Clinical and Histologic Features Associated With Malignant Transformation of Oral Cavity Dysplasia Managed at a Single North American Institution

Key Points

Question  What demographic, social, clinical, physical, and histologic variables are associated with malignant transformation (MT) of oral dysplastic lesions (ODLs)?

Findings  In this cohort study of 264 ODLs in 241 patients, 14% of ODLs underwent MT; 50% underwent transformation by 424 days, and 90% by 1600 days. Nodularity, friability, and mass effect were commonly observed in malignant lesions, and multiple abnormal sites were associated with female sex and a higher percentage of MT.

Meaning  Nodularity, friability, and mass effect were associated with lesions that push, displace, or invade surrounding tissue, and were commonly seen in ODLs that ultimately progressed to cancer; there may be little utility in continuing surveillance beyond 5 years.

Importance  Reports characterizing clinical and histologic features associated with a higher risk for development of malignant lesions in the background of an oral potentially malignant disorder have largely reflected East Asian populations. Long-term studies among the North American population are rare.

Objective  To evaluate risk of malignant transformation (MT) of oral dysplastic lesions by investigating the demographic, social, clinical, and histologic factors that may be associated with an increased rate as well as a decreased time to MT.

Design, Setting, and Participants  This was a retrospective cohort study with medical record review at a single institution from January 1, 2000, to December 31, 2019, with follow-up for 20 years. Patients were excluded if they were younger than 18 years, the first biopsy diagnosis showed cancer, biopsies were never performed, biopsies were taken from sites outside of the oral cavity, there was no additional follow-up after the first visit, or the biopsy specimen was not characterized on the spectrum of dysplasia.

Exposures  Diagnosis of leukoplakia of oral mucosa, unspecified lesions of oral mucosa, or other disturbances of oral epithelium.

Main Outcomes and Measures  Main outcome measures included MT rate and time to MT as well as demographic, social, clinical, physical, and histologic features associated with MT.

Results  Thirty-eight of 264 lesions (14%) in 241 patients (132 men and 109 women; mean [SD] age, 64 [13] years) underwent MT. Of the 38 lesions that underwent MT, 19 (50%) underwent transformation by 424 days, 28 (75%) by 870 days, and 34 (90%) by 1600 days. Nodularity, friability, and mass effect were more commonly observed in malignant lesions (nodularity: 42.9% vs 10.0%; difference, 32.9%; OR, 6.72; 95% CI, 3.03-14.89; friability: 42.9% vs 12.8%; difference, 30.1%; OR, 5.11; 95% CI, 1.66-15.69; mass effect: 54.2% vs 10.4%; difference, 43.8%; OR, 10.16; 95% CI, 4.12-25.09). Men were less likely than women to have multiple lesions in the oral cavity (OR, 0.40; 95% CI, 0.22-0.74). Having multiple abnormal sites was associated with higher percentage of MT (OR, 3.38; 95% CI, 1.63-7.01).

Conclusions and Relevance  In this cohort study, nodularity and mass effect were associated with lesions that may push, displace, or invade surrounding tissue, and these were more likely to be present in oral dysplastic lesions that ultimately progressed to cancer. There may be less utility in continuing surveillance beyond 5 years.

The mucosal lining of the oral cavity represents a dynamic environment of constant mucosal turnover as a means of providing a barrier against toxic, mechanical, and environmental stressors.¹ Hence, mucosal changes of the oral cavity often arise and vanish without lasting sequelae. While most oral cavity lesions are benign, head and neck surgeons should distinguish physical characteristics, in conjunction with social risk factors, that raise suspicion for presence of an oral potentially malignant disorder (OPMD) or cancer itself. Biopsy of these suspect lesions is the definitive means to assign a diagnosis of cancer; however, certain histologic diagnoses represent a category considered OPMD because they exhibit characteristics that may predispose to a risk of malignant transformation (MT).² Oral dysplastic lesions (ODLs), the focus of this study, represent a specific subset of OPMDs that are described based on the degree of dysplasia.

Several studies have attempted to characterize clinical and histologic features of OPMD that demonstrate a higher risk for developing a malignant lesion. Reported transformation rates have ranged from 2% to 8%,^3-5 with follow-up times as long as 5 years. The tongue is the most common site in the oral cavity to experience transformation to squamous carcinoma,⁴ and severity of dysplasia as well as increasing age have been shown to be significant predictors of MT.⁵ Exophytic verrucous hyperplasia, lichen planus, and oral submucosal fibrosis have variously been proposed as harboring the greatest risk of transformation.^3,5,6 A 2020 systematic review and meta-analysis⁷ found different rates of MT among different subgroups of OPMD, with the highest rates found among those characterized as either oral erythroplakia (33.1%) or proliferative verrucous leukoplakia (49.5%).

Many of these studies were conducted among East Asian populations with a high incidence of areca/betel nut use in addition to tobacco and alcohol consumption.^8-10 Among the North American population, long-term follow-up studies of ODLs are rare. Screening interventions can improve early detection and confer a survival benefit in patients with OPMD¹¹; thus, there may be significant implications in developing a surveillance regimen in patients with OMPD.

This study primarily aimed to evaluate risk of MT of ODLs, a subset of OPMD, at a single institution by investigating the demographic, social, clinical, and histologic factors that may be associated with an increased rate as well as a decreased time to MT. Because the decision to pursue biopsy relies on physical findings that elicit suspicion, particular attention was paid to defining associations between physical findings and histologic diagnosis of carcinoma.

A retrospective medical record review was conducted by identifying patients who presented to the principal investigator’s (J.A.R.) clinic with the diagnosis of leukoplakia of oral mucosa (International Classification of Diseases, Ninth Revision [ICD-9] 528.6 or International Statistical Classification of Diseases and Related Health Problems, Tenth Revision [ICD-10] K13.21), unspecified lesions of oral mucosa (ICD-9 528.9 or ICD-10 K13.70 or K13.79), or other disturbances of oral epithelium (ICD-9 528.79 or ICD-10 K13.29) between January 1, 2000, and December 31, 2019. A broad category of ICD diagnoses was used to capture as many oral mucosal lesions as possible in the initial review. All patients were referred for concern for oral lesions that could progress to cancer; the practice is based at a National Cancer Institute–designated comprehensive cancer center. All pathology slides were reviewed by the institution’s pathology department. This study was approved by the Fox Chase Cancer Center Institutional Review Board. Informed consent was waived because all data were deidentified.

Patients were excluded from the study if they were younger than 18 years, the first biopsy diagnosis was cancer, biopsies were never performed, biopsies were taken from sites outside of the oral cavity, there was no additional follow-up after the first visit, or the biopsy specimen was not characterized on the spectrum of dysplasia. For each patient identified, demographic variables, including sex, race and ethnicity, and age, were collected, as well as clinic and biopsy dates. Social habits, including tobacco smoking, use of smokeless tobacco, and alcohol use, were recorded.

Physical findings describing the oral lesion were characterized in a consistent manner by the principal investigator (J.A.R.) and described based on the presence or absence of leukoplakia, erythroplakia, friability, nodularity, ulceration, or mass effect. Leukoplakia and erythroplakia were defined as white and red patches involving oral mucosa, respectively. Lesions that bled easily or fragmented with gentle manipulation were termed friable. Nodularity was characterized as an abnormal bosselation of the mucosa, while ulceration described lesions that eroded the surface layer of the mucosa. Mass effect was used to describe lesions that pushed or displaced nearby structures.

While some physical examination findings are mutually exclusive, such as leukoplakia and erythroplakia, other descriptions can exist concurrently within a lesion. Therefore, additional groupings of lesions by combined physical characteristics were also investigated. Specifically, leukoplakia in conjunction with friability, nodularity, ulceration, or mass effect was examined, as well as erythroplakia with friability, nodularity, ulceration, or mass effect.

The subsite of the oral cavity in which the lesion presented was recorded. If patients had multiple lesions, characteristics were recorded for all subsites. The histology of each lesion that was biopsied was characterized as either no dysplasia, mild dysplasia, moderate dysplasia, severe dysplasia (carcinoma in situ), or invasive carcinoma. Some pathology results were characterized as simply “no dysplasia.” In other cases, there were further descriptions of histologic features that did not fit the spectrum of dysplasia.

The dates of each biopsy were recorded. If patients had multiple biopsies without progression, only the initial documentation of dysplasia or of a worsening in degree of dysplasia was recorded. In the principal investigator’s practice, patients were systematically observed every 6 to 8 weeks for a year, every 3 months for 6 months, every 4 months for 8 months, and then every 6 months thereafter. Patients were instructed in monthly oral self-examination; reasons to report were reviewed at each visit. If a change in clinical condition that warranted biopsy confirmed identified progression, the follow-up schedule returned to that described previously, beginning with 8-week intervals. No effort was made to remove an entire lesion that did not represent at least carcinoma in situ/severe dysplasia if it lacked nodularity, friability, erythroplakia, or mass effect. If an initial biopsy was thought to be unrepresentative of a residual abnormality, then the lesion was excised. No additional detection technologies were used.

All patients with severe dysplasia or invasive carcinoma were treated with definitive resection. Because severe dysplasia is characterized as carcinoma in situ, transformation from severe dysplasia to invasive carcinoma could not be investigated.

Standard descriptive statistics were used to characterize the study population. Although data were collected individually for each patient, for the few patients with suspect lesions in more than 1 subsite of the oral cavity over their course of surveillance, each lesion was recorded as a separate data point. If lesions occurred in subsites that were adjacent to each other at the same time of diagnosis, these were not considered separate subsites. Only synchronous nonadjacent lesions or metachronous lesions were registered in more than 1 subsite.

Transformation was described as a change in pathology from either no dysplasia to mild, moderate, or severe dysplasia; from mild dysplasia to moderate or severe dysplasia; or from moderate dysplasia to severe dysplasia. Whether there was an association with transformation upstaging by more than 1 pathological degree was also assessed. The distribution of time to any progression was characterized using Kaplan-Meier methods.¹² Individuals whose lesion(s) did not progress were censored at the time of last follow-up.

Wilcoxon rank sum tests and Fisher exact tests were used to assess the associations between the presence of transformation, represented as a binary variable, and demographic, social, physical, and histologic factors as described previously. Odds ratios (ORs) with 95% CIs were also described. Because a large majority of patients were White (210 of 241 [87%]), statistics were not undertaken comparing ethnicity and race. Ordinal logistic regression was used to assess the associations between the ordinal measure of transformation with these same factors. Mann-Whitney tests were used to evaluate associations between time to transformation with demographic, social, physical examination, and histologic variables. All tests were 2-sided with a 5% type I error. All data were recorded using Microsoft Excel (Microsoft Corporation), and statistical analyses were performed with SPSS Statistics, version 20 (IBM Corporation).

A total of 841 patient medical records were identified between January 1, 2000, and December 31, 2019, with the diagnosis of leukoplakia of oral mucosa, unspecified lesions of oral mucosa, or other disturbances of oral epithelium. Of these, 321 were excluded because the first biopsy diagnosis was cancer, 28 because biopsies were never performed, 98 because biopsies were taken from sites outside of the oral cavity, 45 because there was no additional follow-up after the first visit, and 101 because the biopsy specimen was not characterized by the spectrum of dysplasia. An additional 7 patient medical records were excluded because there was insufficient information to determine whether the patient fit the inclusion criteria. After the aforementioned criteria were applied, 241 patient medical records remained (Figure 1). The median (IQR) duration of follow-up since initial biopsy was 21 (10-42) months.

There were 132 men and 109 women. Ages ranged from 21 to 92 years (mean [SD], 64 [13] years). The majority of patients were White (n = 210 [87.1%]). There were 4 African American patients (1.7%), 6 Asian/Pacific Islander patients (2.5%), and 1 Hispanic patient (0.4%). The ethnicity was not disclosed in the rest of the population.

Of the 241 patients, 227 (92.4%) had only 1 lesion of the oral cavity while 14 (5.8%) had more than 1 lesion in separate subsites of the oral cavity during their entire follow-up time. There were 264 lesions overall.

Most lesions were located within the tongue subsite of the oral cavity (n = 157 [59.5%]). Other subsites included the floor of mouth (n = 34 [12.9%]), buccal mucosa (n = 31 [11.7%]), alveolar ridge (n = 29 [11.0%]), hard palate (n = 10 [3.8%]), retromolar trigone (n = 2 [0.8%]), and mucosal lip (n = 1 [0.4%]).

Thirty-eight of 264 lesions (14.4%) underwent MT, of which most (n = 24) transformed from moderate dysplasia to carcinoma in situ or invasive carcinoma. However, some underwent MT from no dysplasia to cancer (n = 4) or from mild dysplasia to cancer (n = 10). A majority of patients who underwent MT experienced progression to carcinoma in situ (n = 16) or tumor category pT1 (n = 16). One lesion progressed to pT2, 3 lesions progressed to pT3, and 2 lesions progressed to pT4a. Both abnormalities that progressed from dysplasia to T4a originated on the mandibular alveolus. These patients had been monitored prior to detection of cancer and were only categorized pT4a after definitive assessment of a malignant lesion demonstrated marrow involvement. One of the lesions that progressed from dysplasia to T3 did so after an interval greater than a decade; the patient did not attend scheduled follow-ups. The other patients with lesions that progressed to T3 were seen regularly, but the patients insisted that they experienced a sudden change in examination only shortly before their scheduled follow-up appointments.

Of the 38 lesions that underwent MT, 19 (50%) had transformed by 424 days (13.9 months), 28 (75%) by 870 days (28.6 months), and 34 (90%) by 1600 days (52.6 months). When transformation was examined by 180-day (6-month) intervals, 1080 days (35.5 months) was the first time point at which the number of new transformed lesions approached an asymptote of 0. By 1080 days (35.5 months), 79% of lesions that transformed had done so (Figure 2). The median time to MT for any lesion was 436 days (14.3 months) (range, 58-3911 days [1.9-128.6 months]). A total of 226 lesions never underwent MT with an average follow-up time of 36 months (range, 1-410 months). However, 11 lesions that did not undergo MT still progressed to a more severe form of dysplasia. The mean (SD) time to transformation was 839.3 (1148.1) days for mild dysplastic lesions and 766.5 (893.8) days for moderate dysplastic lesions.

Men were less likely than women to have multiple lesions in the oral cavity (14.2% vs 29.2%; difference, 15.0%; OR, 0.40; 95% CI, 0.22-0.74). Physical characteristics of oral cavity lesions that did and did not undergo MT as well as ORs with 95% CIs are summarized in Table 1.

Malignant transformation did not appear to be associated with age (66.3 years in MT group vs 63.6 years in no MT group; difference, 2.7 years; 95% CI of difference, −7.1 to 2.5 years). There was no significant association between malignant progression and current tobacco use through smoking or with alcohol use. There were too few patients having used smokeless tobacco to be included in a meaningful analysis.

Lesions with leukoplakia were less likely to undergo MT than lesions without leukoplakia (12.1% vs 20.3%; difference, 8.2%; OR, 0.54; 95% CI, 0.26-1.11). Lesions demonstrating erythroplakia more often underwent MT (36.4% vs 13.4%; difference, 23.0%; OR, 3.68; 95% CI, 1.02-13.24). Nodularity, friability, and mass effect were more commonly observed in lesions that progressed to cancer than in those that did not transform to cancer (nodularity: 42.9% vs 10.0%; difference, 32.9%; OR, 6.72; 95% CI, 3.03-14.89; friability: 42.9% vs 12.8%; difference, 30.1%; OR, 5.11; 95% CI, 1.66-15.69; mass effect: 54.2% vs 10.4%; difference, 43.8%; OR, 10.16; 95% CI, 4.12-25.09). Ulceration was observed in both lesions that did and did not undergo MT (21.9% vs 13.3%; difference, 8.6%; OR, 1.82; 95% CI, 0.72-4.55).

Combined features of leukoplakia with friability, nodularity, or mass effect were also more frequently observed in lesions that underwent MT (Table 2; friability: 44.4% vs 2.2%; difference, 42.2%; OR, 5.20; 95% CI, 1.33-20.33; nodularity: 40.0% vs 8.0%; difference, 32.0%; OR, 5.33; 95% CI, 2.31-12.31; mass effect: 57.9% vs 3.5%; difference, 54.4%; OR, 11.10; 95% CI, 4.11-30.02). The same analysis could not be performed with features of erythroplakia because instances in which these characteristics were observed in combination with ulceration (n = 0), friability (n = 3), nodularity n = 2), or mass effect (n = 1) were rare. There was no meaningful difference between the rates in which leukoplakia and ulceration were observed in malignant and benign lesions.

It was most common for moderate dysplasia to transform to carcinoma in situ or invasive carcinoma (n = 24). Only 10 underwent MT from mild dysplasia, and even fewer from no dysplasia (n = 4).

Time to transformation was not associated with age, sex, tobacco use, or alcohol use. There were no significant differences between time to transformation based on leukoplakia, erythroplakia, nodularity, friability, ulceration, and mass effect.

Thirty-seven of 264 lesions (14.0%) occurred in conjunction with another OPMD at or during the course of the patient’s follow-up. Mucosal changes that involved more than 1 subsite had a higher percentage of MT (28.6% vs 10.6%; difference, 18.0%; OR, 3.38; 95% CI, 1.63-7.01). There was no meaningful difference in the time to transformation between lesions that occurred in isolation and those that occurred in conjunction with others.

Despite advances in the understanding of OPMDs, there remains much to be learned regarding risk stratification, appropriate follow-up schedules, and management options for these lesions. This study highlights the importance of identifying risk factors that may be associated with MT in patients with ODLs. Malignant transformation in our study was similar to that reported in prior studies.^13,14

Oral dysplastic lesions were more frequently observed in men, which is consistent with the higher incidence of oral cavity cancers in men.¹⁵ However, ODLs arising in women were more likely to present in conjunction with other lesions of the oral cavity. There were no significant associations between progression of ODLs to cancer and advanced age or social habits. Physical characteristics of each lesion were important in defining clinical features as particularly worrisome. Nodularity, friability, mass effect, and erythroplakia were more likely to be present in ODLs that ultimately developed MT than in ODLs that never progressed to cancer. While there was no meaningful difference between the rates in which leukoplakia and ulceration were each observed in malignant and benign lesions, lesions with leukoplakia in combination with nodularity, friability, or mass effects were more likely to undergo subsequent MT compared with lesions without a combination of these characteristics.

It is important to know the patient population and social habits in which physical characteristics may inform differing natural history. In this study’s predominantly White population, leukoplakia was the only physical examination finding to be more commonly seen in lesions that did not ultimately undergo MT. By contrast, leukoplakia is a more concerning finding among the Indian population, in which chewing areca nut is a common practice.¹⁶ Approximately 16.5% of men in India chew betel nut,⁹ a habit that contributes to the development of oral cavity cancers (hazard ratio, 2.2).¹⁰ However, in the current study’s population, erythroplakia was a more concerning finding that demonstrated an association with MT.

While features such as friability and ulceration can certainly be markers for cancer, they are also commonly seen in the background of mucosal inflammation or irritation. Nodularity and mass effect were associated with lesions that may push, displace, or invade surrounding tissue, and these were (not surprisingly) more likely to be present in the ODLs that ultimately progressed to cancer.

The presence of multiple ODLs in nonadjacent subsites of the oral cavity, whether as synchronous or metachronous lesions, was associated with female sex and showed a significantly higher percentage of MT. Such patients may benefit from more intense follow-up schedules.

Studies with long-term surveillance of ODLs are rare and give little insight into the virtues of observation beyond a certain time frame. In the present study, the overwhelming majority of ODLs (approximately 90%) underwent transformation by the 1600-day mark; there may be diminishing returns from surveillance beyond 5 years if patients know reasons to report.

There are several limitations to this study. Histologic features of lesions were only categorized by degrees of dysplasia. Lesions characterized as “no dysplasia” (such as those with submucosal fibrosis or lichenoid changes) were not additionally addressed and are not the subject of this report.

According to the systematic protocol, patients who did not show interval progression short of a malignant lesion experienced longer follow-up intervals than those with an upstaging episode. Hence, there is a selection bias because lesions of lower severity ultimately experienced less-intense surveillance; time between observations lengthened when lesions did not progress in severity. While most patients strictly adhered to scheduled follow-up times, there were instances in which patients presented earlier than scheduled if they developed symptoms that worried them. Another limitation was the failure to capture symptom status and patient report of physical examination findings at the time of each office visit. Knowledge of the iatrotropic stimulus¹⁷ (new symptoms, change in physical findings, or routine screening examination) that motivated medical attention is important to understand the value of screening. Without such information, we are unable to conclude if routine screening is beneficial beyond follow-up in response to symptoms or physical examination findings.

This study was unable to conclude whether regularly scheduled visits, rather than visits prompted by worrisome signs or symptoms, led to improved oncologic outcomes. This information would provide important insight on the utility of our current follow-up regimen.

South Asian patients were not separately distinguished in patient demographics. They represent an unknown, but small, percentage of patients who may have been at higher risk of progression.

Physical examinations and description of these lesions were performed and characterized by the principal investigator, which may render the experience less generalizable to other practices and institutions. Some features, such as nodularity and mass effect, are subjective descriptors that may vary among physicians. These descriptors are not consistently defined by international standards, and this study attempts to describe each of these features as used in our analysis and practice. Nonetheless, because this report is based on an individual’s personal experience that spanned more than 3 decades, the consistency of documenting physical descriptions and adhering to a set follow-up algorithm constitute strengths. In fact, most of the lesions that progressed to cancer were detected as carcinoma in situ or T1 cancers. Without more data for each oral cavity subsite, this series cannot distinguish whether transformation time differed significantly among subsites.

An association of MT with continued tobacco and alcohol use was not demonstrated in this study cohort. Perhaps once dysplastic lesions have developed, the mucosal injury is already sufficiently profound such that additional exposures constitute only small added insults. Although we documented pack-year and smoking rate as well as drinks per week at initial visits, subsequent tobacco and alcohol use was not systematically recorded. In addition, some patients who deny substance abuse may be dissembling. To show an association between MT and ongoing exposure to tobacco and alcohol would require a larger set of patients with more robust data acquisition.

In this cohort study, findings suggest that ODLs that exhibit erythroplakia, nodularity, and mass effect should prompt meticulous follow-up in view of their likelihood of malignant progression. They represent a potential population for research to define propensity to progress. Because approximately 90% of ODLs progressing to cancer did so by 1600 days, there may be less utility in continuing surveillance beyond 5 years.

Accepted for Publication: August 29, 2021.

Corresponding Author: Jessica A. Tang, MD, Department of Otolaryngology–Head and Neck Surgery, Temple University, 3440 N Broad St, Kresge West, Ste 300, Philadelphia, PA 19140 (jessica.tang@tuhs.temple.edu).

Author Contributions: Dr Tang and Ms Amadio had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Tang, Ridge.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: All authors.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Tang.

Administrative, technical, or material support: Amadio, Ridge.

Supervision: Ridge.

Conflict of Interest Disclosures: None reported.