Malignant transformation of oral leukoplakia: a retrospective cohort study of 218 Chinese patients

Oral squamous cell carcinoma (OSCC) is widely recognized as the most common type of head and neck cancer, with a ~50% survival rate over 5 years despite various treatments in the past three decades [1, 2]. Oral leukoplakia (OL) is defined as "A white plaque of questionable risk having excluded (other) known diseases or disorders that carry no increased risk for cancer", which is the best-known potentially malignant disorder of the oral mucosa [3, 4]. Reports indicate that 15.8-48.0% of OSCC patients were associated with OL when diagnosed [5‐8]. Pooled estimate of annual rate of OL malignant transformation is 1.36% (95% confidence interval, 0.69%-2.03%) in various populations and geographical areas [9]. Possibly due to the rarity of cohort studies in developing nations, a concrete conclusion on the global annual rate of OL transformation is not currently available [10].

Histopathologically, oral epithelial dysplasia currently is the most important prognostic indicator for determining the malignant transformation risk of OL [11]. Traditionally, OL lesions are classified as non-dysplasia (hyperplasia) and dysplasia (mild, moderate or severe). At a workshop coordinated by the WHO Collaborating Centre for Oral Cancer and Precancer in the United Kingdom on issues related to oral potentially malignant disorders, a binary classification (no/questionable/mild-low risk; moderate/severe-high risk) of the lesion was recommended. The move was intended to reduce the subjectivity inherent in grading dysplasia, which may increase the likelihood of agreement between pathologists [12]. Kujan et al. recently tested the new binary system of grading oral dysplasia and supported this view [13]. However, the biological significance of this grading system needs to be researched in longitudinal studies to explore its relevance in predicting malignant transformation risk of epithelial precursor lesions [12].

The risk factors of clinical features and lifestyle habits associated with transformation of OL into carcinoma have been evaluated in previous studies. Napier and Speight recently reviewed clinical predictors of malignant transformation in oral leukoplakia, such as age, gender and lesion site, but the results from different study populations vary [10]. The role of smoking and ethanol intake as important risk factors related to malignant transformation remains controversial and unclear [14‐16]. Thus, assessment of these factors for OL malignant transformation is still needed.

The objective of the present study is to estimate the malignant transformation rate of a retrospective cohort of 218 patients with OL (mean follow-up of 5.3 years), explore the usefulness of the new binary system of grading dysplasia and identify significant risk factors of OL malignant transformation in China.

In the present study, we evaluated the malignant transformation rate of Chinese patients with OL, and identified the risk factors of malignant transformation during a relatively long follow-up period. Of the 218 cases, 39 (17.9%) patients developed invasive cancer, with a mean malignant transformation period of 5.2 years. Dysplasia was an independent risk factor for OL malignant transformation, but age, gender, site, diet habit, smoking and ethanol intake were not risk factors (Table 3).

In our series, according to the WHO criteria, we considered the time elapsed from the initial diagnosis of OL to the development of cancer. In this context, we excluded patients with diagnosis of OL concomitant OSCC at the first visit or patients with a followed-up period of less than 12 months after the initial diagnosis of OL. Our recorded annual malignant rate of 3.38% is higher than the rate documented in the literature (0.69%-2.03%) [9‐11]. Herein, various treatments on the OL patients were not considered because few prevention studies have shown effectiveness in preventing the transformation of OL to cancer [18].

Although it is a well-known fact that the histological classification of OL lesions is imperfect, which may involve subjectivity, we cannot do without it to date [12]. Notwithstanding it had been elucidated that oral lesions with epithelial dysplasia more often develop into cancer than those with hyperplasia in previous findings [11, 14, 19, 20], few studies have examined the risk of malignant development in different grades of dysplastic OL. In an American population, moderate or severe dysplastic OL was associated with 2.30-fold increased risk of malignant transformation, compared with mild dysplasia or hyperplasia [21]. In a Dutch study [22], OL diagnosed with moderate or severe epithelial dysplasia had a significantly higher risk of malignant development than leukoplakia with lower dysplasia grades (P < 0.01). These findings were similar to ours. The degree of dysplasia was associated with OL malignant transformation (Table 3).

According to the new binary grading system proposed by WHO, we found the high-risk dysplastic OL was associated with a 4.57-fold (P < 0.001) increased risk of malignant transformation, when compared to low-risk dysplasia by Cox regression. Consistent with this result, patients with high-risk dysplastic OL had significantly higher oral cancer incidence than low-risk dysplasia, particularly during the first 2-3 years of follow-up by Kaplan-Meier curve (P < 0.001; Figue 1A). Similar findings were reported by Ho et al and Silverman et al[14, 16]. It is suggested that rigorous follow-up in the first 2-3 years for patients with diagnosis of dysplastic OL may be important to detect early malignant events.

We observed the average age at diagnosis of OL is 52.7 years, while other study populations had an average age closer to 60 years [16, 20]. The peak incidence of OL was in the fifth decade of life in our study, earlier than the sixth decade in other reports [20, 23]. Gender predilection is not of existence in our area, but significant gender differences (ratio M: F = 10.6: 1) were found in Taiwan, China [24]. The predominant sites of lesions are the tongue and buccal mucosa, and few lesions were located on the floor of mouth, whereas the tongue and floor of mouth were reported as the most common sites in Western countries [6, 16, 20]. These were probably due to the ethnic population and geographic difference in our cohorts compared to previous reports. Nevertheless, these factors were not related to malignant transformation of OL in our series. Moreover, It may be generally accepted that smoking and ethanol intake play significant roles in the development of OL, but the roles of those in the malignant transformation of OL is conflicting and yet unclear. The studies by Silverman et al [16] and Schepman et al [20] demonstrated an increased risk of malignant transformation in the non-smoking cohort, while the study by Ho et al [14] and our present study found smoking was not a significant factor in transformation risk. Ethanol intake was also not a risk factor for malignant transformation of OL [14, 15]. Further prospective cohort studies are needed to investigate the roles of lifestyle habits in the malignant process of OL.

In summary, we evaluated the usefulness of the new binary system of grading dysplasia proposed by WHO in prediction of OL malignant transformation risk. High-risk dysplasia was a significant indicator for OL malignant transformation. It is thus import to detect early malignant events of OL with the diagnosis of high-risk dysplasia in rigorous followed-up in the first 2-3 years.