Cocaine-induced midline destructive lesions: clinical, radiographic, histopathologic, and serologic features and their differentiation from Wegener granulomatosis.

Abstract

Introduction Habitual nasal insufflation of pulverized cocaine may cause mucosal lesions. Mild changes cause nasal stuffiness, headache, or hyposmia. If cocaine use becomes chronic and compulsive, progressive damage of the mucosa and perichondrium leads to ischemic necrosis of the septal cartilage and perforation of the nasal septum. Occasionally, cocaine-induced lesions cause extensive destruction of the osteocartilaginous structures of the nose, sinuses, and palate that mimics the clinical picture of other diseases associated with necrotizing midfacial lesions (1,4,17,20,31,38,43,49,58,62,63,66,74). The mucosal damage induced by cocaine is multifactorial. The vasoconstrictive effect of the drug is thought to be the most important factor (8,17,20,38). However, the irritant effect of adulterants of the drug, the traumatic effect on the mucosa caused by cocaine crystals insufflated at high velocity, and recurrent nasal infections all seem to contribute to chronic tissue destruction (20,38). Progressive nasal obstruction, epistaxis with crusting, and ulceration of the nasal mucosa with or without septal perforation are also characteristic manifestations of nasal involvement by Wegener granulomatosis (WG). The differentiation of cocaine-induced midline destructive lesions (CIMDL) and limited WG may be difficult, particularly if the patients do not readily admit to their substance abuse. Antineutrophilic cytoplasmic antibodies (ANCA) directed against proteinase 3 (PR3) or myeloperoxidase (MPO) are sensitive and specific markers for the idiopathic small vessel vasculitides including WG (30). It is generally believed that the presence of a positive ANCA test result with either of the 2 antigen specificities facilitates the differential diagnosis of WG. However, instances of positive ANCA test results have been reported in patients with lesions attributed to cocaine abuse (4,29). We found positive ANCA test results in an unexpectedly large proportion of patients with CIMDL. In several instances their lesions were clinically indistinguishable from WG limited to the upper respiratory tract (28). This seems to limit the usefulness of routine ANCA testing for the unequivocal differentiation of cocaineinduced nasal lesions from limited WG. We performed the present study to identify clinical, radiographic, and histopathologic features that allow the distinction of patients with severe CIMDL from those with WG and to further characterize their ANCA. Patients and Methods Patients Between January 1991 and December 1999, 18 cocaine abusers with midline destructive lesions were evaluated in the Department of Otolaryngology of the University of Brescia, Italy (Table 1). The patients ranged in age from 22 to 66 years (median, 37 yr). Ten were men and 8 women. The follow-up period ranged from 12 to 108 months (median, 34 mo). At the time of first observation, all patients except 1 admitted to cocaine use. In 7 patients reliable information on abuse duration and dose could not be obtained, 2 patients had a history of abuse with an undetermined dose lasting 6 and 8 years, respectively. One patient had been using 1–3 g per week irregularly. The remaining 8 patients had been using cocaine for 2–30 years, at a dose ranging from 1 to 15 g per week.TABLE 1: Clinical features and laboratory values at presentation of patients with CIMDL and WGThe control population consisted of all 21 consecutive WG patients who underwent a nasal biopsy in the Department of Otolaryngology of the University of Brescia during the same time frame (see Table 1). Patients’ ages ranged from 30 to 64 years (median, 45 yr). Nine were men and 12 women. The follow-up period ranged from 13 to 108 months (median, 52 mo). Nineteen patients were evaluated at the time of first diagnosis, 2 at the time of their first relapse. Five WG patients had generalized disease with renal involvement, 3 had limited disease involving only the nasal and tracheal mucosa. The remaining 13 patients had lung involvement and other systemic symptoms but no kidney involvement. All patients had biopsy-proven WG or satisfied the Chapel Hill Consensus Conference definition (35) of the disease. Clinical and laboratory evaluations Physical examination included inspection of the face, oral cavity, and oropharynx; inspection of the nasal cavities and nasopharynx using 0° and 30° rigid telescopes, 4 mm in diameter, and the flexible fiberscope. Multiple biopsies and samples for bacterial and fungal cultures were taken under endoscopic guidance. To monitor the clinical course of the disease, digital images of the relevant endoscopic features were archived. Erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), complete and differential blood counts, liver and kidney function tests, urinalysis and microscopy, and serology for hepatitis virus B and C and human immunodefiency virus (HIV), were performed in every patient. In addition, antinuclear antibodies (ANA), rheumatoid factor (RF), and complement levels were determined in most patients. ANCA determination