3.1:Included studies
A total of 10 systematic reviews and meta-analyses were included after applying eligibility criteria [21–30] (Fig. 1). These studies collectively synthesized evidence on risk factors associated with OPMDs and prognostic determinants in OSCC (Table 1). Systematic Reviews (Mata-analysis) were published between 2019 and 2024, originating from diverse regions including Spain, India, Chile, and Portugal, with frequent contributions from European and South Asian institutions. Notably, several studies originated from high-impact journals such as Oral Oncology, Cancers, and Oral Diseases, reflecting strong research interest in the topic.
3.2: Pre-registration and Funding
Across the included reviews, most were both systematic reviews and meta-analyses, though a few were purely systematic in nature. A significant number of studies reported PROSPERO registration, reflecting methodological transparency and adherence to best practices in evidence synthesis. Funding sources were variably declared; while many SRs conducted without funding support, a few received government or institutional support, especially from Health departments and Research councils. Keim-del Pino et al. (2024) from Spain, published in Cancers, conducted a systematic review and meta-analysis registered with PROSPERO CRD42024511457, without external funding [28]. Dwivedi et al. (2024) from India, published in the Asian Pacific Journal of Cancer Prevention, received financial support from the Department of Health Research, Ministry of Health & Family Welfare, and registered the review under PROSPERO CRD42021282787 [24]. Rivera et al. (2020) from Chile, in Molecular and Clinical Oncology, conducted a systematic review without declared funding and registered it under PROSPERO CRD42018086476 [26]. González-Moles et al. (2020) from Spain, publishing in Oral Oncology, also declared no funding and mentioned PROSPERO registration, although without number [30]. Similarly, Lorenzo-Pouso et al. (2022) from Spain, in the Journal of Oral Pathology & Medicine, reported no funding and was registered with PROSPERO CRD42022299026 [21]. An earlier review by González-Moles et al. (2019) in Oral Oncology also noted no funding and was registered under PROSPERO CRD42019128539 [23]. Monteiro et al. (2020) from Portugal, publishing in Oral Diseases, declared partial funding from Henry Schein Cares and CAPES-FAPESC, with PROSPERO registration number CRD42020163464, while their later review in 2022 in the same journal reported no funding and was registered under CRD42022329326 [29]. Another study by González-Moles et al. (2021) in Cancers was funded by the Plan Andaluz de Investigación and registered under PROSPERO CRD42021267896 [22]. Finally, Saluja et al. (2019) from India, in Cancer Epidemiology, Biomarkers & Prevention, received funding from the Department of Health Research, India, though PROSPERO registration was not reported [27]. This diversity in authorship, geography, funding, and transparency contributes to a broad and balanced representation of the current evidence landscape regarding risk and prognosis in OPMDs and OSCC.
3.3:Type of studies Included in SR/MA
The included SRMA demonstrated a wide range of scope and methodological diversity in terms of review types, number of primary studies, and study designs. Keim-del Pino et al. conducted a combined SRMA including 110 observational studies, making it one of the most comprehensive in terms of study volume [28]. Dwivedi et al. also performed both a SRMA, incorporating 49 studies, of which 40 were case-control, 7 cross-sectional, and 2 cohort studies [24]; however, it should be noted that not all included studies focused solely on oral cancer or OPMDs, as the review encompassed various cancer types. Rivera et al. carried out a systematic review limited to 4 retrospective cohort studies [26]. González-Moles et al. contributed multiple reviews: one included 27 studies without a detailed report on study design; another review included 82 studies, of which 74 were retrospective longitudinal and 8 were prospective longitudinal studies; and yet another review analyzed 23 studies qualitatively and 21 quantitatively, primarily comprising retrospective longitudinal studies with a single prospective longitudinal study [22, 23, 30]. Lorenzo-Pouso et al. conducted both a systematic review and meta-analysis that synthesized 10 longitudinal observational studies [21]. Monteiro et al. also conducted two reviews: the first was a systematic review involving 46 studies, primarily retrospective cohort studies (n = 38), along with 6 randomized clinical trials (RCTs), 1 prospective cohort study, and 1 quasi-experimental study [25]. The second review, a combined systematic review and meta-analysis, included 6 studies, comprising 5 retrospective cohort studies and 1 randomized clinical trial [29]. Lastly, Saluja et al. included 18 studies, mostly retrospective in nature, in their SRMA [27]. Most SRs have a predominance of retrospective observational studies with relatively fewer high-quality prospective studies and RCTs.
