Oral potentially malignant disorders: risk of progression to malignancy

The location of a lesion within the mouth may influence the risk of malignant transformation, but this is almost certainly related to etiologic factors and therefore may vary by geographic location and local habits. For example, in betel quid chewers, the buccal mucosa is likely to be the most affected site, whereas in reverse smokers, it may be the palate. The lateral border of the tongue and the floor of the mouth are anatomically contiguous and, together, are the most common site for OPMDs and oral cancer in the developed world, where smoking of tobacco and alcohol consumption are the most important etiologic factors. In a UK study of 630 patients with dysplastic lesions, over 95% were leukoplakias; the most common sites (42% of lesions) were the lateral and ventral tongue and the floor of mouth. In addition, lesions at this site were more likely to show severe epithelial dysplasia. Conversely, only 21% of lesions arose on the buccal mucosa, and these were mostly mild dysplasia. In a similar study in Australia, Dost et al. found that 40% of lesions arose on the tongue and the floor of mouth and that these were more likely to be dysplastic or malignant (odds ratio [OR] 2.6; P = .005). Thirty-one percent of lesions were on the buccal mucosa, but these were less likely to be dysplastic, and none progressed to malignancy. In a cross-sectional study of 3256 leukoplakias in the United States, the highest prevalence of severe dysplasia or carcinoma in situ (CIS) was in the floor of mouth (13.5%) and tongue (5%). These data suggest that these sites are at the highest risk, but studies of actual malignant transformation have shown variable findings.

In Hungary, although only 8.2% of leukoplakias arose on the tongue, these accounted for 37.5% of the lesions that underwent malignant transformation, equivalent to a transformation rate of 27% for tongue leukoplakias. The floor of mouth also had a high transformation rate, with 13% of lesions developing into cancer. In contrast, although most of the leukoplakias (63%) were found on the buccal mucosa, only 4% of these lesions progressed. Similar data have been reported in England, where 2 studies showed high transformation rates of 24% and 16% for leukoplakias in the floor of the mouth (sublingual keratosis).

In contrast, some studies have been unable to establish a strong correlation between site and malignant transformation. Schepman et al. studied 101 patients with lesions on the tongue or the floor of mouth; 15 (14.9%) developed oral cancer, compared with 5 of 65 (7.7%) whose lesions were located elsewhere, but this was not statistically significant. In a second study by Dost et al., the malignant potential of 383 dysplastic lesions in 368 patients was determined. Although the tongue (48.8% of lesions) and the floor of mouth (11.5%) together were the most common sites and the tongue had the highest transformation rate of 1.4% per year, the relationship between site and transformation was not significant. Holmstrup et al. followed up 236 patients with 269 lesions and found a malignant transformation rate of 12% for lesions treated surgically and 4% for lesions only observed. The only significant prognostic factors were size and type of lesion (homogeneous vs nonhomogeneous). The site of the lesions was not significant.

In summary, most studies and clinical papers do emphasize the lateral and ventral tongue and the floor of the mouth as areas of particular clinical concern, and this may be attributed to their overexposure to carcinogens as a result of pooling of saliva in alcohol and tobacco users.

In their systematic review, Warnakulasuriya and Ariyawardana found that on a global basis, the buccal mucosa was the most common site overall (18.4% of lesions) but had the lowest rate of malignant transformation (3.35%), whereas the tongue accounted for 16.14% of lesions but was the most common site for transformation, with a rate of 24.22%. The next most common was the combined “tongue and floor of mouth” site, at a rate of 14.85%. However, these data may hide some geographic variations because they may apply to populations with the most common habits of tobacco and alcohol use. In other populations, other sites may be more important and be associated with specific tobacco habits. For example, in Andhra Pradesh, India, 71% of leukoplakias were located on the palate and associated with reverse smoking; in Kerala, 65% were on the buccal mucosa and associated with chewed tobacco.