All sera from CIMDL and WG patients were tested for ANCA in 3 different laboratories. ANCA tests were first performed at the time of the patients’ clinical evaluation in the laboratory of the Department of Clinical Immunology of Spedali Civili, University of Brescia, Italy. Sera were analyzed by indirect immunofluorescence microscopy (IIF) on ethanol-fixed blood donor neutrophils following the standard procedure delineated at the first ANCA workshop (77). Sera were further tested by enzyme-linked immunosorbent assays (ELISA) for the presence of antibodies reacting with PR3 (PR3-ANCA) and MPO (MPO-ANCA). PR3, purified as previously described (65), and MPO obtained from a commercial source (Calbiochem Biosciences, Inc. La Jolla, CA) served as antigens in the assays. Aliquots of the sera were subsequently analyzed at the Mayo Clinic (Rochester, MN) by IIF and ELISA for MPO-ANCA following a standardized test algorithm (68). For some of the WG patients only convalescent sera were available for testing at Mayo (see Table 2). In addition, the sera were analyzed in the research laboratory of 1 of the authors (US) by IIF for PR3-ANCA and human leukocyte elastase (HLE)-ANCA using ethanol-fixed HMC-1 cells that expressed recombinant PR3 (rPR3) or HLE (rHLE) (56,69). Furthermore, the sera were tested for PR3-ANCA by capture-ELISA using purified neutrophil PR3 (Athens Research, Athens, GA) as target antigen (71). IIF results obtained using neutrophils were expressed qualitatively as showing characteristic cytoplasmic staining (C-ANCA) or perinuclear staining (P-ANCA). IIF results using HMC-1 cells expressing rPR3 or rHLE were reported as positive when characteristic IIF staining was detected at a serum dilution of 1:4 or higher that was not detectable on sham-transfected HMC-1 (HMC-1/VEC) control cells (69). Solid-phase assay results were expressed in arbitrary units.TABLE 2: ANCA test results in 18 patients with CIMDL and 21 patients with nasal destructive lesions caused by WG *Imaging studies Sixteen of the 18 CIMDL patients underwent cross-sectional imaging. A total of 24 examinations were evaluated: 7 computed tomography (CT) studies in 5 patients, and 17 magnetic resonance (MR) studies in 11 patients. Four patients had more than 1 imaging study (range, 2–5). Sixteen of the 21 WG patients were evaluated by cross-sectional imaging: 2 by CT and 14 by MR. The imaging studies were obtained shortly after the endoscopic examination in all patients (range, 2–25 d). Of the CIMDL patients, 4 patients had additional examinations (n = 8) during the follow-up (range, 7–52 mo). All MR studies were performed at the Department of Radiology of the University of Brescia using superconductive equipment (Siemens Magnetom SP 1.5 T, Symphony 1.5 T). In all MR studies, spin echo or turbo spin echo T2 sequences were acquired along with enhanced T1 sequences. All CT examinations were performed without contrast enhancement. Two of 5 CT studies were performed at the Department of Radiology, the others at different hospitals. All imaging studies were reviewed by 1 of the authors (RM). We analyzed the degree of septal destruction, that is, its area, obtained by multiplying the maximum diameters of the eroded septum, erosion of adjacent nasal structures (inferior, middle, superior turbinates), lateral nasal wall (medial antral wall, lamina papyracea) and floor of the nasal cavity (hard and soft palate). A score was obtained by assigning 1 point for each single structure involved. We evaluated abnormal changes of signal intensity of the mucosa on plain and enhanced MR sequences. In addition, we evaluated changes in size and signal of Waldeyer ring sites and abnormalities of the middle ear, and compared imaging studies obtained at the time of the first endoscopic examination. Follow-up studies were considered separately. Histopathologic evaluation A total of 44 mucosal biopsies from nasal cavities and paranasal sinuses were evaluated from the 18 CIMDL patients. Five patients had more than 1 biopsy (range, 2–16). Sections were stained with hematoxylin-eosin. Orcein staining was used to evaluate elastic fibers, and periodic-acid Schiff and Ziehl-Neelsen stains were used to identify fungi and mycobacteria, respectively. Polarizing filters were used to identify birefringent foreign material. Additional sections were immunostained for lymphoid pan-B-cell, CD20; pan-T-cell, CD3; Natural Killer (NK) cell, CD56 (Bio-SPA, Milan, Italy); and nonlymphoid, pan-macrophage, CD68 leukocyte-associated antigens. Stains for EBV-associated latent membrane protein 1 and RNA were performed using immunohistochemistry (anti-LMP1) and in situ hybridization (EBER-1,2), respectively. Twenty-nine nasal biopsies were obtained from 21 patients with WG and evaluated in the same fashion. (Unless otherwise indicated, all antibodies used were purchased from Dako, Milan, Italy.) Statistical analysis Quantitative variable analysis was performed using the non-paired Student t-test. The qualitative histopathologic variables between the 2 groups were compared using the Fisher exact test. The radiologic variables between the 2 groups were compared using the chi-square test and the Pearson correlation