A Disease Activity Score for ENT Involvement in Granulomatosis With Polyangiitis (Wegener s)

Transcription

1 The Laryngoscope VC 2013 The American Laryngological, Rhinological and Otological Society, Inc. A Disease Activity Score for ENT Involvement in Granulomatosis With Polyangiitis (Wegener s) Marcos Martinez Del Pero, MRCS; Afzal Chaudhry, PhD, FRCP; Niels Rasmussen, MD; Piyush Jani, FRCS (ORL-HNS); David Jayne, MD, FRCP Objectives/Hypothesis: Accurate assessment of disease activity in patients with otorhinolaryngological manifestations of granulomatosis with polyangiitis (Wegener s) (ENT/GPA) is necessary for treatment decisions and clinical trials. We have designed a disease activity score (ENT/GPA DAS) for this purpose. Study Design: A prospective cross-sectional study. Methods: GPA patients seen in a tertiary center were systematically assessed for disease activity and/or infection in the ear, nose, and throat region using European Vasculitis Study Group guidelines. An ENT disease activity score was developed and validated, and compared to the ENT domain of the Birmingham Vasculitis Activity Score (ENT/BVAS). Results: One hundred forty-four patients were studied, of whom 87% (125/144) had ENT involvement. ENT items of disease activity were correlated with expert clinical assessment. Discriminant correlation tests were performed to control for infection. Six items were retained to form the ENT/GPA DAS: reported bloody rhinorrhoea, granulation, ulcers and/or friable mucosa in the upper airway on endoscopic evaluation, objective stridor, sudden sensorineural hearing loss, and inflamed tympanic membrane/middle ear without infection. Nasal crusting was excluded. Individual items of ENT/GPA DAS had higher sensitivities and comparable specificities in predicting disease activity than ENT/BVAS items. Overall ENT/GPA DAS demonstrated higher sensitivity and lower specificity for disease activity in ENT/GPA when compared to overall ENT/BVAS. Conclusions: A tool to assess ENT disease activity in GPA has been developed. It is potentially superior to existing tools but requires further testing for intra- and interobserver reliability. Key Words: Granulomatosis with polyangiitis, Wegener s, granulomatosis, ENT, assessment. Level of Evidence: 2c. Laryngoscope, 123: , 2013 INTRODUCTION Granulomatosis with polyangiitis (Wegener s) (GPA) is a primary systemic vasculitis that commonly affects the ear, nose, and throat (ENT). 1,2 Multiple sites can be affected with epithelial damage, secondary infection, and irreversible damage. Scoring systems have been devised to quantify both disease activity and chronic damage in order to guide therapy and as outcome measures for clinical trials. 3 The scoring systems rely on the subjective evaluation of the examiner to discriminate disease activity from damage or other causes. In the ENT system, the presence of infection makes From the Department of Otolaryngology (M.M.DP., P.J.) and Department of Medicine (A.C., D.J.), Addenbrooke s Hospital, Cambridge University Hospitals NHS Foundation Trust, Cambridge, U.K.; and the Department of Otolaryngology (N.R.), Rigshospitalet, Copenhagen, Denmark. Editor s Note: This Manuscript was accepted for publication on August 22, Presented at the Royal Society of Medicine, London, U.K., March 5, All the work was financed by Addenbrooke s Hospital, Cambridge. The authors have no other funding, financial conflicts, or conflicts of interest to disclose. Send correspondence to Marcos Martinez Del Pero, ENT Department, Box 48, Addenbrooke s Hospital, Hills Road, Cambridge CB2 0QQ, U.K. marcos@doctors.org.uk DOI: /lary assessment of disease activity more difficult. For example, crusting or a sensation of nasal obstruction can represent either disease process. Moreover, infection and active vasculitis often coexist and infection may aggravate vasculitis. 4 Current vasculitis assessment employs the Birmingham Vasculitis Activity Score (BVAS). 5 It is divided into nine organ domains, and the ear, nose and throat (ENT) domain encompasses five items (Table I). Although BVAS has proven useful as an outcome measure in several trials and its use is recommended by rheumatology societies, 6 disease activity of the ENT system is difficult to assess. For instance, the definition of sinus involvement is not specific enough and does not follow current recommendations, such as those from the European position article. 