Oral leukoplakia varies in clinical appearance, and this has led to several definitions that may be complex and confusing. Some lesions are uniformly white or plaque-like, with a flat or wrinkled surface that may contain fine cracks or fissures (“homogeneous leukoplakia”) (Figure 1), whereas others are “nonhomogeneous.” Nonhomogeneous lesions may have various appearances, resulting in a number of clinical terms used. They may have a warty, nodular, or verrucous surface pattern (“nodular” or “verrucous” leukoplakia) or may contain red areas or be speckled (“speckled leukoplakia”) (Figure 2). Speckled lesions may also be called “erythroleukoplakia,” but this should not be confused with erythroplakia, a term that should only be used to describe lesions that are uniformly red.

There is consensus in the literature that nonhomogeneous lesions have a greater risk of malignant transformation compared with homogeneous lesions,^{,  ,  ,  ,}  but it can be very difficult to compare studies because of the different locations and different ways of reporting malignant transformation. For example, most large studies in India are population or community based, often involving house-to-house surveys, whereas most studies in the developed world are based on hospital populations or retrospective analyses of pathology records. However, a systematic review found a global leukoplakia prevalence rate of 2.6% and a malignant transformation rate of 1.36% per year. Warnakulasuriya and Ariyawardana carried out a systematic review of 24 studies and found an overall malignant transformation rate of 1.5% to 34%, but there was wide variation in the location or type of study. They found 8 studies that compared homogeneous and nonhomogeneous lesions, and all showed that nonhomogeneous lesions had the highest rate of progression. In their summary analysis, Warnakulasuriya and Ariyawardana calculated an overall transformation rate of 3% for homogeneous lesions and 14.5% for nonhomogeneous lesions (P = .001).

One of the largest series was published by Bánóczy, who summarized a series of reports^{,  ,  ,}  on a cohort of 670 Hungarian patients. Over a 30-year follow-up period, 6% (40 cases) of leukoplakias progressed to cancer. They divided the lesions into 3 clinical variants: (1) simplex (homogeneous); (2) verrucous; and (3) erosive (speckled) (the latter 2 being nonhomogeneous). None of the homogeneous lesions (simplex) progressed to cancer, whereas the transformation rate for verrucous lesions was 4.6% and that for erosive lesions was 28%. The overall rate for the nonhomogeneous group was 13.4%. Expressing these data in a different way, there were 82 erosive lesions, which accounted for 74% of the oral cancers. The remaining 26% arose from 173 verrucous lesions. Although the actual data may vary, Bánóczy's finding that verrucous and, especially, speckled lesions have the highest rates of progression has been confirmed a number of times.

In the largest Indian study to date, the overall malignant transformation rates were very low (0.3%–2.19%), but homogeneous lesions were much less likely to become malignant. In California, Silverman showed that nonhomogeneous lesions (erythroleukoplakia) transformed in 23.4% of cases compared with 6.5% of homogeneous lesions. In a series of Norwegian patients, in 14 of 157 patients, leukoplakia developed into cancer, but only 1 of these arose in a homogeneous lesion (1.6%), whereas the remainder arose from nonhomogeneous lesions (13 of 97; 13.4%). Of these, 28 were described as nodular, and 8 became malignant (28.6%). In the study by Holmstrup et al., nonhomogeneous lesions showed a 7-fold increase in the risk of progression to cancer compared with homogeneous lesions (OR = 7.0; 95% confidence interval [CI] 1.7-28.5). Of all the factors that they analyzed, the type of lesion was the most significant in predicting prognosis. A characteristic type of nonhomogeneous leukoplakia—proliferative verrucous leukoplakia—has been described and is discussed in the following sections.

