Diagnostic Value of Ki-67 and Cytokeratin 13 Immunohistochemistry in Oral Precancerous Lesions

As oral precancerous lesions (OPL) have been considered as oral intraepithelial neoplasia, it is important to distinguish OPL from reactive atypical epithelium, clinically and pathologically. To achieve this, many researchers have identified certain molecular markers for the pathological diagnosis of OPL. Ki-67, cytokeratin 13, and/or cytokeratin 17 detection by immunohistochemistry has been reported to be useful for the differential diagnosis of oral precancerous lesions and reactive atypical epithelium. However, appropriately designed statistical studies have not been performed to assess their diagnostic potential. In this study, we aimed to assess the diagnostic potential of Ki-67, CK13, and CK17 for OPL by using a logistic regression model.

We initially examined the oral mucosal biopsy specimens of all patients managed between 2002 and 2006 at our department. Our exclusion criteria were (1) a previous diagnosis of oral squamous cell carcinoma or OPL, (2) OPL with concomitant oral squamous cell carcinoma at the first visit, (3) and a diagnosis of papilloma. Epithelial dysplasia was classified as low-grade (none or mild dysplasia) or high-grade (moderate dysplasia, severe dysplasia, or CIS). Subepithelial inflammation was classified as low (none or mild) or high (moderate or severe). Ki-67 immunoreactivity was scored as 0—sporadic positive cells in the parabasal layer, similar to normal mucosa; 1—applicable to neither score 0 nor 2; or 2—positive cells localized in the basal layer and/or in more superficial layers than in normal epithelium. CK13 was scored as 0—strong and diffuse expression, similar to normal mucosa; 1—weak and/or patchy expression; or 2—no expression. CK17 was scored as 0—no expression, similar to normal mucosa; 1—weak and/or patchy expression; or 2—strong and diffuse expression. A logistic regression model was used to test the diagnostic value of high-grade dysplasia. Goodness-of-fit of the model was tested using the Hosmer–Lemeshow test, and potential multicollinearity was tested using the variance inflation factor (VIF). The sensitivity, specificity, and positive or negative likelihood ratios (LR+ and LR−, respectively) corresponding to different markers versus a diagnosis of high-grade dysplasia were calculated. Receiver operating characteristic (ROC) curves were plotted and areas under these curves (AUC) were calculated. Variables with univariate associations (P ≤ 0.3) were used as candidate factors in the multivariate models. P< 0.05 denoted statistical significance.

The study included 76 patients (mean age, 62.2 years) at initial diagnosis. The most common involved site was the tongue (n = 21). Low-grade and high-grade dysplasia were identified in 52 and 24 cases, respectively. In univariate logistic analyses, inflammation, Ki-67 expression, CK13 expression, and CK17 expression were significantly associated with the presence of high-grade dysplasia (P< 0.001, = 0.001, and = 0.02, respectively). In multivariate logistic analyses, Ki-67 and CK13 expression were significant predictive factors for high-grade dysplasia (odds ratio, 2.32 and 5.19; 95% confidence interval [CI], 1.14–4.71 and 1.69–15.9, respectively). When cutoffs were chosen by posterior probability ≥ 0.5, the sensitivity, specificity, and AUC of Ki-67 were 50%, 86%, and 0.72, respectively. Likewise, the sensitivity, specificity, and AUC of CK13 were 83%, 63%, and 0.76, respectively. Ki67 together with CK13 yielded a sensitivity of 58%, specificity of 88%, and AUC of 0.82. Although Ki-67 and CK13 were independent markers for OPL diagnosis, they did not perform well individually. However, when the two markers were combined, their more robust predictive role became evident.

References: 1)Izumo T. Oral premalignant lesions: from the pathological viewpoint. Int J Clin Oncol 2011; 16(1): 15-26. 2)Mikami T, Cheng J, Maruyama S, et al. Emergence of keratin 17 vs. loss of keratin 13: their reciprocal immunohistochemical profiles in oral carcinoma in situ. Oral Oncol 2011; 47(6): 497-503.