3.4: Population and Lesions
Sample size
The included systematic reviews and meta-analyses encompassed diverse populations, primarily focusing on patients with OPMDs and OSCC, with notable variability in follow-up durations and language restrictions. Keim-del Pino et al. examined only OLP cases and did not specify the follow-up period; notably, they did not restrict their review to English-language publications [28]. In contrast, Dwivedi et al. included both OSCC and OPMD populations but applied language restrictions, excluding non-English articles, though follow-up duration was not detailed [24].
Rivera et al. studied OPMD cases, including oral dysplasia, leukoplakia, and erythroplakia, with English-only inclusion, and did not report follow-up duration [26]. González-Moles et al. included both OLP and OSCC cases, reporting a wide range of follow-up durations from 10 days to 660 months, and applied language restrictions [22]. In another review, Lorenzo-Pouso et al. focused on oral erythroplakia, with a reported mean follow-up of 6.66 years (range 3.83–16 years), also limited to English-language studies [21]. A further review by González-Moles et al. on various OPMDs, including OLP, oral lichenoid lesions, lichenoid reactions, and dysplastic OLP, provided a detailed distribution of follow-up durations across studies: 29 studies with ≥ 6 months, 53 with < 6 months, 16 with ≥ 12 months, and 66 with < 12 months, all with English-language restrictions [23]. Monteiro et al. (Monteiro L et al. 2020; 2022) authored two reviews centered exclusively on oral leukoplakia [25, 29]. One review (Monteiro L et al. 2020) did not report follow-up duration [25], while the other (Monteiro L et al. 2022) included studies with follow-up periods ranging from 32 to 90 months [29]; both were restricted to English-language studies [25, 29]. In contrast, one review by González-Moles et al. that analyzed both OSCC and OPMD (specifically proliferative verrucous leukoplakia - PVL) reported a mean follow-up of 65.63 months and was among the few reviews that did not impose language restrictions [23]. Lastly, Saluja et al. evaluated various OPMDs such as OLP, oral leukoplakia, and oral erythroplakia, with reported follow-up ranging from 0.66 years to 16 years, and applied English-only inclusion criteria. Overall, most reviews focused on OPMD populations, particularly OLP and oral leukoplakia, with considerable variation in follow-up duration and a prevalent
3.5: Prognostic Risk Factors Assessed
Across the included systematic reviews and meta-analyses, a different prognostic risk factors were investigated for their potential role in predicting malignant transformation of OPMDs and prognosis in OSCC. The spectrum of risk factors spanned molecular biomarkers, clinical characteristics, lifestyle habits, and histopathological features. Keim-del Pino et al. explored an extensive panel of 104 biomarkers associated with cancer hallmarks, assessing their expression in OLP using immunohistochemistry techniques [28]. The comparator groups included both healthy individuals and patients diagnosed with OSCC, providing a contrast to identify specific biomarkers that may predict progression. Dwivedi et al. focused on non-tobacco product (NTP) usage—including betel quid, areca nut, and paan—as risk factors [24]. They defined NTP exposure clearly and compared outcomes between exposed and unexposed groups across both case-control and cohort designs, though the scope extended beyond OSCC and OPMD to other cancer sites [24]. Rivera et al. assessed key prognostic biomarkers such as the degree of dysplasia, expression of retinal aldehyde dehydrogenase 1 (ALDH1A1), prominin-1 (PROM1), and podoplanin (PDPN) [26]. These biomarkers were evaluated using logistic regression or Cox proportional hazard models, with comparator groups based on absence of dysplasia or negative biomarker expression [26]. Several studies by González-Moles et al. investigated traditional clinical risk factors, including age, sex, tobacco use, and alcohol consumption [22, 23, 28]. These factors were evaluated by stratifying the populations into groups such as smokers vs. non-smokers, males vs. females, and age groups above or below 40 years. Another study by the same group focused on the malignant transformation rate (MTR) of OLP and proliferative verrucous leukoplakia (PVL), without comparator groups but using transformation rates to assess prognostic implications [23].