7 Moreover, BVAS does not require assessment by ENT clinicians or audiologists despite categories such as subglottic stenosis and hearing loss (conductive or sensorineural). Although BVAS performs well 8 as an overall measure of disease activity and there have been studies assessing its correlation to disease activity in GPA, 9 there are no studies focusing on the ENT BVAS items. We aimed to develop a more detailed and objective assessment tool, and to compare this tool with BVAS, in order to optimize ENT evaluation for the clinical setting and for clinical trials involving GPA patients. et al. ENT/GPA Disease Activity Score

2 Items Bloody nasal discharge/ nasal crusts/ulcers and/orgranulomata Paranasal sinus involvement TABLE I. BVAS ENT Items. Description Bloody, mucopurulent, nasal secretion, light or dark brown crusts frequently obstructing the nose, nasal ulcers and/or granulomatous lesions observed by rhinoscopy Tenderness or pain over paranasal sinuses with pathologic imaging (CT, MR, X-ray, ultrasound) Subglottic stenosis Stridor and hoarseness due to inflammation and narrowing of the subglottic area observed by laryngoscopy Conductive hearing loss Hearing loss due to middle ear involvement confirmed by otoscopy and/or tuning fork examination and/or audiometry Sensorineural hearing loss Hearing loss due to auditory nerve or cochlear damage confirmed by audiometry MATERIALS AND METHODS Patients GPA patients attending a vasculitis clinic at a tertiary center were assessed by a single ENT specialist trained in vasculitis. 10 The diagnosis of GPA was made using the European Vasculitis Study Group (EUVAS) criteria based on the Chapel Hill Consensus conference definitions. Prior ethical approval was obtained and patients gave written informed consent. Design and Validation We have applied a previously described design and validation process consisting of four steps: 11,12 Face Validity. A structured assessment was carried out using an all-encompassing database of ENT signs and symptoms. 10 This database includes items from existing scoring systems and additional items suggested by experts in the field. Disease activity or remission was determined using the European League Against Rheumatism recommendations. 13 Infection was defined as purulent-looking discharge, and in the nasal cavity and ears this was confirmed with positive microbiology tests. Noninfection was established by the absence of purulent discharge and normal mucosa. In cases of crusting of the nose, if there was dry black/bloody crusting overlying friable mucosa, with no associated purulent discharge, it was classified as noninfected and attributable to disease activity. Purulent nasal discharge mixed with blood-stained crusting and friable mucosal lining was deemed as mixed disease activity and infection. Dry crusting overlying pale, nonfriable mucosa was considered a consequence of damaged mucosa in patients in remission. Content Validity. Individual ENT signs and symptoms were compared against a gold standard assessment as to whether the disease was active or not. An experienced vasculitis physician performed the assessment based on the patient s history, examination of affected systems excluding the objective ENT assessment, and relevant laboratory parameters. Univariate analysis was performed for each sign or symptom for the three regions: the ear, nose and head and neck. Signs and symptoms significant at the P < 0.05 level were entered into a multivariate regression analysis using activity or remission as dependent variables. Following this analysis, a set of items of ENT disease was selected. Construct Validity. In order to assess the specificity of potential ENT items for disease activity as compared to infection, they were tested for the degree of correlation with disease activity (convergent validity) and with infection (discriminant validity). Potential disease activity items should show correlation with the presence of active disease but not with infection. Criterion-Oriented Validity. The