coefficient. For all comparisons, p values of 0.05 or less were considered significant. Results Clinical and laboratory findings All patients with CIMDL sought an otolaryngologic consultation because of longstanding symptoms including nasal obstruction, epistaxis, and severe facial pain. The most common findings at the first visit were diffuse necrotizing ulcerative lesions, extensive crusting, and septal perforation. Destruction of the septum and inferior turbinates was invariably found in all patients. In more severe cases it extended to the middle and superior turbinates and the lateral wall of the nose. The latter was entirely absorbed in 3 patients. Hard and soft palate perforations were present at initial presentation in 2 patients and became evident during follow-up in 3 additional patients (Figure 1). The lesions gave rise to dysphagia and nasal reflux, and substantially affected the patients’ quality of life. An example of progressive destruction of the midline structures is shown in (Figure 2).Fig. 1: Different examples of palate perforation in patients with cocaine-induced midline destructive lesions. The lesion can vary from a small defect to a large erosion involving the entire hard palate and extending to the alveolus.Fig. 2: Temporal progression of a midline destructive lesion in a patient with cocaine-induced midline destructive lesions over 3 years. In 1996, a large hard palate perforation was evident at inspection of the oral cavity (upper left panel). Magnetic resonance (MR) studies also demonstrated destruction of the inferior, middle, and superior turbinates on both sides and partial reabsorption of the left medial wall (upper right panel). By 1999, the lesion had progressed dramatically to destruction of the central part of the anterior skull base. The exposed dura and both frontal recesses are clearly visible in the endoscopic picture taken with a 70° scope (lower left panel). MR image shows the presence of a huge meningoencephalocele (lower right panel).None of the CIMDL patients complained of symptoms suggesting ear, orbit, or lung involvement at initial presentation. During the course of the disease, 2 patients with severe destruction developed acute orbital symptoms caused by propagating infections associated with pseudotumor, proptosis, and diplopia. One patient developed a Staphylococcus aureus lung abscess. All these lesions resolved rapidly with appropriate antibiotic therapy. At presentation, constitutional symptoms such as fever, malaise, weight loss, as well as arthralgia or myalgia were universally absent in this patient group, and none of the patients had any symptoms or laboratory findings that would indicate a systemic disease process. The clinical presentation of patients with WG was different. The degree of nasal destruction was less severe than in patients with CIMDL, and signs of other organ involvement were present in most of the WG patients. All 21 patients presented with nasal crusting, but only 3 with a septal perforation. None of the WG patients had involvement of the turbinates, lateral nasal wall, or palate. A comparison of the affected sites is shown in Figure 3.Fig. 3: Comparison of anatomic sites involved in cocaine-induced midline destructive lesions (CIMDL) and Wegener granulomatosis (WG) patients. The percentages indicate the proportion of patients of each group that had involvement of the indicated sites.In patients with WG, nasal symptoms were usually only 1 of several complaints. Clinical evidence of involvement of sites other than nose and sinuses were detected in all patients. Systemic symptoms and alterations in the blood tests were absent in only 2 WG patients at the time of initial presentation. Both were women with subglottic stenosis as the only extranasal manifestation of the disease. All other patients had constitutional symptoms and arthralgia or myalgia. The lungs, ears, and cranial nerves were the other sites most frequently affected. Laboratory abnormalities were significantly more frequent in WG than CIMDL patients. Among the 21 WG patients anemia of chronic disease was found in 16, elevation of white blood counts in 12, and of platelet counts in 10. Increased levels of CRP and ESR were detected in 15 and 14 of the WG patients, respectively. Of the 18 CIMDL patients, 2 had anemia, 5 had elevated white blood counts, and 2 had elevated platelet counts. Abnormal CRP and ESR values were found in 9 and 8 CIMDL patients, respectively, but the mean values were significantly lower than those found in WG patients (see Table 1). Microhematuria and proteinuria were present in 13 of 21 WG patients, even in the presence of normal serum creatinine values. Several WG patients also had ANA (6 of 18 tested) or RF (5 of 11 tested). All CIMDL patients had normal liver and renal function test results and tested negative for hepatitis B antigen, hepatitis C, and HIV antibodies. Fifteen of 17 CIMDL patients tested at the initial visit had positive nasal cultures for S. aureus. Thirteen of 21 WG patients were on broad-spectrum antibiotics at the time of diagnosis and, therefore, were not tested for S. aureus. Five of the remaining 7 WG patients not exposed to antibiotics had positive S. aureus