Diagnostic sensitivity and specificity of laryngoscopic signs of reflux laryngitis

Objective. To assess the diagnostic sensitivity and specificity of the laryngoscopic signs of reflux laryngitis. Material and methods. Using original quantitative evaluation system, the laryngoscopic signs of 108 patients with reflux laryngitis and 90 healthy people were subjected to comparative analysis in this study. Summing up all evaluations of laryngeal changes, laryngoscopic reflux index was proposed. Results. Mucosal lesions and edema of vocal cords along with mucosal lesions of the interarytenoid notch were found to be most significant for diagnostics of reflux laryngitis. Presence of mucosal lesions of the interarytenoid notch (roughness, hypertrophy, keratosis, granuloma) increases the odds ratio to attribute the patient to the reflux laryngitis patient group 21 times (OR=21.32, 95% CI 4.38–103.93; P<0.001). Mucosal lesions (hypertrophy, keratosis, granuloma) and edema of vocal cords were determined as the most sensitive and rather specific laryngoscopic signs. Roughness and grade II hypertrophy of interarytenoid notch were found to be the most specific (98%) and sensitive (56%) laryngoscopic signs of reflux laryngitis. Abnormal values of laryngoscopic reflux index (higher than 5 points) reflect a diagnostic sensitivity of 96% and specificity of 97% differentiating patients with reflux laryngitis from healthy persons. Conclusions. Mucosal lesions and edema of vocal cords along with mucosal lesions of the interarytenoid notch are found to be the most significant laryngoscopic signs for diagnostics of reflux laryngitis. Laryngoscopic reflux index was found to be the most sensitive and specific diagnostic criterion for reflux laryngitis. Medicina (Kaunas) 2008; 44(4)


Introduction
Patients with typical and atypical symptoms of gastroesophageal disease (GERD) appeal both, to the gastroenterologist as well as to the ear, nose, and throat specialists. However, less than 50% of the patients present typical complaints of GERD: heartburn or regurgitation (1,2). Most of atypical GERD patients suffer from laryngopharyngeal symptoms (hoarseness, throat cleaning, cough, "globus" sensation), accompanied with various GERD-related laryngeal findings that are called reflux laryngitis (RL) (2)(3)(4).
GERD-related laryngopharyngeal signs are often established after laryngoscopy, because the most distinct pathological findings occur in the larynx. It depends on physiological and topographical relation between larynx and esophagus (1,(3)(4)(5)(6). The entirety of GERD-related pathological findings in the larynx (reflux laryngitis) and concomitant laryngopharyngeal symptoms are considered as laryngopharyngeal form of GERD (LF GERD) (5).
Reflux laryngitis is acknowledged as one of the most common manifestations of LF GERD (1,4,5,7). However, the data in the literature regarding the diagnostic criteria of RL are rather controversial because of low specificity of laryngoscopic signs and poor correlation between RL and pharyngeal pH-metry findings (8)(9)(10). Therefore, the actual scientific and practical needs require to revise the significance of laryngeal findings in order to evaluate the sensitivity and specificity of laryngoscopic signs in diagnostics of RL and to avoid over-diagnosis of the disease.
The aim of this study was to assess the diagnostic sensitivity and specificity of the laryngoscopic signs of reflux laryngitis.

Patients
The study was carried out at the Departments of Otorhinolaryngology and Gastroenterology, Kaunas University of Medicine Hospital, Lithuania. A total of 127 consecutive patients who appealed to otorhinolaryngologist because of atypical GERD symptoms (hoarseness, throat clearing, "globus" sensation, throat itching, cough), which lasted more than 3 months, were examined.
Reflux esophagitis (RE) according to Los Angeles (1994) classification (11) was proved endoscopically only in 36 of the 108 RL patients (33.3%). However, histologically RE was diagnosed in all 108 cases as the histological signs of RE (12) were found in all biopsy specimens. Nonerosive esophagitis was detected in 107 cases (99%), and erosive esophagitis was diagnosed in 1 case.
In the group of endoscopically proved esophagitis, grade A esophagitis was diagnosed in 72.2% of patients (n=26). Grade B esophagitis was not so prevalent and accounted for 27.8% of cases (P<0.001) (10).
Consequently, 19 patients without endoscopically and/or histologically proven RE were excluded from the study group despite some laryngoscopic signs of RL were found in these cases.
The control group consisted of 90 randomly selected healthy subjects (36 males, 54 females; mean age, 36.9±11.5 years). They had neither chronic laryngeal diseases nor any other long-lasting voice disorders, and none of them had ever consulted an otolaryngologist concerning the voice problems. In this respect, these persons considered themselves as healthy and therefore they could serve as controls.
The groups of the patients and the controls were dominated by people of 20-60 years of age, who had never undergone any voice training, and who were mostly office workers. The proportions of subjects in these groups did not differ significantly in respect to gender, voice training, social status, and smoking habits (P<0.05).