The use of the terms homogeneous leukoplakia and nonhomogeneous leukoplakia has been severely criticized because this may overemphasize the importance of white lesions and divert clinicians' attention away from the more significant and dangerous red lesions. As noted earlier, the highest rates of malignant change were noted in lesions that were described as “speckled” or “erosive” or as “erythroleukoplakia”—in other words, emphasizing that these lesions are not, in fact, true “leukoplakias” (white plaques) but have a red component. The seminal work of Mashberg is well known, and although he did not primarily study OPMDs, he showed that most early oral cancers are red or have a significant red component (“erythroplasia”).^,  In his first major study on this subject, Mashberg identified 158 asymptomatic, early lesions in 125 patients. One hundred and twelve lesions were invasive SCCs, and 46 were CIS. His major finding, which formed the basis of his later work and opinions,^,  was that greater than 90% of the lesions had a red component and that only 14 (9%) were described as white (of which 9 were located on the lips). Sixty lesions (38%) were entirely red, and an additional 98 (62%) were red and white (speckled, stippled, or patchy). In later studies (reviewed by Mashberg and Samit), Mashberg confirmed these findings in a cohort of 236 asymptomatic cancers, of which only 6% were white. The remainder were entirely red (32%), had a predominant red component (32%), or were mixed (29%). In his reviews,^,  Mashberg emphasized the importance of redness as an early sign of cancer and expressed dismay at the overemphasis of “leukoplakia” as a premalignant condition, which, he claimed, is a major cause of the lack of progress in early diagnosis of cancer.

Erythroplakia is a well-defined clinical lesion,^,  but Mashberg's opinion is correct in that erythroplakia is often not included in studies of OPMDs^{,  ,  ,  ,}  —probably because it is relatively uncommon and therefore more difficult to study. In their survey of 50,915 Indian patients, Mehta et al. found 881 leukoplakias (1.73%) but only 9 cases of erythroplakia (0.02%). In the United States, only 58 cases were identified among 64,345 biopsy samples (0.09%). Reichart and Phillipsen reviewed a number of population studies that showed a prevalence rate of 0.02% to 0.83%. As suggested by Mashberg,^,  most erthroplakias show evidence of invasive carcinoma or CIS; thus, the issue of progression to malignancy is a moot point because the lesion should probably be managed as malignant at first presentation.

In Shafer and Waldron's series of 58 cases, 51% were found to be invasive cancer and 40% severe epithelial dysplasia or CIS at first biopsy. The remaining 9% were mild or moderate dysplasia. In the case of erythroplakias that do not indicate invasive cancer at first biopsy, the majority will have high-risk histologic features. The best estimates for malignant progression of these lesions can be derived from data for the malignant transformation of lesions showing severe epithelial dysplasia (including CIS). A number of relevant studies are summarized in Table II.^{,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,} 

In the study by Holmstrup et al., size was the only other factor that showed a statistical correlation with malignant transformation. Specifically, lesions greater than 200 mm² were more likely to progress compared with lesions less than this size (OR = 5.4; 95% CI 1.1-26.1). In a study of 50 patients from Northern Ireland, Napier et al. found that the risk of transformation was 6 times greater in patients with large confluent lesions that extended over more than one anatomic site compared with those with smaller localized lesions. Of 12 patients with such lesions, 7 developed oral cancer. They also found 11 patients with lesions at one anatomic site, 5 of whom developed cancer, a risk 4 times greater than in those with multiple lesions. It was notable that of the 27 patients with multiple nonconfluent lesions, only 5 progressed to malignancy. Warnakulasuriya and Ariyawardana, on the basis of evidence from their systematic review, suggested that any leukoplakia that exceeds 200 mm² has a higher risk for malignant transformation.

Most studies agree that although OPMDs are chronic and persistent, the risk of malignant change is higher within the first 5 years after diagnosis. In their systematic review, Warnakulasuriya and Ariyawardana found 5 studies that had analyzed transformation by age group. Overall, although lesions tended to progress early in their natural history, transformation rates were higher in older individuals. In a large Swedish study, involving 782 patients with leukoplakia, the highest transformation rate (6.4% in 5 years) was found in those aged 70-89 years, compared to less than 1% in all age groups below 50 years. In Bánóczy's Hungarian series, the peak incidence of leukoplakia was in the sixth decade, but the highest rates of transformation were in the seventh decade (7.1%) or in patients over the age of 71 years (8.2%). They also noted, however, a transformation rate of 7.4% in the fourth decade. In his series of 157 Norwegian patients with leukoplakia, Lind found that lesions tended to progress soon after the initial diagnosis. Of the 14 cases that progressed, 5 patients developed oral cancer in the first year and 6 within the next 8 years. In the Netherlands, it was noted by Schepman et al. that the number of transformed lesions increased with longer follow-up times. Those authors estimated that 50% of patients with leukoplakia would develop a tumor within 200 months of diagnosis, but the median time to malignant transformation was only 32 months. They found that overall, 12.4% of lesions underwent malignant transformation and that at first presentation, the mean age of the patients who developed cancer was 67.1 years, compared with 55.8 years for patients who did not develop cancer.