sensitivity and specificity of the new assessment tool based on the selected items for disease activity was compared to the performance of the ENT domain of BVAS. A gold standard assessor was used to judge disease activity. TABLE II. Demography. Sex n female 69 (47.9) Age years (median, IQ range) 57.7 ( ) Disease duration months (median, IQ range) 65.2 ( ) No ENT disease activity 98 (68) Prevalence of organ involvement n * E 13 (8) EL 25 (17) EK 10 (7) ELK [33] 18 (13) K 4 (3) L 3 (2) Localized 59 (41) Early systemic 8 (6) Generalized&sol;severe renal [17] 77 (53) Incidence of organ involvement during disease activity (in brackets patients with activity in specified organ only) ENT 44 (22) EL 9 (6) EK 6 (2) ELK 2 (1) Renal 5 (0) Pulmonary 7 (1) General 23 (4) Cutaneous 5 (0) Membranous and eyes 6 (0) Nervous system 10 (2) Distribution within ENT of activity and damage n * Activity in the ear 9 (6) Activity in the nose 46 (32) Activity in the head and neck 15 (10) Damage in ears 88 (61) Damage in nose 75 (42) Damage in head and neck 30 (21) Damage in ENT system 121 (84) *Percentages are calculated over a total of n ¼ 144. E ¼ ENT involvement; Early systemic ¼ any except renal or imminent vital organ failure; Generalized ¼ renal with serum creatinine level <500 mmol/l þ/ other vital organ failure; K ¼ kidney involvement; L ¼ lung involvement; Localized ¼ upper þ/ lower respiratory tract. 623

3 TABLE III. Univariate Analysis of ENT Items. Active Remission P value Ear items Reported HL* No reported HL 1 73 Tinnitus No tinnitus Otalgia No otalgia Otorrhoea No otorrhoea OME&sol;inflammed TM Scarred&sol;normal TM Rinne s negative Rinne s positive 2 80 Objective HL No objective HL 1 18 Nose items Obstruction No obstruction Bloody rhinorrhoea E-7 No bloody rhinorrhoea Normal olfaction Abnormal olfaction Nasal crusting E-7 No nasal crusting Sinusitis No sinusitis Normal AR E-5 Abnormal AR Mucosal discoloration No mucosal discoloration AR crusts E-5 No AR crusts AR ulcers No AR ulcers RR Normal E-8 RR abnormal RR crusts E-6 No RR crusts RR granulation E-6 No RR granulation Head and Neck Items Voice change No voice change Subjective stridor No subjective stridor Objective stridor E-7 No objective stridor Pharyngitis No phayrngitis Larynx inflammation E-5 No larynx inflammation 4 98 (Continued) TABLE III. (Continued). Statistical Analysis The sample size required was calculated using a formula for one sample to measure proportions: n 1:962 d 2 Active Remission P value Subglottic stenosis No subglottic stenosis Tracheal stenosis No tracheal stenosis Scarred includes patients with perforated tympanic membranes, retractions. Patients with infection excluded (n ¼ 4). *HL ¼ hearing loss; TM ¼ tympanic membrane; AR, RR. pð1 pþ where p is the expected proportion of patients with active disease, and d is the prespecified percentage within which the specified amount lies with 95% confidence. If the expected number of patients with active disease is 30% and the range is 8%, the following result is obtained: n 1:962 0:3ð1 0:3Þ 126 0:082 Continuous variables were described as medians and interquartile ranges (IQR), and categorical variables were described as counts and percentages. Continuous variables were compared with the Wilcoxon rank-sum test, and categorical variables were compared with the Fisher s exact test. To assess potential signs and symptoms for the new scoring system, we constructed multivariable logistic regression models for the outcomes of disease activity. The independent variables used were the signs and symptoms that showed a significant association with disease activity in the univariate analysis. The sensitivity, specificity, and correlation coefficient (Spearman s rank order correlation) of each of the individual new items, as well as the overall ENT/GPA DAS, for identifying disease activity was calculated. Similar calculations were performed for the individual and overall ENT/BVAS components / score. All tests were two-sided and P values <0.05 were considered significant. Analyses were conducted with SPSS version 16 (SPSS Inc., Chicago, IL) and NCSS 2007 version 07.12(NCSS LLC, Kaysville, UT). RESULTS All 144 patients with GPA described previously were included in the analysis (Table II) 10. Face Validity The inclusion in the database of items from current scoring systems, along with expert advice, fulfill face validity. Content Validity Analysis (Tables III & IV). Ear items. Reported hearing loss; otorrhoea; and the examination of the tympanic membrane, revealing either inflammation in the absence of infection or a new 624

