Differences in the diagnostic value between fiberoptic and high definition laryngoscopy for the characterisation of pharyngeal and laryngeal lesions: A multi‐observer paired analysis of videos

Abstract Objectives High definition laryngoscopy (HDL) could lead to better interpretation of the pharyngeal and laryngeal mucosa than regularly used fiberoptic laryngoscopy (FOL). The primary aim of this study is to quantify the diagnostic advantage of HDL over FOL in detecting mucosal anomalies in general, in differentiating malignant from benign lesions and in predicting specific histological entities. The secondary aim is to analyse image quality of both laryngoscopes. Design Retrospective paired analysis with multiple observers evaluating endoscopic videos simulating daily clinical practice. Setting A tertiary referral hospital. Participants In 36 patients, both FOL and HDL videos were obtained. Six observers were provided with additional clinical information, and 36 FOL and HDL videos were evaluated in a randomised order. Main outcome measures Sensitivity, specificity, positive and negative predictive value and diagnostic accuracy of observers using both flexible laryngoscopes were calculated for detection of mucosal lesions in general and uncovering malignant lesions. Sensitivities were calculated for prediction of specific histological entities. Image quality (scale 1‐10) was assessed for both flexible laryngoscopes. Results HDL reached higher sensitivity compared to FOL for detection of mucosal abnormalities in general (96.0% vs 90.4%; P = .03), differentiating malignant from benign lesions (91.7% vs 79.8%; P = .03) and prediction of specific histological entities (59.7% vs 47.2%; P < .01). Image quality was judged better with HDL in comparison with FOL (mean: 8.4 vs 5.4, P < .01). Conclusions HDL is superior to FOL in detecting mucosal anomalies in general, malignancies and specific histological entities. Image quality is considered as superior using HDL compared to FOL.


| INTRODUC TI ON
Visualisation of the upper aerodigestive tract is essential in the diagnostic process of pharyngeal and laryngeal lesions. 1 Especially distinguishing benign from malignant mucosal lesions is of importance because early detection of pharyngeal or laryngeal cancer improves survival. [2][3][4] Although first described in 1954, flexible fiberoptic laryngoscopy (FOL) is still commonly used worldwide. 5,6 In FOL glass fibres transmit light through the fibrescope to compose the acquired image. 7 A disadvantage of the commonly used FOL could be the overlooking of small epithelial changes. Also, the differentiation between benign and malignant tumours in vivo seems to be difficult. 4 Very early malignant disease presents as low-contrast unspecific mucosal changes with superficial reddening and superficial roughness. 4,8 Insufficient optical resolution negatively influences its correct interpretation, and it depends on the clinical experience whether this will be detected. [8][9][10] Imaging like computed tomography, magnetic resonance imaging and fluorodeoxyglucose-positron emission tomography provides information about the extension of tumours and lymph node metastasis but fails to identify superficial mucosal abnormalities. 10 In the last decades, medical imaging technology has improved tremendously and this has resulted in more detailed medical photos. 11 Next to FOL, standard definition endoscopy, also called digital chip-on-tip endoscopy, was introduced in 1983. [12][13][14][15] This endoscope consists of a small light-sensitive charge-coupled device (CCD) in the tip of the endoscope which functions as a miniature TV camera. 7,[13][14][15] The image is transmitted through the endoscope to a video processor in the form of a digital signal. 15 Image quality of distal chip technology is validated better in comparison to FOL. 16 Nowadays, high definition images, which have 850.000 to 1 million pixels, can be used in flexible high definition laryngoscopy (HDL). 13 HDL is a new diagnostic tool presumably leading to better interpretation of the inspected mucosa and earlier detection of head and neck cancer. However, the advantage of HDL has never been quantified before. HDL needs to pass the ubiquitous technology adoption life cycle and, like any other technology, has to be adopted by five categories of consumers. 17 Innovators and early adopters are willing for changes and adopt new ideas but late majority need extensive statistical evidence of new technology to be convinced of the benefits for the patient. 18 Other barriers could be the believe of medical professionals that the new technology does not show an improved performance in comparison with the standard technology. 17 The primary aim of this study is to quantify the advantage of HDL over FOL in detecting mucosal anomalies in general, differentiating malignant from benign lesions and predicting specific histological entities in pharynx and larynx. The secondary aim is to analyse the image quality of both laryngoscopes.

| Ethical considerations
The Institutional Review Board of the University Medical Centre Groningen assessed this retrospective study and judged that there was no need for approval based on the Dutch Medical Research Law (Wet medisch-wetenschappelijk onderzoek met mensen [WMO]).

| Patients
In this study, we included archived pharyngeal and laryngeal endoscopic videos of 51 patients collected routinely during diagnostic procedures between June 2014 and October 2017. Patient data, videos and histopathological results were assembled from the electronic patient records. Inclusion criteria were availability of both one FOL video and one HDL video of the same lesion. Both videos had to be recorded within a maximum of 3 months without treatment between both endoscopies. Videos of normal pharynges and larynges were also included in order to assess the detection rate of lesions. In total, 15 lesions were excluded because either an FOL or HDL video was not available (n = 10), the lesion altered between the two video recordings (n = 1), the histological diagnosis was not possible to classify (n = 2) or the interval between the two recordings extended 3 months (n = 2). Figure 1 shows examples applied in our study group.
Conclusions: HDL is superior to FOL in detecting mucosal anomalies in general, malignancies and specific histological entities. Image quality is considered as superior using HDL compared to FOL.