Large studies from India and the developing world support the view that lesions are likely to develop and progress to cancer in older individuals because the highest incidences of leukoplakia, rather than oral cancer, are consistently found in the younger age groups.^{,  ,  ,}  For example, the highest incidence of leukoplakia in Kerala was found in 35 to 54 year olds, whereas the highest incidence of oral cancer was found in the 55 to 74 year olds.

Warnakulasuriya and Ariyawardana's review identified 12 studies that had examined sex and malignant transformation. In 9 of the 12 studies, the rate of transformation was greater in females than in males, with an overall rate of 13.1% for females and 1.7% for males (P < .001). The highest rate in females was 40.6%, reported from Northern Ireland, and the lowest was found in Denmark (3.7%). In males, the highest rate was 28.6%, in the United Kingdom, and the lowest was 0%, in The Netherlands. Another study in The Netherlands, however, found 76 males with leukoplakia, of whom 4 developed oral cancer (5.3%), but of 90 females with leukoplakia, the lesions progressed to cancer in 6 (17.7%). In a Norwegian study, Lind found lesions in 102 males and 55 females, but oral cancer developed in 8 males (7.8%) and 6 females (10.9%). Nonetheless, it is still unclear why women are more predisposed to malignant transformation compared with men. Global genomic arrays may illustrate a differential gene expression that explains this sex predilection.

Despite the fact that lesions are less common in females, they have a higher risk for malignant transformation. In Sweden, Axell found a male:female ratio of 6:1, but this decreased to 5:2 among nonsmokers. In the United States, the male:female ratio was reported as 2:1. In large Indian studies,^{,  ,  ,  ,}  females were found to be much less likely to have lesions, but this was related to the fact that they were less likely to use tobacco.

Tobacco use and consumption of alcohol are well established etiologic factors for the development of OPMDs, but there is also good evidence that risk of progression is related to the practice of, or continued practice of, these habits. Lesions in nonsmokers (sometimes referred to as “idiopathic leukoplakia”) seem to be at a higher risk of progression to cancer. In Schepman's study in The Netherlands, the 3 factors associated with the greatest risk of transformation were nonhomogeneous lesions (P = .01), female sex (P < .025), and being a female nonsmoker (P < .05). There were no statistically significant associations with smoking in males or with alcohol consumption in either sex.

In another study in The Netherlands, 3 of 46 patients developed oral cancer; all were nonsmokers. In California, Silverman et al. found that lesions in smokers and nonsmokers showed quite different patterns of behavior. In their cohort of 257 patients, 45 developed oral cancer during an average follow-up time of 7.2 years. There were 133 lesions in patients who were smokers and continued to smoke; of these, the lesions transformed to oral cancer in 21 (16%) and either regressed or disappeared in 49 (37%). Of 74 nonsmokers, oral cancer developed in 18 patients (24%), but the lesions regressed or resolved in only 2 patients (3%). The remaining 50 patients stopped smoking after diagnosis; among these, 6 (12%) developed oral cancer, and in a further 22 (44%), the lesions became smaller or resolved. In a Swedish study, the cumulative frequency of oral cancer development was 3.1% over 5 years in non–tobacco users compared with only 0.4% in smokers. In her large Hungarian studies, Bánóczy found that the proportion of patients who smoked was higher (87%) among those with leukoplakia not progressing to cancer compared with those in whom carcinoma developed (78%); Bánóczy concluded that there was a greater tendency for malignant transformation in leukoplakias not associated with tobacco. In a study in England, Ho et al. found that the most significant predictor of malignant change was nonsmoking status. They studied 91 patients with histologically diagnosed dysplasia, of whom 20 were never-smokers, 29 were moderate smokers, and 42 were heavy smokers. After 5 years of follow-up, 13 never-smokers (43%) developed cancer compared with 11% and 4%, respectively, in the other 2 groups (P = .001).