Lorenzo-Pouso et al. evaluated the impact of tobacco, betel quid, alcohol consumption, and epithelial dysplasia in patients diagnosed with oral erythroplakia (OE), identifying their prognostic relevance by comparing patients with and without these exposures [21]. Monteiro et al. conducted two reviews concentrating on tissue-based biomarkers in patients with oral leukoplakia [25, 29]. In their broader review, they comprehensively assessed over 40 molecular markers, including p53, Ki-67, CD133, EGFR, SOX2, Nanog, and microRNAs, among others [25]. The association of these biomarkers with cancer incidence was evaluated using immunohistochemistry, gene expression analysis, and loss of heterozygosity (LOH) studies [25]. In a more focused review, they reported on podoplanin expression as a potential indicator of malignant progression [29]. Saluja et al. reviewed biomarkers associated with cancer stem cell–like characteristics (CSCs), which are hypothesized to play a critical role in tumor initiation, resistance, and progression [27]. The markers included CD133, CD44, CD24, OCT4, SOX2, NANOG, ALDH1, and ABCG2, which were identified via specific expression patterns distinguishing CSCs from other tumor populations [27].
Overall, the umbrella review reveals a complex interplay of clinical and molecular factors that hold potential prognostic value in assessing the malignant transformation risk of OPMDs and the progression of OSCC. The diversity in measurement methods, comparator groups, and study designs highlights the need for standardized biomarker validation and longitudinal follow-up to strengthen prognostic precision in clinical practice.
3.6: Outcomes Evaluated
The included reviews encompassed a range of prognostic and clinical outcomes across the included systematic reviews and meta-analyses. The primary outcomes most frequently assessed were malignant transformation of OPMDs into OSCC, along with overall survival (OS), disease-free survival (DFS), cancer-specific survival (CSS), and recurrence rates [24, 27, 30]. Secondary outcomes included treatment response, disease progression, and the impact of conventional and molecular risk factors on prognosis.
Several reviews focused on specific prognostic indicators, such as the expression of cancer hallmarks in OLP and the differential expression of proteins detected via immunohistochemistry (IHC) [22, 23, 25]. These analyses sought to uncover early oncogenic molecular mechanisms driving malignant transformation in OLP, with the goal of identifying preventive or predictive biomarkers.
In studies evaluating transformation, a key emphasis was placed on histological outcomes such as the progression from epithelial dysplasia to invasive SCC, including the prevalence and grade of dysplastic changes (low-grade, high-grade, or absent) [25, 26, 30].
Meta-analyses specifically focusing on OLP-OSCC evaluated mortality rates, tumor staging (T1/T2, N+), histological grade, and clinical staging (I/II), and compared these with conventional OSCC cases lacking OLP history [22, 23, 28]. Covariates such as age, sex, tobacco and alcohol use, tumor location, and multiple tumor development were commonly assessed for their impact on patient prognosis.
Moreover, several reviews targeted molecular markers such as podoplanin and cancer stem cell (CSC) markers, analyzing their utility in predicting malignant transformation and distinguishing between indolent and high-risk lesions [29, 30]. These studies concluded that evaluating a panel of CSC markers—rather than relying on single-marker analysis—was more effective in stratifying progression risk in patients with OPMDs like oral leukoplakia and epithelial dysplasia.
Overall, the outcomes explored in the reviews reinforce the need for integrative prognostic models that combine clinical features, histological grading, and biomarker profiles to better predict malignant transformation and improve survival outcomes in OPMD and OSCC patients.
3.7: Summary Synthesis of Effect Measures and Statistical Outcomes
Across the included systematic reviews and meta-analyses, a diverse set of prognostic outcomes and effect estimates were reported. Various statistical measures, including odds ratios (OR), hazard ratios (HR), relative risks (RR), and pooled proportions (PP), were used to evaluate associations between clinical, lifestyle, histopathological, and molecular risk factors with malignant transformation, survival, and recurrence in OPMDs and OSCC.
