97 studies for eligibility, of which 59 were excluded after full-text reading (Appendix 2). Subsequently, 38 studies [5,6,7,8,9,10, 15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45] were included for qualitative analysis and quantitative synthesis. A flow chart detailing the process of identification, inclusion, and exclusion of the studies is shown in Fig. 1.

Fig. 1
figure 1

PRISMA flow diagram of literature search and selection criteria

Full size image

Study characteristics

The included studies were conducted as retrospective analysis in which data were collected from patient’s records. The 38 included studies were published in eighteen countries: Australia [23, 36,37,38] Brazil [19, 21, 24,25,26, 32, 34], Canada [7, 35], China [5, 18] [43, 45], France [16], Germany [6, 30], India [15, 27, 42], Ireland [28], Israel [20, 31], Italy [8], Japan [41], Singapore [44], South Korea [29, 40], Sri Lanka [4], Spain[17], Taiwan[22], Thailand [10], and USA [9, 32, 39] from 1977 to 2021. Table 1 summarizes the descriptive characteristics of the included studies. Most studies only reported the proportion of patients who were tobacco and alcohol consumers, without specifying the type, quantity, and frequency of the habit. Therefore, it was not possible to stratify groups according to the amount of tobacco or alcohol consumed.

Table 1 Summary of included articles’ descriptive characteristics
Full size table

Risk of bias within studies

Regarding the overall risk of bias, studies were considered at low risk, ranging an overall proportion of 88.8 to 100% yes answers to the JBI checklist [13]. The main source of bias was the small sample sizes in at least one of the study groups in sixteen studies [6, 7, 18,19,20,21,22,23, 28,29,30,31,32, 34, 36, 40, 44] and data analysis being conducted with insufficient coverage of the identified sample [29, 31]. A weighted risk of bias bar plot and risk of bias graph is depicted in Fig. 2.

Fig. 2
figure 2

Risk of bias graph: authors’ judgments about each risk of bias item presented as percentage across all included studies

Full size image

Certainty of evidence

GRADE analysis rendered a very low overall certainty of evidence due to varying proportions of (1) inconsistency, for having different methods to define quantity and frequency of consumption, and due to high methodological heterogeneity, which results from differences in the way studies were conducted; (2) indirectness, for having no healthy control groups; and (3) imprecision, due to wide confidence intervals for both outcomes. However, it is important to notice that certainty of evidence starting in a low rank is due to the observational nature of studies.

Synthesis of results

A total of 15,832 OSCC patients were included in this SR. The pooled young group comprised 2439 patients, being 861 women (35.3%) and 1578 men (64.7%). The pooled older group encompassed 13,393 patients, being 4803 women (35.8%), 8554 man (63.9%) and 36 with sex data not being reported (0.3%). The tongue was the commonest reported oral subsite of cancer onset. In the young patient’s OSCC group, meta-analyses results demonstrated that 39.5% (CI = 31.7% to 47.9%, I2 = 78%) of the patients reported tobacco smoking (Fig. 3). Additionally, 30.9% (CI = 22.7% to 40.5%, I2 = 83%) reported alcohol consumption (Fig. 4). The concurrent habit of smoking and drinking alcoholic beverages was reported by 24.4% of the young patients (CI = 14.9% to 37.3%, I2 = 69%) (Fig. 5).

Fig. 3
figure 3

Forest plot depicting the overall proportion of tobacco smokers among young and older cancer patients. Squares represent sample sizes; horizontal lines regard to confidence intervals, and the diamond represents the pooled prevalence from all studies included in meta-analysis. Figure generated with the software R Statistics version 4.0.2 (The R Foundation)

Full size image
Fig. 4
figure 4

Forest plot depicting the overall proportion of alcohol consumption within young and older oral cancer patients. Squares represent sample sizes; horizontal lines regard to confidence intervals, and the diamond represents the pooled prevalence from all studies included in meta-analysis. Figure generated with the software R Statistics version 4.0.2 (The R Foundation)

Full size image
Fig. 5
figure 5

Forest plot depicting the overall proportion of concurrent tobacco and alcohol consumption within young and older oral cancer patients. Squares represent sample sizes; horizontal lines regard to confidence intervals, and the diamond represents the pooled prevalence from all studies included in meta-analysis. Figure generated with the software R Statistics version 4.0.2 (The R Foundation)

