This study aimed to investigate the clinical efficacy of using broadband white light (BWL) to observe morphologic appearance, narrow-band imaging (NBI) to observe intraepithelial microvasculature, and both BWL and NBI for the detection of high-grade dysplasia and carcinoma in oral leukoplakia. Among 317 patients (274 males and 43 females; aged 52.4 ± 10.7 years), the odds ratio (95% confidence interval) for detecting high-grade dysplasia and carcinomatous lesions based on morphologic appearances of BWL, and microvasculature patterns of NBI, were 39.12 (9.33–64.10), and 97.16 (38.19–247.21), respectively, which were significantly better than BWL ( p < 1 × 10 −15 ). The sensitivity, specificity, positive and negative predictive values, and accuracy of use of traditional BWL classification, NBI classification, and combined BWL and NBI classification for detecting high-grade dysplasia and carcinomatous lesions were 96.30, 60.08, 33.12, 98.75, 66.25, 39.92, and 3.70%; 87.04, 93.54, 73.44, 97.23, and 92.43%; and 100.00, 60.08, 33.96, 100.00, and 66.88%, respectively. In conclusion, the diagnostic accuracy by NBI classification of oral leukoplakia based on the intraepithelial microvasculature patterns is significantly better than BWL indicating that NBI is a promising non-invasive tool in detecting high-grade dysplasia and carcinomatous lesions in oral leukoplakia.
Clinically oral leukoplakia is divided into homogeneous and non-homogeneous types, with the latter further subdivided into speckled, nodular, and verrucous types based on the outer appearance. This traditional classification based on the outer morphologic appearance of oral leukoplakia is a major indicator of carcinoma and epithelial dysplasia. The rate of occurrence of dysplastic change and carcinoma is higher in non-homogeneous leukoplakia, but epithelial dysplasia or carcinomatous changes do occur in homogeneous leukoplakia. Since the presence of dysplastic areas in the epithelium of the upper aerodigestive tract is thought to be associated with a high likelihood of progression to cancer, identification of dysplastic lesions in oral leukoplakia is an important issue. Previous reports have revealed high-grade dysplasia to be a critical risk factor related to the malignant changes of leukoplakia. Thus, detecting high-grade dysplasia, carcinoma in situ, and invasive carcinoma (HGD/Tis/CA) within whitish patch lesions is important for further surgical intervention and prognosis.
Narrow-band imaging (NBI) is an innovative optical technology that enhances the contours and patterns of vessels, or intraepithelial microvasculature, in the surface of the mucosa by employing the characteristics of the light spectrum. This newly invented endoscopic technique has been used for detecting precancerous and neoplastic lesions in oral cavity mucosa. The purpose of the present study was to investigate the role of NBI in examining oral leukoplakia, under the hypothesis that NBI may provide more information in addition to the traditional classification under broadband white light (BWL). The specific aim was to analyze and compare the clinical efficacy of the morphologic appearance by BWL, the intraepithelial microvasculature pattern by NBI, and a combination of both light systems for detecting HGD/Tis/CA in oral leukoplakia.
Materials and methods
This study was approved by the Institutional Review Board of Chang Gung Memorial Hospital (relevant reference number: 100-2699C). Records of patients with oral leukoplakia who had undergone flexible endoscopy with BWL and NBI at the Department of Otolaryngology-Head and Neck Surgery of Chang Gung Memorial Hospital, Keelung City, from April 2009 to August 2011 were reviewed retrospectively.
The examinations had been conducted using ENF type V2, type VQ, and type VT (Olympus Medical Systems Corp., Tokyo, Japan) NBI endoscopes, light source (CLV-160B; Olympus Medical Systems, Tokyo, Japan), and a central video system (CV-160B; Olympus Medical Systems). A button on the control section of the video-endoscope allowed switching between the conventional BWL and NBI views. The examinations were first conducted using BWL illumination in a wide view to observe the entire lesion and its surrounding mucosa. The same procedure was performed under NBI illumination for detailed analysis of the appearance of the intraepithelial microvasculature and screening of random sites of normal mucosa.
The images were recorded and transferred to a computer hard drive for analysis. The clinical morphologic appearance under BWL was analyzed first, including homogeneous and non-homogeneous leukoplakia. The latter was categorized as the verrucous, nodular, or speckled type. Imaging analysis of the microvascular (regular; irregular) organization and the intraepithelial papillary capillary loop (IPCL) features by NBI was conducted based on the intraepithelial microvasculature patterns of the oral mucosa.