Dwivedi et al. reported strong associations between tobacco exposure and the development of oral cancer (OR = 5) as well as OPMDs (OR = 15.3), both with statistical significance (p < 0.01) and high heterogeneity (I² > 80%) [24]. Rivera et al. evaluated cancer stem cell (CSC) markers, identifying ALDH1A1 and PROM1 as significantly associated with malignant transformation, with HRs ranging from 2.9 to 8.9 [26]. Saluja et al. found CSC marker expression in OPMDs to be predictive of increased OSCC risk, reporting a pooled RR of 3.31, with CD133 showing the highest individual risk (RR = 4.20) [27].
Multiple pooled analyses by González-Moles et al. revealed a global malignant transformation rate of 1.16% in oral lichen planus (OLP), with considerable heterogeneity (I² = 72.97%, p < 0.001) [22, 23]. Prognostic variables such as tobacco use, alcohol consumption, lesion site (e.g., tongue), histopathological grade, and method of diagnosis were consistently associated with increased malignant risk. Lorenzo-Pouso et al. quantified the pooled malignant development rate in oral erythroplakia at 19.9% (95% CI: 1.6–41.4), with very high heterogeneity (I² = 91.74%) [21].
Monteiro et al. evaluated tissue molecular biomarkers in oral leukoplakia and reported a pooled HR of 3.72 (95% CI: 2.40–5.76) for malignant transformation, with no observed heterogeneity (I² = 0%), indicating a highly consistent effect across included studies [25]. Other included studies also explored secondary outcomes such as disease progression and recurrence; however, these were variably reported and not uniformly defined across reviews.
Subgroup analyses, where performed, revealed additional insights into the impact of variables like lesion site, sex, age, and concurrent risk habits (e.g., alcohol and tobacco). Only 4 studies performed sensitivity analysis. Sensitivity analyses confirmed the robustness of findings in reviews by González-Moles et al. and Saluja et al., with no significant alteration of outcomes after excluding low-quality or high-bias studies [22, 27]. Overall, the evidence across reviews emphasizes the significant prognostic value of both molecular markers (e.g., CSC, podoplanin) and clinical parameters (e.g., dysplasia grade, tobacco use, lesion location) in determining malignant transformation and survival outcomes in OPMDs and OSCC.
4.1 Quality Assessment
Most included reviews used validated tools such as QUIPS (Quality In Prognosis Studies tool.), REMARK (REporting recommendations for tumor MARKer prognostic studies), Joanna Briggs Institute, and the Newcastle-Ottawa Scale to assess methodological quality. A large proportion of primary studies demonstrated moderate to high risk of bias, especially in domains related to study confounding, prognostic factor measurement, and statistical analysis and reporting, while only a few showed consistently low risk across all areas. Publication bias was assessed in most reviews using funnel plots and Egger’s tests, with no significant bias detected in key reviews by Keim-del Pino et al. [28], Dwivedi et al. [24], and González-Moles et al. [22], though some studies did not perform this assessment. REMARK-based evaluations generally reflected high reporting quality, albeit with notable omissions like sample size justification and missing data handling. Joanna Briggs-based assessments revealed variable methodological rigor, especially concerning sampling strategies and subgroup classifications.
The systematic reviews and meta-analyses collectively highlight the potential prognostic value of certain biomarkers—particularly podoplanin, LOH, and cancer stem cell markers—in predicting malignant transformation in OPMDs, including OLP, leukoplakia, and proliferative verrucous leukoplakia (PVL) [21–30]. Several studies found significant associations between these biomarkers and increased cancer risk, especially in cases with specific clinical features like erosive or atrophic lesions, tongue localization, and risk habits such as tobacco or alcohol use [22–25, 30]. However, widespread clinical implementation is hindered by substantial limitations, such as small sample sizes, lack of standardized diagnostic and reporting criteria, absence of individual participant data, inadequate follow-up durations, and high heterogeneity across studies. These limitations underscore the need for future high-quality, longitudinal studies with larger, well-defined cohorts, standardized methodologies, and detailed stratified data reporting to validate the biomarkers’ clinical utility and enhance risk stratification in OPMDs.
4.2 Quality Assessment of Included SR (AMSTAR 2)
Out of 10, 7 critically low, 1 low, 2 moderate level of quality based on AMSTAR2 assessment criteria (Appendix II). Most of included SR did not provided information regarding source of funding in the primary studies