Keypoints
• High definition laryngoscopy (HDL) could lead to better interpretation of the pharyngeal and laryngeal mucosa than regularly used fiberoptic laryngoscopy (FOL).
• Six observers evaluated FOL and HDL videos (paired analysis) of 36 patients in a randomized order.
• Sensitivities of HDL were significantly higher compared with FOL for detecting mucosal lesions in general, differentiating malignant from benign lesions and for predicting specific histological entities.
• Although FOL is still commonly applied, it is advised to use HDL in the standard flexible endoscopic evaluation of the pharyngeal and laryngeal mucosa.

| Procedure
For FOL, a flexible fiberoptic rhinolaryngoscope ENF GP (Olympus Medical Systems), which was connected to a Matrix E camera processer (Xion Gmbh), and for HDL, a flexible video rhinolaryngoscope ENF VH (Olympus Medical Systems), which was coupled to a HD monitor, were used. After using FOL for examination, patients gave informed consent and were included in a prospective HD-NBI study.
These patients (89%) underwent HDL as an additional examination for that study. In four patients (11%), FOL was followed by HDL due to uncertainty of the diagnosis. When a patient had a suspicious lesion, a biopsy was taken or the lesion was resected completely. The pathologist assigned the diagnosis of the biopsy.
The videos were edited to acquire fragments with a maximum  (Table 1). Observers had the possibility to choose from a variety of diagnoses even if there were no cases with these diagnosis included. Thirdly, observers had to judge the image quality on a scale from zero to ten (ie 0 = very poor image quality, 10 = excellent image quality). Each observer had a maximum of 30 seconds to judge each video. In this way, six observers assessed 72 videos (ie both FOL and HDL performed in 36 patients) which resulted in 432 observations. All six observers had at least 5 years of experience in the field of laryngology and/or head and neck oncology.

| Statistical analysis
Statistical analysis was performed using SPSS version 22.0 (IBM Corp.). The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and diagnostic accuracy were calculated of both FOL and HDL for the detection of a mucosal lesion and of a malignant lesion in the pharynx and larynx. Sensitivities for prediction of a specific histological entity were calculated for each laryngoscope. The chi-squared test was used to analyse differences in sensitivities, specificity, PPV, NPV and diagnostic accuracy between both techniques. Furthermore, for each video the mean sensitivity of all six chosen diagnoses was calculated. Wilcoxon signed-rank test for matched pair samples was performed to compare the values of FOL and HDL and to evaluate the difference in image quality between both laryngoscopes. In this study, a P-value < 0.05 was considered as significant.

| Detection of mucosal lesions in general and differentiating malignant from benign lesions
As demonstrated in  Table 3). As shown in the overview of sensitivities in Figure 2, HDL reached significantly higher sensitivities compared to FOL for the detection of mucosal lesions, for differentiating malignant from benign lesions and for predicting specific histological entities.

| Key findings
To the best of our knowledge, this is the first study using videos directly comparing the commonly used FOL with the recently developed HDL to quantify the value in the diagnostic process of pharyngeal and laryngeal lesions. This study showed that HDL is superior

| Strengths of the study
This study was designed to combine a simulation of daily clinical practice with a reliable method of paired statistical analysis with multiple observers to obtain a scientifically solid conclusion which could be translated into daily clinical care. Daily clinical practice was simulated by using pharyngeal and laryngeal videos instead of photographs. By presenting information about the patients' risk factors TA B L E 4 (a) Sensitivity per diagnosis for prediction of a specific histological entity and (b) paired analysis for overall sensitivity and image quality of FOL and HDL

| Comparison with other studies
Our study shows that HDL has a significantly higher sensitivity for detecting a malignant lesion than FOL. This is of great importance because a timely detection of malignant lesions is an important factor for the prognosis of a patient. [2][3][4] Additionally, early detection is essential not only in primary cases but also in the followup in patients who already have head and neck cancer in their medical history. 21  The overview of the sensitivities (Figure 2) shows differences in detecting mucosal lesions in general, in differentiating malignant from benign lesions and in predicting specific histological entities very clearly. Comparing these sensitivities with each other, it is obvious that they are highest for detecting mucosal lesions in general and lowest for recognition of specific histological entities. This is a logical result because it is more difficult to detect a specific histological entity than to detect an unspecific anomaly in general. Moreover, the overview explicitly illustrated that HDL has higher sensitivities than FOL.

| Clinical applicability
Nowadays, the availability of the HDL has increased. Due to the evidence of our study and to the reflection of normal circumstances resembling daily clinical practice, our advice is to use HDL instead of FOL in the standard endoscopic evaluation of the pharynx and larynx in the outpatient setting, especially if a malignancy is expected.

| CON CLUS ION
HDL is superior to FOL in diagnosing pharyngeal and laryngeal lesions. Sensitivities of HDL were significantly higher compared with FOL for detecting mucosal lesions in general, differentiating malignant from benign lesions and for predicting specific histological entities. The HDL image quality was considered better. Although FOL is still commonly applied, in our opinion, it is of great importance to soon adopt HDL as the gold standard in flexible endoscopic evaluation of the pharyngeal and laryngeal mucosa.