Endoscopic examination and surgical intervention, such as biopsy or surgical extirpation under local or general anaesthesia, had been performed after obtaining informed consent. Paraffin-embedded specimens were collected from the archives of the Department of Pathology, Chang Gung Memorial Hospital at Keelung. Haematoxylin and eosin-stained slides were reviewed by two independent pathologists (Dr. L.-C. Chang and Dr. H.-P. Chien) to confirm lesion diagnosis and determine the degree of epithelial dysplasia. Immunohistochemical staining was performed to confirm the pathologic diagnosis if necessary. Epithelial dysplasia was diagnosed according to the WHO 2005 classification. The inclusion criteria for this study were an oral whitish patch with pathologic diagnosis of hyperkeratosis, parakeratosis, squamous hyperplasia, grades of epithelial dysplasia, including low, intermediate and high ones, carcinoma in situ, or squamous cell carcinoma. The exclusion criteria were previous surgery on the oral whitish patch, white sponge nevus, frictional keratosis, morsicatio buccarum, chemical injury, acute pseudomembranous candidasis, leukoedema, discoid lupus erythematosus, skin graft, hairy leukoplakia, squamous papilloma, warts, lichen planus, lichenoid reaction, Fordyce spots, submucosa fibrosis, verrucous hyperplasia, or verrucous carcinoma. If more than one IPCL type was seen in some whitish patches by NBI, the most advanced type detected was determined as the IPCL type of the lesion.
Each patient’s chart records were reviewed, including their demographic data, site of the lesion, information regarding homogeneous or non-homogeneous leukoplakia, morphology of vascular architecture or IPCL, and histopathology. Any history of betel quid, alcohol, and tobacco use was obtained by detailed questioning during the patient’s first visit to the otolaryngology clinic of the hospital. The criteria for positive assignment were: one quid of betel or more daily, for at least one year; one cigarette or more per day for at least one year; and drinking alcohol for more than 4 days a week for at least one year.
Using the histopathologic findings, which served as the final diagnostic standard, two forms of analysis were performed by researchers blind to the sources of the data: determination of the correlation between pathology and imaging by BWL and NBI by analysis of the sensitivity, specificity, positive and negative predictive values, accuracy, and false positive and negative percentage in detecting HGD/Tis/CA; and comparison of traditional classification of leukoplakia based on observation of morphological appearance using BWL and observation of pattern of intraepithelial microvasculature using NBI.
Results are presented descriptively, with factors related to the clinical appearance of oral leukoplakia grouped and analyzed using χ 2 test. Odds ratio (OR) and 95% confidence intervals (CIs) were calculated using a two-tailed test of significance ( p < 0.05) for each risk factor. The authors used the following parameters: when the 95% CI did not include 1.0, the resulting OR of the risk factor was statistically significant; if the value of the OR was greater than 1.0, the risk was increased; and if the value was less than 1.0, the risk was reduced or protective.
The significance of differences in age between groups of homogeneous and non-homogeneous oral leukoplakia was analyzed by the Mann–Whitney U -test. A score of p < 0.05 indicated statistical significance for nonparametric data analyzed by one-tailed test.
The method described by Simel et al. was used for predictions and diagnostic tests. A comparison between two criteria was based on changes of the log-odds ratio for the two tables with the usual Pearson χ 2 on a 2 × 4 table in which each row is obtained by treating each 2 × 2 table as a one-way table with 4 cells. The statistical analyses were conducted using MATLAB program (MathWorks Inc., Natick, MA, USA).
317 consecutive patients who received ENF type V2, type VQ, and type VT NBI flexible endoscopic examinations and biopsy or laser surgery for oral leukoplakia between April 2009 and August 2011 were enrolled, including 274 males and 43 females, whose age ranged from 22 to 82 years ago (mean ± standard deviation: 52.4 ± 10.7 years). The patients studied formed part of a group of 154 patients with oral leukoplakia who had constituted the authors’ previous study, and 163 patients were further enrolled in this study. All of the patients were newly diagnosed and received either biopsy or both biopsy and surgical excision for oral leukoplakia. Patient demographics, including gender age, and oral habits; topographic locations included lip, buccal, gum, floor of the mouth, palate, retromolar, dorsal tongue, and ventrolateral tongue areas; images of oral leukoplakia examined by BWL and NBI ( Table 1 ). No significant differences were found between patients diagnosed with homogeneous and non-homogeneous leukoplakia in terms of basic demographic data, including gender, age, and oral habits ( Table 2 ).
|Characteristics||Case no. (%)|
|Age, years (mean ± standard deviation)||52.4 ± 10.7|
|Betel quid chewing|
|Dorsal tongue||11 (3.5)|
|Ventrolateral tongue||31 (9. 8)|
|Floor of the mouth||3 (1.0)|
|Gross morphological appearance by BWL|
|Homogeneous leukoplakia||160 (50.5)|
|Non-homogeneous leukoplakia||157 (49.5)|
|IPCL pattern by NBI|
|IPCL type I||150 (47.3)|
|IPCL type II||103 (32.5)|
|IPCL type III||49 (15.5)|
|IPCL type IV||15 (4.7)|
|Squamous hyperplasia||127 (40.0)|
|Low-grade dysplasia||96 (30.3)|
|Intermediate-grade dysplasia||40 (12.6)|
|High-grade dysplasia/carcinoma in situ||30 (9.5)|
|Squamous cell carcinoma||24 (7.6)|
|Homogeneous leukoplakia ( n = 160)||Non-homogeneous leukoplakia ( n = 157)||p value|
|Age (years) ‡||51.0 ± 10.3 §||53.9 ± 11.0 §||0.096|
|Alcohol drinking †||0.092|
|Betel quid chewing †||0.132|
Pathologic results were categorized into non-HGD/Tis/CA and HGD/Tis/CA groups. The former included squamous hyperplasia, low-grade dysplasia, and intermediate-grade dysplasia, and the latter included high-grade dysplasia, carcinoma in situ, and invasive carcinoma. Among the 160 cases of homogeneous leukoplakia, 158 were categorized as non-HGD/Tis/CA group and 2 as HGD/Tis/CA. Among the 157 cases of non-homogeneous leukoplakia, 105 were categorized as non-HGD/Tis/CA and 52 as HGD/Tis/CA. Based on the traditional BWL classification, patients with non-homogeneous leukoplakia had significantly increased risk of occurrence of pathologic HGD/Tis/CA ( p < 0.001, OR = 39.12, 95% CI = 9.33–64.10, Table 3 ).
|Non-HGD/Tis/CA *||HGD/Tis/CA||Odds ratio (CI † 95%)||p -Value|
|Classification of oral leukoplakia||<1 × 10 −15 ‡|
|Traditional classification based on morphological outer appearance of BWL||<0.001|
|Non-homogeneous leukoplakia||105||52||39.12 (9.33–64.10)|
|NBI classification based on intraepithelial microvascular pattern||<0.001|
|Low-risk NBI pattern (IPCL type I and type II)||246||7||1.0|
|High-risk NBI pattern (IPCL type III and type IV)||17||47||97.16 (38.19–247.21)|
IPCL type I (regularly organized arborescent IPCL, defined as IPCL type I, Fig. 1 ) and type II (tortuous or dilated IPCL, defined as IPCL type II, Fig. 2 ) were defined as the low-risk NBI pattern for occurrence of pathologic HGD/Tis/CA in oral leukoplakia, and IPCL type III (twisted, elongated, or winding IPCL, defined as IPCL type III, Fig. 3 ) and type IV (IPCL destruction, defined as IPCL type IV, Fig. 4 ) defined as the high-risk NBI pattern. Among the 253 cases of low-risk NBI pattern, 246 were categorized as non-HGD/Tis/CA and 7 as HGD/Tis/CA. Among the 64 cases of high-risk NBI pattern, 17 were categorized as non-HGD/Tis/CA and 47 as HGD/Tis/CA. On the basis of this NBI classification, patients with the high-risk NBI pattern had a significantly increased risk of occurrence of pathologic HGD/Tis/CA than those with the low-risk NBI pattern ( p < 0.001, OR = 97.16, 95% CI = 38.19–247.21, Table 3 ). Use of NBI to observe intraepithelial microvascular patterns was better in predicting HGD/Tis/CA compared to use of BWL to observe morphological outer appearance ( p < 1 × 10 −15 , Table 3 ).