These data suggest that although tobacco is an important etiologic factor for the formation of keratotic lesions, other factors must be important for these lesions to progress to malignancy. The increased risk in nonsmokers suggests that an underlying genetic predisposition may also be involved in at least a proportion of cases. It should be noted, however, that not all studies have been able to demonstrate a relationship between tobacco use and progression. Warnakulasuriya and Ariyawardana did not identify any study that analyzed smoking as a factor, and neither Dost et al. nor Holmstrup et al. found any significant associations between smoking and progression to malignancy. With regard to alcohol consumption, there is significant evidence to implicate it as an etiologic factor, but no evidence for its role as a risk factor for progression of established lesions.^{,  ,  ,  ,}  The use of betel quid and areca nut are also risk factors and is discussed under the section on oral submucous fibrosis.

Although it is established that OPMDs are statistically more likely to become malignant, it is not inevitable that every lesion will progress to cancer. Although the clinical parameters described earlier may help in clinical risk assessment, the gold standard diagnostic procedure is biopsy and histologic examination. This will enable the pathologist to rule out other specific diagnoses and to evaluate any tissue changes, the most significant of which are features of cytologic atypia and distorted epithelial architecture, which together are referred to as oral epithelial dysplasia (OED).^,  It is not inevitable, however, that a dysplastic lesion will transform into cancer, and nondysplastic lesions may also progress. Silverman et al. found that 36% of dysplastic lesions progressed to carcinoma but also that 16% of leukoplakic lesions without evidence of dysplasia progressed to cancer. Clinical or histologic biomarkers are urgently needed to improve the ability to distinguish lesions that may progress to cancer from those that will not,^,  but given the complexity of the oral carcinogenesis process and the lack of any proven predictive biomarkers in large prospective cohorts, surgical biopsy and histopathologic grading of OED is regarded as the gold standard in managing these patients.^{,  ,  ,  ,  ,} 

Historically, several terms, including dyskeratosis, epithelial atypia, squamous intraepithelial neoplasia, and squamous intraepithelial lesions, have been used to describe epithelial changes that precede the development of oral cancer. However, for lesions of the oral cavity, OED is currently the preferred term.^{,  ,  ,}  The latest 2017 WHO classification system uses a combination of architectural and cytologic features (Table III) in an attempt to provide a more objective approach to the diagnosis and grading of OED. Pathologists should use these criteria to grade lesions, but several other grading systems have been developed, and there is, as yet, no international consensus regarding which should be used.^,  A major problem is that the process of grading is subjective and has low levels of intra- and interobserver agreement.^{,  ,  ,  ,}  Many pathologists grade lesions into mild, moderate, and severe dysplasia on the basis of the extent and degree of the changes listed in Table III.^{,  ,  ,}  Mild dysplasia defines a lesion in which the architectural and cytologic changes are minimal and limited to the lower third of the epithelium, and moderate dysplasia demonstrates changes involving up to two-thirds of the full thickness of the epithelium. Severe dysplasia displays more marked changes, which may extend beyond the lower two-thirds of the epithelium. CIS is regarded as the most severe form of severe dysplasia and involves changes through the entire epithelial thickness, but in the oral cavity, even these lesions may show a degree of surface keratinization. In an attempt to improve the reliability of grading, a number of authors have developed a binary grading scheme,^{,  ,}  which is supported by the WHO, although it is recognized that further validation and evidence from clinical studies are needed. The binary system divides lesions into “high risk” and “low risk” on the basis of the number of features identified on histologic examination. High-risk lesions show at least 4 architectural features and 5 cytologic features (Table III), whereas low-risk lesions show fewer changes. In their studies, Kujan et al.^,  demonstrated that the binary system showed superior reliability and intra- and interobserver agreement compared with the 5-scale 2005 WHO grading system.