4 TABLE IV. Multivariate Analysis of ENT Items. OR 95% CI P value Nose Items Bloody rhinorrhoea Nasal crusting Olfactory dysfunction Nasal obstruction Crusting on rhinoscopy Granulation on rhinoscopy Discolouration on rhinoscopy Ulcers on rhinoscopy Head and Neck items Voice change Objective stridor Pain Odinophagia > Abnormal pharynx on endoscopy Abnormal subglottis on endoscopy otitis media with effusion (OME), were discriminatory in determining disease activity. Negative Rinne test approached significance. In this sample, Rinne test and reported hearing loss were reliable in terms of sensitivity and specificity (96.5% and 79% and 100% and 87.5%, respectively). Among formally tested patients, hearing loss was not a predictor of disease activity (Table III). The ear items were not included in the multivariate analysis because there was insufficient discriminating data. Nose Items. Analysis of nasal symptoms revealed nasal crusting, bloody rhinorrhoea, nasal obstruction, and abnormal olfaction as significant symptoms in predicting disease activity in the univariate analysis. Normal rhinoscopy and crusting and granulation seen on rhinoscopy were also significant in the univariate analysis (Table III). Friable, erythematous mucosa and the presence of ulcers, granulation, and crusting seen on rigid rhinoscopy were independently associated with disease activity in the multivariate analysis (Table IV). Head and Neck Items. Both univariate and multivariate analysis of head and neck features showed that objective stridor (assessed by the clinician) was a significant predictor of disease activity. Laryngeal or subglottic inflammation was significant by univariate analysis (Table III). These two features and dysphagia were not included in the multivariate analysis because there was insufficient discriminating data (Table IV). Infection was found to be associated with active disease in the nose (P ¼ 0.021) but was not associated with ear and head and neck items (P ¼ and P ¼ 0.104). Construct Validity Analysis Convergent and divergent analysis of items affecting the ears was not possible, but univariate analysis showed that an inflamed tympanic membrane or granulation in the middle ear was associated with disease activity (P ¼ 0.009). Figure 1 shows convergence results, and Figure 2 shows divergence results of nasal and head and neck items. Nasal crusting showed convergence with disease activity, but did not discriminate from infection. For this reason an item was created of nasal inflammation on nasal endoscopy, excluding crusting. The sensitivity and correlation results are shown in Table V and illustrated in Figure 3. Criterion-Oriented (Concurrent) Validity Analysis Total ENT BVAS and individual ENT BVAS items were correlated with disease activity (Table VI). However, BVAS had a sensitivity of 45.5%, but a specificity of 89.6%, whereas the items proposed have a sensitivity of 97.8% and a specificity of 52.6%. Fig. 1. Convergent correlations. Abnormal endoscopy includes ulceration, granulomata, or inflamed mucosa in the nasal cavity and in the hypopharynx, larynx, or subglottis. 625

5 Fig. 2. Divergent correlations. Abnormal endoscopy includes ulceration, granulomata, or inflamed mucosa in the nasal cavity and in the hypopharynx, larynx, or subglottis. The proposed new scoring system (ENT/GPA DAS) is shown in Table VII. The scoring system will be binary (0 ¼ not present, 1 ¼ present) using BVAS rules: items are only scored if they are new or worse in the preceding 4 weeks. DISCUSSION The Outcome Measures in Rheumatology process (OMERACT) recognizes scoring systems as a useful tool to assess chronic diseases and judges them in terms of truth, discrimination, and feasibility. 6 It has recognised BVAS as a core tool to use clinically and in research in order to measure disease activity, outcomes of the treatment of vasculitis, and prognosis. 