Full size image

In the older patient’s OSCC group, meta-analyses demonstrated 48.4% (CI = 37.8 to 59.2%, I2 = 94%) of the patients reported tobacco smoking (Fig. 3). Furthermore, 45.8% (CI = 35.6 to 56.5%, I2 = 95%) reported alcohol consumption (Fig. 4). The concomitant habit of smoking and drinking alcoholic beverages was informed by 47% (IC = 26% to 69.1%, I2 = 93%) of the older OSCC patients (Fig. 5).

The proportion of older individuals reporting habits of tobacco and alcohol consumption, either isolated or concurrently, was higher than in young patients. However, in the proportion meta-analysis of patients reporting no habits, a higher rate was observed in the young patients’ group (28.1%, IC = 8.9 to 61%, I2 = 80%) than in the older patients’ group (24%, IC = 19.3% to 29.5%, I2 = 0%) (Fig. 6).

Fig. 6
figure 6

Forest plot depicting the overall proportion of young and older oral cancer patients reporting no habits. Squares represent sample sizes; horizontal lines regard to confidence intervals, and the diamond represents the pooled prevalence from all studies included in meta-analysis. Figure generated with the software R Statistics version 4.0.2 (The R Foundation)

Full size image

Discussion

This SR and meta-analysis included 38 studies in which tobacco and alcohol consumption were investigated in young and older OSCC patients. In the current study, the number of male patients in both samples outweighed the number of female individuals, which is in accordance with the previous literature [1, 2, 46]. The tongue was the most common site of occurrence of OSCC, conforming to other studies [1, 47].

Despite some studies in this SR demonstrating a small proportion of smokers in the young group [4,5,6,7], tobacco was still the most frequently reported risk factor attributed to OSCC in both young and older patients. Overall, the meta-analysis of studies reporting tobacco smoking demonstrated that 39.5% of the young OSCC patients reported smoking habits, and almost half of the older patients were tobacco smokers.

Tobacco smoke has over 70 carcinogens which have been evaluated by the International Agency for Research on Cancer (IARC) as having sufficient evidence for carcinogenicity in either laboratory animals or humans, acting both as a tumor initiator and promoter [48]. Tobacco-specific nitrosamines, polycyclic aromatic hydrocarbons, and volatile aldehydes present in several forms of tobacco derivates have been attributed to mutations in oncogenes, loss of cell cycle control, and decrease of apoptosis, contributing to carcinogenesis [49, 50].

Some studies included in this SR highlighted that those individuals reaching the age of 40 years who started smoking early in adolescence might have been exposed to tobacco carcinogens for over 20 years [51, 52]. Therefore, the exposure time would be long enough to induce oral carcinogenesis in these young individuals [51, 52].

However, a noteworthy proportion of patients in both groups did not report any recognized risk factors for OSCC and should be closely evaluated in future studies [15, 21]. Specific risk factors for OSCC not addressed in the present study, which are more common in certain global areas, could potentially contribute to the development of OSCC at younger ages [1, 53]. For instance, exposure to secondhand smoke (SHS) from childhood should be investigated in further studies as SHS has recently been implicated as a significant risk factor for OSCC [54].

Additionally, in Southeast Asia, where OSCC ranks among the most frequent cancer types, there is a traditional habit of chewing areca nut derivates (in forms of pan/paan, gutka, betel quid, etc., with or without tobacco), which is reported as a risk factor for OSCC [2, 53]. These habits could reflect on the high incidence of OSCC not only in the elderly but also in the young individuals as observed in studies included in this SR. These studies involved patients from countries such as India, Bangladesh, Pakistan, and Sri Lanka [2, 4, 22, 27], in which a high frequency of betel quid consumption has been reported [2, 56].

Environmental smoke exposure [54], polymorphism in genes relating to the detoxification of carcinogens [55], anemia, chronic mechanical irritation [56], periodontal pathogens and microbial dysbiosis [57], and other possible risk factors need to be investigated to unveil etiological agents for OSCC in this young group of patients [3].

Some articles included in this SR have shown a high rate of young patients reporting alcohol consumption, with an increased intake among women [10, 22, 24,25,26, 28], as well as early and heavy alcohol consumption habits [43]. However, the overall proportion of older patients who reported drinking habits was markedly higher than the proportion of young patients reporting alcohol consumption. The evaluation by the IARC Working Group (Monograph 100E-2012), based on a significant number of analytical epidemiological studies with different designs (both cohort and case–control) from several geographical regions, provide sufficient evidence that alcohol consumption is a carcinogenic agent especially to oral and pharyngeal cancer [58].

The independent effect of alcohol consumption on the risk of oral and pharyngeal cancer is demonstrated in studies including non-smokers [59]. The risk increases with added amounts of alcohol drinking in most of these studies [60, 61]. However, it should be observed that intrinsic susceptibility to metabolize carcinogens derived from alcohol, particularly acetaldehyde, may vary among individuals [62]. This could partially explain why some patients may develop cancer at younger ages.

This SR revealed that in six studies [4, 10, 15, 19, 21, 33], about a quarter of the young patients and 47% of older individuals with OSCC reported the concurrent habit of alcohol consumption and tobacco smoking. According to the results of a recent meta-analysis including fifteen primary studies, the synergistic consumption of alcohol and tobacco both smoked and smoke-less significantly increased the odds for the occurrence of oral squamous cell carcinoma, and its concurrent consumption was more harmful than the sole consumption of alcohol or tobacco in most included studies [63].

In a review, Poschl and Seitz [64] discuss some possible mechanisms of alcohol as a cocarcinogen. Locally, alcohol acts as a solvent that increases the absorption of carcinogens into the mucosa. Ethanol facilitates the uptake of environmental carcinogens, especially from tobacco smoke, through the cell membrane of oral keratinocytes made more permeable by its direct effect [40]. Therefore, alcohol may contribute to oral carcinogenesis solely or as an adjuvant factor [65].

The main risk of bias in the included studies of this SR was related to small sample sizes and data analysis being conducted with insufficient coverage of the identified sample. When primary studies rely upon collected data from patients’ records, the quality of evidence could be compromised, as data are often improperly completed and important information about patients’ lifestyles might be missing. Cohort studies would be appropriate to aid the identification of risk factors for OSCC at young ages.

The main limitation of this SR concerns the overall low certainty of evidence in primary studies. Other limitations include the retrospective nature of primary studies, lack of healthy control groups, differences in reporting quantity and frequency of tobacco and alcohol consumption, having not the same age cutoffs for young and older patients, and lack of reporting confounding factors. Tobacco and alcohol consumption were usually reported as a general term without specifying the types, frequency, or duration of use. Additionally, the relation between alcohol consumption and smoking history and demographic variables such as gender and ethnicity were not approached in the current SR and should be in-depth addressed in further research.

For future studies, it is highly recommended for researchers to specify the frequency of consumption as well as the type of beverages, for the alcoholic percentual may vary significantly between fermented beverages and distilled spirits. The same applied to tobacco use, i.e., future studies should specify the form of tobacco use (smoking, chewing, snuffing, etc.), the daily frequency and quantity of years that the patient has smoked, and the exposure to passive smoking. Moreover, a meta-analysis of association, including studies addressing both patients with OSCC and healthy individuals who have alcohol intake or smoking history, is recommended for future research. This was not feasible in the current study due to the lack of a healthy control group.

Conclusions

According to this systematic review, tobacco smoking and alcohol consumption had an elevated prevalence in both young and older groups of patients. However, the comparison in the proportion of individuals reporting tobacco and alcohol consumption demonstrated that these habits were more prevalent in the older group (48.4% and 45.8% respectively) than in the young group (39.5% and 30.9%, respectively). Nontraditional risk factors for OSCC in young patients need to be investigated in future research, and individualized preventive measures according to the main risk factors in either group should be readily addressed in clinical practice.

Registration and protocol

This systematic review was developed according to the Preferred Reporting Items for Systematic Reviews and Meta Analyses Checklist (PRISMA) [11]. The study protocol was registered at the Prospective Register of Systematic Reviews (PROSPERO; Centre for Reviews and Dissemination, University of York; and the National Institute for Health Research) [12] under the registry code CRD42017065583.