The ultimate goal of OED diagnosis and grading is to provide patients with the best management and care. Clinicians are more likely to intervene when they come across a patient with moderate or severe OED, whereas a “wait and see” policy may be adopted for lesions showing only mild dysplasia. Dost et al. demonstrated that 4.1% of mild dysplasias underwent malignant transformation and advocated surgical removal of lesions in all OED cases, irrespective of their histopathologic grade. Nevertheless, whether or not complete resection of all OED cases is warranted will not be discussed here, given that this topic is beyond the scope of our review and has been well reviewed by others.^{,  ,  ,} 

As it stands, the grade of OED is still the most important prognostic factor for malignant transformation.^{,  ,}  However, even in the case of severe epithelial dysplasia, studies have shown that the malignant transformation rate varies considerably, from 3% to 50% (Table II) and appears to be heavily dependent on study design and population characteristics. Most authors agree that the risk of progression increases with the grade of OED. A meta-analysis by Mehanna et al. pooled data from 14 nonrandomized cohort studies and estimated that the mean rate of malignant transformation was 12.1% (95% CI 8.1%–17.9%), with wide variation among the different studies (0%–36.4%) and a follow-up time ranging from 0.5 to 16 years. They also found a significant correlation between the OED grade and malignant transformation. The estimated mean progression rate of mild/moderate dysplasia was 10.3% (95% CI 6.1%–16.8%) compared with 24.1% (13.3%–39.5%) for severe dysplasia/CIS. Sperandio et al. retrospectively examined the prognostic value of OED grade in a large cohort of 1401 patients. Overall, oral cancer developed in 49 patients (3.5%) during a follow-up period of 5 to 15 years. The transformation rate of nondysplastic lesions was 0.012% (14 of 1182 patients) compared with 6% (6 of 105 patients), 18% (14 of 76 patients), and 39% (15 of 38 patients) for mild, moderate, and severe dysplasias, respectively. In the study by Warnakulasuriya et al., the severity of OED was the most significant prognostic indicator (P < .0001). The transformation rates for mild, moderate, and severe dysplasia were 4.8%, 15.7% and 26.7%, respectively. These studies strongly support OED grade as an important marker for risk of malignant progression.

Using their binary system, Kujan et al. graded 68 lesions as “high risk” or “low risk” on the basis of a combination of the architectural and cytologic features shown in Table III. Of 35 lesions, 28 (80%) graded as high risk progressed to cancer, compared with 5 of 33 low-risk lesions (15%). The binary grading system had good prognostic value, with sensitivity and specificity of 0.85 and 0.80, respectively. In a similar study, Liu et al. tested the diagnostic utility of the binary OED grading system in a cohort of 138 patients, with a mean follow-up period of 5.1 years. Of the low-grade lesions, 17 of 92 (18.5%) progressed to malignancy compared with 20 of 46 (43.5%) high-grade lesions (P = .004). They found that the degree of dysplasia was an independent marker for predicting malignant transformation, with high-grade dysplasia showing a 2.78-fold (95% CI 1.44-5.38) increased risk of cancer progression. Nankivell et al. further validated the reliability and prognostic usefulness of the binary OED grading system. These studies showed that the binary system may have better prognostic value and superior reproducibility compared with the current more widely used 3-scale grading system. With further evaluation, it is believed that the binary system will become the standard grading scheme.^,  There is also good evidence that the accuracy and prognostic reliability of dysplasia grading can be improved with careful training and consensus reporting by at least 2 pathologists.^{,  ,  ,  ,}  Nevertheless, as indicated in Table II, epithelial dysplasia, whether reported as “severe” or “high risk” is strongly associated with an increased risk of malignant progression.