6,13 Mahr et al. assessed the correlation to disease activity and weight of each item using a modified version of BVAS for GPA (BVAS/WG), but they did not attempt to redefine the ENT items. 9 The ENT items in BVAS appear nonspecific (e.g., paranasal sinus involvement), and there are items that require specialist input, although the scoring system does not make it essential Items TABLE V. Proposed Scoring System Correlation with Disease Activity as Assessed by Single Observer. Sensitivity Specificity Spearman Correlation Coefficient (r) P value Bloody rhinorrhoea E-9 Objective stridor E-8 Inflammation on rhinoscopy E-6 Inflammation on E-7 flexible laryngoscopy Inflamed TM/middle E-6 ear mass Sudden sensorineural N.A. N.A. N.A. N.A. hearing loss Total ENT/GPA DAS E-16 Fig. 3. ROC curve for each scoring system assessed against the assessors impression of disease activity or remission. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.] Items TABLE VI. BVAS Correlation with Assessed Disease Activity. Sensitivity Specificity Spearman Correlation Coefficient (r) P value Crusting E-5 Sinusitis SGS CHL SNHL Total ENT BVAS E-6 CHL ¼ conductive hearing loss; SGS ¼ subglottic stenosis; SNHL ¼ sensorineural hearing loss. 626

6 Items Bloody rhinorrhoea Objective stridor Inflammation on nasal examination TABLE VII. New ENT Scoring System: ENT/GPA DAS. Inflammation on flexible laryngoscopy Inflamed TM*/middle ear Sudden sensorineural hearing loss *TM ¼ tympanic membrane. Description Daily blood stained nasal discharge Stridor assessed by doctor Ulcers, granulation, friable mucosa on rigid nasendoscopy. Excluding crusting Ulcers, granulation, friable mucosa in the larynx Persistent inflammation or granulation tissue in tympanic membrane/middle ear 30db drop in 3 frequencies within 72 hours There is a recognised lack of a gold standard or biomarkers to measure the presence of disease activity. 6 Treatment decision was not used as localized ENTdisease activity is often treated expectantly. Serological tests have not been shown to correlate well with disease activity, so it was not used either. 19 Another difficulty in measuring disease activity accurately is the heterogeneity of the condition and how it manifests differently in each patient. We have therefore used the evaluation of an experienced vasculitis physician as the gold standard for disease activity and the BVAS score to represent the overall clinical impression. For the initial design of the database, a single observer seemed helpful, but the present scoring system will further need to be checked for inter and intraobserver reliability and change over time. (e.g., subglottic stenosis or hearing loss). We have designed a new scoring system for the ENT system specifically in GPA using recognised validation steps. The new scoring system, ENT/GPA DAS has six items and yielded better sensitivity and specificity than BVAS. The selected items did not correlate to infection, which is the main confounding factor in disease activity of the ENT system. Crusting on examination or as a symptom was removed as it correlated to both disease activity and infection in this cohort. Change in sense of smell was also significant in the univariate analysis, but not in the multivariate analysis, and was removed. Sudden sensorineural hearing loss was included as it is a recognised complication of GPA affecting the ENT system and would trigger escalation in immunosuppression. 14,15 Middle ear inflammation, as opposed to conductive hearing loss, was included because it is partly caused by disease in the nose. Items that were considered, but not included were facial pain, mouth ulcers, and cranial nerve palsies (excluding sensorineural hearing loss). Mouth ulcers and cranial nerve palsies occur in GPA and can be associated with ENT disease, 16 but, in the context of disease activity scores, they are grouped under neurology and mucus membranes/ eyes. The proposed scoring system is designed to complement BVAS, and those items would create overlap. Facial pain occurs in relation to disease activity, but was considered to be nonspecific as it is commonly associated with chronic rhinosinusitis and can present as atypical facial pain. 17 The patients in this cohort have the gender distribution and age seen in other published cohorts, 14,16,18 and the sample size is comparable to other cohorts used to describe disease phenotypes. 