Methods
The laryngoscopic signs of 108 RL patients and 90 healthy people of the control group were subjected to comparative analysis in this study.
Laryngoscopic examination was carried out during video laryngostroboscopy (VLS) using Kay Elemetrics RLS (model 9100) device with 70° rigid telescope. Mucosal changes in four laryngeal regions -1) arytenoids (A), 2) the interarytenoid notch (IN), 3) vestibular folds (VF), 4) vocal cords (VC) -were evaluated according to the original quantitative evaluation system of the GERD-related laryngeal findings (13). Three types of mucosal changes were evaluated: 1) mucosal lesions/alteration of the epithelium (reduced mucosal light reflex, hypertrophy, roughness, keratosis, and granuloma), 2) erythema, and 3) edema (Table 1). According to this system, overall 12 laryngoscopic parameters were assessed. For the quantification of the evaluation, the following indices of laryngoscopic signs have been established: arytenoid index (AI), interarytenoid notch index (INI), vestibular fold index (VFI), vocal cord index (VCI). Each index was expressed in appropriate points (Table 1). Summing up all evaluations (indices) of mucosal changes in different laryngeal regions, laryngoscopic reflux index (LRI) was proposed. The possible scores of LRI were from 0 to 20 points.
Data analysis. Statistical analysis of the data was performed using SPSS (Statistical Package for Social Sciences) 10.0 for Windows. Nonparametric statistics Mann-Whitney U test, chi-square and two-tailed t tests as well as parametric analysis of variance were used. Differences between variables were considered to be reliable if the level of significance p or a (type I error) was <0.05 and b (type II error) was £0.2. When performing analysis of receiver operating characteristics (ROC) curves as presented in the SPSS program, limit (normal/abnormal) values of LRI were established making it possible to differentiate RL patients from healthy persons. Parametric values that were found to be larger than the limiting values were considered as pathological.
Diagnostic specificity and sensitivity of laryngoscopic signs and LRI were calculated. For determination of the most important laryngological signs in the diagnostics of RL, a binary logistic regression was carried out.

Laryngoscopic findings
When assessing the rate of laryngoscopic findings, it was found that signs of erythema and edema of mucosa of the entire larynx were statistically significantly (P<0.001) more common in the group of RL patients to compare with the controls. The same statement applies to the signs of reduced mucosal light reflex of arytenoids (A), vestibular folds (VF), vocal cords (VC) as well as to hypertrophy, roughness, and granuloma of the mucosa of interarytenoid notch (IN), VF, and VC. Mucosal roughness, keratosis, and grade II-III hypertrophy of IN (14) was found only in the RL patients' group ( Table 2). The laryngoscopic findings did not depend on age of the patients (P>0.05).
The mean LRI (11.78±3.64 points) of RL patients was found to be significantly greater than the mean LRI in the control group (1.70±1.96 points) (a<0.01,

Medicina (Kaunas) 2008; 44
The limit LRI value of 5 points differentiating RL patients from healthy persons was determined following analysis of receiver operating characteristics (ROC) curves.
The diagnostic sensitivity and specificity of the laryngoscopic signs As follows from Fig. 1, the diagnostic sensitivity and specificity of 9 out of 12 laryngoscopic signs tested was higher than 50%. Mucosal lesions (reduced mucosal light reflex, hypertrophy, keratosis, granuloma) and edema of VCs were determined as the most sensitive and rather specific laryngoscopic signs, i.e. diagnostic specificity of these signs was 90-91%. (IN) were the most sensitive laryngoscopic findings. However, the limits of diagnostic specificity of these signs varied from 66% (edema) to 71% (mucosal lesions). According to the statistical analysis data, roughness and grade II hypertrophy of IN (14) were found to be the most specific (98%) and sensitive (56%) laryngoscopic signs of RL. However, from the diagnostic point of view, the LRI itself as the resumptive of all laryngoscopic signs was found to be the most sensitive and specific. Abnormal values of LRI (higher than 5 points) corresponded to a diagnostic sensitivity of 96% and specificity of 97%.

The most important laryngoscopic signs for diagnosing reflux laryngitis established by logistic regression analysis
Logistic regression analysis revealed the following three laryngoscopic signs that were the most significant for diagnosing of GERD-related reflux laryngitis: 1.Mucosal lesions of vocal cords: reduced mucosa light reflex, hypertrophy, roughness, keratosis, granuloma, 2.Edema of vocal cord, 3.Mucosal lesions of the interarytenoid notch: reduced mucosa light reflex, roughness, hypertrophy, granuloma (Fig. 2).
The combination of these three laryngoscopic signs separates reliably (P<0.05) the RL patients from healthy persons in 95.9% of cases.
As the result of analysis of the data, it was established that mucosal lesions of the interarytenoid notch (hypertrophy, roughness, granuloma) have the highest diagnostic value among the investigated laryngoscopic signs diagnosing RL ( Table 4). Presence of mucosal lesions of the IN increases the odds ratio to attribute the patient to the RL patients' group 21 times (OR=21.32, 95% CI 4.38-103.93; P<0.001).

Discussion
It has been reported in the literature that among the patients seeking help for otolaryngologist, up to 10% of persons were found to have reflux-related symptoms, i.e. possible LF GERD (3,15). Therefore, diagnostics of LF GERD remains a rather complicated clinical problem. Despite the attempts to use contemporary, sophisticated, and "objective" methods of diagnostics, the result is quite disappointing.
Many studies have clearly showed that 24-h esophageal pH monitoring unfortunately is not a perfect "gold standard" for diagnosing atypical GERD (16)(17)(18). Overall, only 54% of patients with suspected GERD and related laryngoscopic signs present abnormal esophageal acid exposure, irrespective of the location of the pH probe (distal, proximal esophagus or hypopharyngeal) (18). Majority of the patients (60-70%) with laryngeal symptoms related to suspected GERD have normal esophageal endoscopies (1,7,13,19). Therefore, an empirical treatment with proton pomp inhibitors is not always successful in cases of atypical GERD unresponsive to treatment (8,20,21).
As reflux laryngitis (RL) is considered as one of the most common manifestations of LF GERD, diagnostics of that clinical entity appears as one of "key  stones" solving the problem. However, controversy in the literature regarding the diagnostic criteria of RL makes the task even more complicated (8)(9)(10). Some studies resume the mucus stasis, erythema, edema, hypertrophy, ulceration, and granulation of the posterior larynx (arytenoids, interarytenoid notch) as sufficient for diagnostics of RL (22,23). However, Koufman (1991) postulates that RL may present as erythema, edema, and epithelial changes in all parts of the larynx while rather subjectively evaluated signs as mucosal edema and erythema (89% and 87%, respectively) are predominant (1). On the other hand, pathological changes in posterior larynx typical for RL may present also among persons without GERD (10,24,25). According to Reulbach study (n=100), signs of RL were determined in 64% of the community-based cohort of adults over 40 years of age. Only 12% of examined persons had a completely normal laryngeal examination (25). Moreover, some laryngoscopic signs of RL (laryngeal tissue irritation) were found in 85-87% healthy volunteers in Hicks' (n=105) and Vavricka's (n=132) studies (10,24). There could be several reasons explaining discrepancy among the laryngoscopic findings mentioned above: different patients' inclusion and exclusion criteria, distinction in assessment of the laryngoscopic findings, different laryngoscopic techniques used (flexible or rigid laryngoscope). Milstein et al. (2005) revealed that laryngeal signs of RL were more likely to be detected using flexible than rigid laryngoscopy in the same individual (75% vs. 54%) (26).
Data of the present study stress on completeness of assessment of laryngoscopic signs of RL. Despite the fact that roughness and grade II hypertrophy of IN (Fig. 3) were found to be specific (98%) and sensitive (56%) laryngoscopic signs of RL, only the LRI was found to be the most sensitive and specific from the diagnostic point of view. Abnormal values of LRI (higher than 5 points) corresponded to a diagnostic sensitivity of 96% and specificity of 97%.
According the data of logistic regression analysis, the following three laryngoscopic signs were found to be the most significant for diagnostics of GERD related RL: a) mucosal lesions of vocal cords (reduced mucosa light reflex, hypertrophy, roughness, keratosis, granuloma), b) edema of vocal cords, and c) mucosal lesions of the interarytenoid notch (reduced mucosa light reflex, roughness, hypertrophy, granuloma). Combination of these three laryngoscopic signs separates reliably the RL patients from healthy persons in 95.9% of cases.
On the other hand, data of the present study confirmed the highest diagnostic value of mucosal lesions of the interarytenoid notch (hypertrophy, roughness, granuloma) among the investigated laryngoscopic signs diagnosing RL. The odds ratio to attribute the patient to the RL patients group increases 21 times in case of presence of mucosal lesions of the interarytenoid notch.
To increase the reliability of diagnostics of RL, we should be more careful and incredulous evaluating sensitive but less specific laryngeal and rather "subjective" signs such as mucosal edema and erythema and rely on more evident and "objective" signs such as mucosal lesions (hypertrophy, roughness, keratosis, granuloma) like it was done by gastroenterologists turning away from Hetzel's to Los Angeles classification of the esophagitis (27).

Conclusions
1. Laryngoscopic reflux index is sensitive and specific tool for the diagnostics of reflux laryngitis. Abnormal values of laryngoscopic reflux index (higher than limiting laryngoscopic reflux index value of 5 points) reflect a diagnostic sensitivity of 96% and specificity of 97% differentiating patients with reflux laryngitis from healthy persons.
2. Mucosal lesions and edema of vocal cords along with mucosal lesions of the interarytenoid notch are found to be the most significant for diagnostics of gastroesophageal reflux disease-related reflux laryngitis. Presence of mucosal lesions of the interarytenoid notch increases the odds ratio to attribute the patient to the reflux laryngitis patients' group 21 times.