Several studies, however, have not been able to find a relationship between epithelial dysplasia and malignant progression. Warnakulasuriya and Ariyawardana found 5 reports that had correlated OED grade to malignant transformation, only 3 of which found a significant relationship.^{,  ,}  Others have not been able to find a statistical correlation and have shown that nondysplastic or low-grade lesions may also progress to cancer^{,  ,  ,}  This contradicts the view that leukoplakic lesions without dysplasia or mildly dysplastic lesions are entirely harmless and has led to calls to drop the “wait and see” approach in favor of more active surgical intervention in the management of OPMDs.^{,  ,} 

Despite these uncertainties, the majority of the literature supports the view that OED carries a significant risk for malignant transformation and that it must be considered in the process of managing the clinical risk of progression of OPMDs. However, clinicians should acknowledge that pathologists report on the tissue they receive and that histopathologic findings are highly dependent on correct and representative sampling of the lesion. Holmstrup et al. compared the biopsy diagnosis and the final diagnosis after surgical removal in 101 OPMDs. In 35% of cases, they found that the biopsy had underdiagnosed the extent of disease, including 7 cases of SCC that had shown only moderate, mild, or no dysplasia on incisional biopsy and may have been regarded as “low risk” lesions. The authors concluded that histologic grading of OED on biopsies is unreliable and that other factors must be considered for the prediction of malignant transformation of OPMDs.

The Holy Grail in terms of risk assessment is to discover a biomarker that can be used in a histologic or chairside test to predict malignant transformation of oral lesions. There have been many studies investigating various potential markers, but to date, no single biomarker has proved to be of clinical value. In a systematic review, Smith et al. found more than 2500 publications addressing the issue of biomarkers in dysplasia, but only 13 met the criteria of longitudinal design with adequate follow-up and well-defined diagnostic criteria. Overall, they found that 113 biomarkers had been analyzed. The most common of these was p53 (found in almost 90 papers and in 6 of the 13 included studies), followed by proliferation markers (Ki67 and PCNA; 20-40 papers). Other markers, including cell cycle proteins, loss of heterozygosity (LOH), and a range of cell surface and stromal proteins were studied in relatively few papers. The authors recognized the limitations of a review of this sort and the variability in the quality of identified papers. Nevertheless, they were able to suggest that LOH, survivin, matrix metalloproteinase-9, and DNA content may be associated with risk of progression. Overall, however, they concluded that there is currently no strong evidence for the use of any biomarker in the prognosis of OPMDs.

Others have come to similar conclusions and have been unable to identify any markers that have yet been fully evaluated or are suitable for use in a routine diagnostic setting.^{,  ,  ,  ,}  Many studies have examined various aberrations in the genome as potential markers of progression (reviewed in some reports^{,  ,  ,}  ). These include LOH and gene expression signatures, but no single aberration has been found to be predictive. Analysis of DNA content (ploidy) provides a simple measure of gross genetic aberrations and is known to be highly associated with malignancy.

Analysis of DNA aneuploidy may be one potentially useful biomarker that has yet to be fully exploited. Recently, Alaizari et al. undertook a meta-analysis to determine whether aneuploidy is a useful marker to predict malignant progression in OPMDs. They identified 5 studies that had assessed the predictive value of ploidy analysis^{,  ,  ,  ,}  and found that aneuploid lesions had a 3.12-fold (95% CI 1.86-5.24) increased risk of malignant transformation. All studies also showed that aneuploidy was associated with increasing severity of dysplasia. Sperandio et al. for example, found a significant correlation between dysplasia grade and DNA ploidy (P < .001). Forty-nine of 110 cases (44.5%) of severe dysplasia were aneuploid compared with only 14% of mild dysplasias and 9.5% of nondysplastic lesions. Of 32 lesions that progressed to malignancy 20 (62.5%) were aneuploid compared with only 39 of 241 (16.2%) lesions that did not progress. Torres Rendon et al. found that 14 (33.3%) of 42 OED lesions that progressed to cancer were aneuploidy, compared with only 5 (11.3%) of 44 that did not (P = .01). Of 19 OED lesions that were aneuploid, 74% showed malignant progression, compared with only 42% of the diploid lesions. Although further work and prospective trials are needed, these studies suggest that aneuploid dysplastic lesions have a higher risk of malignant progression. Aneuploidy can now be measured by using image cytometry, and its application to cytology samples may soon result in rapid and reliable noninvasive tests for risk assessment of oral lesions.