14,18 A sample size calculation was performed for the proportion of patients with disease activity, and the current sample exceeded the expected number of patients by 15%. The use of one observer with ENT experience and vasculitis training made the assessment consistent. The steps used for the design and initial validation of the scoring system are well recognised and have been used to validate other recognised scoring systems in vasculitis and otolaryngology. 5,12 CONCLUSION This project finalizes the design of a scoring system for ENT disease in granulomatosis with polyangiitis (Wegener s): ENT/GPA DAS. It is objective and easy to score using tools commonly available in the ENT outpatient clinic. It has two items relating to the ears, the nose, and the throat. The current BVAS is adequate in measuring global disease progression and has proven useful in predicting and assessing outcome. The new scoring system for ENT disease in GPA will need to be checked for reliability and interobserver variation before it can be applied to the clinic or in research projects. Acknowledgment The Cambridge Biomedical Research Centre supported this work. BIBLIOGRAPHY 1. Jennings CR, Jones NS, Dugar J, Powell RJ, Lowe J. Wegener s granulomatosis a review of diagnosis and treatment in 53 subjects. Rhinology 1998;36: Gottschlich S, Ambrosch P, Kramkowski D, et al. Head and neck manifestations of Wegener s granulomatosis. Rhinology 2006;44: Flossmann O, Bacon P, de Groot K, et al. Development of comprehensive disease assessment in systemic vasculitis. Ann Rheum Dis 2007;66: Popa ER, Tervaert JW. The relation between Staphylococcus aureus and Wegener s granulomatosis: current knowledge and future directions. Intern Med 2003;42: Suppiah R, Mukhtyar C, Flossmann O, et al. A cross-sectional study of the Birmingham Vasculitis Activity Score version 3 in systemic vasculitis. Rheumatology (Oxford) 2011;50: Merkel PA, Herlyn K, Mahr AD, et al. Progress towards a core set of outcome measures in small-vessel vasculitis. Report from OMERACT 9. J Rheumatol 2009;36: Fokkens W, Lund V, Mullol J. European position paper on rhinosinusitis and nasal polyps Rhinology Suppl 2007: Merkel PA, Cuthbertson DD, Hellmich B, et al. Comparison of disease activity measures for anti-neutrophil cytoplasmic autoantibody (ANCA)- associated vasculitis. Ann Rheum Dis 2009;68: Mahr AD, Neogi T, Lavalley MP, et al. Assessment of the item selection and weighting in the Birmingham vasculitis activity score for Wegener s granulomatosis. Arthritis Rheum 2008;59: Martinez Del Pero M, Rasmussen N, Chaudhry A, Jani P, Jayne D. Structured clinical assessment of the ear, nose and throat in patients with granulomatosis with polyangiitis (Wegener s). Eur Arch Otorhinolaryngol [Epub ahead of print]. 11. Cronbach LJ, Meehl PE. Construct validity in psychological tests. Psychological Bulletin 1955;52: Goodwin LD. Changing conceptions of measurement validity: an update on the new standards. J Nurs Educ 2002;41: Hellmich B, Flossmann O, Gross WL, et al. EULAR recommendations for conducting clinical studies and/or clinical trials in systemic vasculitis: 627

7 focus on anti-neutrophil cytoplasm antibody-associated vasculitis. Ann Rheum Dis 2007;66: Hoffman GS, Kerr GS, Leavitt RY, et al. Wegener granulomatosis: an analysis of 158 patients. Ann Intern Med 1992;116: Kempf HG. Ear involvement in Wegener s granulomatosis. Clin Otolaryngol Allied Sci 1989;14: Stone JH. Limited versus severe Wegener s granulomatosis: baseline data on patients in the Wegener s granulomatosis etanercept trial. Arthritis Rheum 2003;48: Cashman EC, Smyth D. Primary headache syndromes and sinus headache: An approach to diagnosis and management. Auris Nasus Larynx 2012;39: Reinhold-Keller E, Beuge N, Latza U, et al. An interdisciplinary approach to the care of patients with Wegener s granulomatosis: long-term outcome in 155 patients. Arthritis Rheum 2000;43: Jayne DR, Gaskin G, Pusey CD, Lockwood CM. ANCA and predicting relapse in systemic vasculitis. QJM 1995;88: