No reliable standardized test is available to quantify degree of epithelial dysplasia (ED) in oral leukoplakia. Because silver stainable nucleolar organizer region (AgNOR) analysis is a simple and inexpensive way to discriminate resting cells from those entering the cell cycle, mean AgNOR count might show promise as a quantifiable test to standardize definitive diagnosis of degree of ED. Objective: To establish a diagnostic cut-off mean AgNOR count value for ED and to validate it by bootstrap simulation. Methods: One hundred new biopsies of clinical leukoplakia between April and July 2001 at Dr. R. A. Dental College, Koltata, India, were examined for presence or absence of ED using hematoxylin and eosin stain as a gold standard criterion. Results were compared against their mean AgNOR counts using standard protocol. An optimal cut-off for mean AgNOR count was selected employing two-graph Receiver Operating Characteristic (ROC) analysis. Simulation was done with 10,000 resampled data sets (with replacement) from the study sample using parametric and non-parametric bootstrap to correct for bias and to estimate standard errors (SE) of the cut-off point. Results: ROC analysis provided AUC=74% (95% CI: 64%-83%); non-parametric method suggested a mean AgNOR count cut-off =2.42; and parametric method suggested mean cut-off =2.57. Non-parametric bootstrap SE was 0.080323 (estimated bias= 0.002075); whereas parametric Bootstrap SE was 0.089655 (estimated bias= -0.011356). 95% CIs (BCa: bias corrected and accelerated) for the optimal cut-off from parametric method [2.31; 2.66] was lower than from non-parametric method [2.43; 2.82]. Conclusion: Applying a conservative diagnostic paradigm by taking the lowest 95% CI estimate from bootstrap analyses, we suggest that mean AgNOR value of 2.3 be used as a diagnostic cut-off for defining oral mucosal ED.

Back to the Oral Medicine & Pathology Program
Back to the IADR/AADR/CADR 82nd General Session (March 10-13, 2004)

Top Level Search