In a meta-analysis of studies up to 1997, Miller and Johnstone found that human papilloma virus (HPV) was more likely to be detected in precancerous lesions than in normal oral mucosa. HPV was detected in 10% of normal tissues compared with 22.2% of nondysplastic leukoplakias and 26.2% of dysplastic lesions. Subsequently, the relationship between high-risk HPV and oropharyngeal SCC has become well established, but the role of HPV in OPMDs remains uncertain. HPV-associated epithelial dysplasia is now well recognized, and increasing numbers of cases are being reported.^,  However, there are few studies with long-term follow-up, and the risk of progression of HPV positive lesions is not yet known. Recently, however, a study by Lerman et al. suggested that HPV-positive OED may have a high risk of malignant progression. They undertook a detailed analysis of 53 cases of HPV-associated OED. All cases showed the characteristic histologic features associated with HPV infection, and most (88.7%) were found to arise in males and involved the tongue or the floor of the mouth (77%). Eight of the 53 cases (15%) were associated with invasive SCC. HPV-associated epithelial dysplasia consistently shows an association with infection by high-risk HPV (HPV-16).^{,  ,}  The gold standard for diagnosis is to demonstrate HPV infection by in situ hybridization (ISH), but p16 expression may be a good surrogate marker for high-risk HPV types. Woo et al. and Lerman et al. found that all lesions that were HPV positive on ISH also showed strong p16 expression on immunocytochemistry. However, although p16 expression is sensitive, it is not specific because HPV-negative epithelial dysplasias may be p16 positive.^,  In the study by McCord et al., 11 cases of high-grade dysplasia showed strong diffuse positivity for p16, but only 7 of these were HPV positive on ISH. Although p16 is now accepted as a surrogate for HPV expression in oropharyngeal carcinoma, further work is needed to determine its value in the diagnosis of HPV-associated epithelial dysplasia.

Proliferative verrucous leukoplakia (PVL) is a multifocal, recurrent, and exophytic variant of leukoplakia with a high rate of malignant progression. In a systematic review and meta-analysis, Abadie et al. examined 23 reports with follow-up data. They found a malignant transformation rate of 63.9%. Most were single case reports, but of the 8 that reported on a series of more than 10 cases, the rate of progression varied from 33% to 100%.

PVL is diagnostically challenging and can often only be diagnosed retrospectively after careful clinical and pathologic correlation shows a history of persistent or recurrent and multifocal lesions. Early lesions are often flat, plaque-like leukoplakias, but as the disease progresses, lesions become multifocal (proliferative) and increasingly exophytic or nonhomogeneous (verrucous). Current evidence indicates a higher frequency in older females (>60 years of age), with the gingiva being the most commonly involved site, followed by the buccal mucosa and the tongue.^{,  ,  ,}  Gingival and palatal lesions are also the most likely to undergo malignant transformation.^{,  ,}  The etiopathogenesis is poorly understood, and there appears to be no correlation with alcohol consumption, tobacco chewing, smoking, or HPV infection.^{,  ,}  The overall mortality rate is also somewhat unclear, with some studies reporting a 5-year survival rate of 60%, whereas others have suggested it to be less aggressive. In their review, Pentenero et al. estimated an overall mortality rate of 30%.

Histologically, PVL can show a range of features that may change over time. These include simple hyperkeratosis, as well as verrucous hyperplasia with a verruciform surface pattern and wide, bulbous rete pegs. However, although there are prominent architectural changes, cytologic atypia is minimal, and conventional OED grade cannot be used as a predictive marker. A “pushing” invasive front, below the level of the basement membrane of the adjacent normal mucosa, is suggestive of progression to verrucous carcinoma, but lesions often progress to conventional invasive oral SCC. Immunocytochemical staining for proliferation markers (MCM-2) and DNA ploidy analysis have been suggested to aid in the prediction of malignant transformation. Because of their extent, lesions are usually excised conservatively, but recurrence rates remain high (>70%), even if a more radical approach is undertaken.^{,  ,}  The average time for malignant transformation is estimated to be 5 to 6 years after the initial presentation. Close follow-up and repeat biopsies are essential to ensure early diagnosis and appropriate treatment.