nasal cultures. ANCA results All patients were tested for ANCA at initial presentation because WG was considered in the differential diagnosis of all necrotizing nasal lesions. A surprisingly large number of ANCA tests were positive in patients with CIMDL. The ANCA test results are summarized in Table 2. Only 5 of the 18 CIMDL patients were negative by all ANCA tests performed. In 8 patients, IIF was strongly positive for P-ANCA. All sera were negative for MPO-ANCA by ELISA. Three of the P-ANCA-positive sera reacted with rHLE, and 4 were positive in 1 or more of the tests performed to detect PR3-ANCA (direct ELISA, capture ELISA, or IIF on HMC-1/PR3 cells). Two of the P-ANCA-positive sera reacted with both HLE and PR3. Three of the P-ANCA-positive sera were negative for PR3-ANCA and HLE-ANCA. Five patients with CIMDL had C-ANCA by IIF. All of them were positive for PR3-ANCA in at least 2 of the target antigen-specific assays. The IIF pattern of C-ANCA in the 5 patients was indistinguishable from that typically found in WG. IIF titers and direct ELISA assay units were as high as those encountered in WG patients. Two of the 5 C-ANCA-positive, PR3-ANCA-positive sera also reacted with HLE. Two others were completely negative in the PR3-ANCA capture ELISA although they displayed strong reactivity in the other PR3-ANCA assays. Of the 21 WG patients, only 2 were ANCA negative (see Table 2). These were the patients with biopsy-proven disease affecting only the nose and subglottic area. Four patients had a P-ANCA pattern with corresponding positive MPO-ANCA ELISA test results, and 14 had a C-ANCA pattern with matching PR3-reactivity. Only 1 C-ANCA -positive WG patient reacted strongly with MPO but not PR3, a phenomenon that has been reported rarely (61). None of the ANCA-positive sera from WG patients reacted with HLE. Taken together, the data indicate that routine ANCA testing alone does not allow the differentiation between cocaine-induced and WG-associated nasal lesions. However, more detailed analyses of ANCA specificities suggest that the ANCA immune response in patients with CIMDL differs from that usually encountered in WG patients. In WG it is much more target-antigen restricted. Radiographic findings To identify imaging characteristics that help to differentiate cocaine-induced lesions from WGassociated lesions, we analyzed the cross-sectional imaging studies obtained at initial presentation in patients of both groups (Table 3).TABLE 3: Radiographic changesAll CIMDL patients had septal perforations. Twelve of 16 (75%) had at least partial destruction of the inferior turbinate, which was bilateral in 9. Ten of the 16 cocaine abusers (62.5%) had partial or total destruction of at least 1 middle turbinate, which was bilateral in 5. Erosion of the superior turbinates was identified in 2 patients. The lateral nasal wall was eroded in 5 of the 16 (31.25%) patients. All had destruction of the medial maxillary wall; in 1 patient the erosion extended to the lamina papyracea. The floor of the nasal cavity was eroded in 4 of 16 patients (25%). In 1 of them the involvement extended to the soft palate. In contrast, the imaging studies obtained at initial presentation in WG patients revealed a nasal septal perforation in only 2 of 16 (12.5%). The difference in frequency of nasal septal perforation was statistically significant (p < 0.0005) with a positive predictive value of 88.9% for CIMDL. The involvement of a second nasal structure in addition to the nasal septum was discriminating between the 2 groups, as it was present in 75% of CIMDL patients but not in any of the WG patients. Accordingly, the average radiographic score of affected structures was significantly higher in CIMDL patients (4.06; range, 1–10), compared with WG patients (0.13; range, 0–1; p < 0.001). A correlation was present between the extent of nasal septum perforation (mm 2 ) and the score. The correlation index was 0.803 in CIMDL patients (p < 0.001) and 0.708 in WG patients (p = 0.002). In addition, the occurrence of hard palate perforation in CIMDL patients was correlated with the destruction of bilateral inferior (Pearson 0.509; p = 0.044) or bilateral middle (Pearson 0.545; p = 0.029) turbinates. In CIMDL patients, areas of an abnormal signal of nasal or paranasal mucosa such as hypointensity on T2 and reduced or nonhomogeneous enhancement were detected on 4 (36.36%) and 6 (54.54%) of the 11 MR scans, respectively. On the 14 MR scans obtained in WG patients, hypointensity on T2 and reduced or nonhomogeneous enhancement were present in 3 (21.4%) and 6 (42.8%) patients, respectively. Although MR abnormalities of the nasal mucosa were detected more frequently in CIMDL than in WG patients, the difference was not statistically significant. However, when only the mucosa of the central facial structures, that is, the nasal septum and the adjacent inferior and middle turbinates, was considered, reduced or nonhomogeneous enhancement was significantly more frequent in CIMDL than in WG patients (45.45% versus 16.66%; p < 0.05). Abnormal thickening of the mucosa of the paranasal sinuses was observed more frequently in WG (78.57%) than in CIMDL patients (54.54%), but the difference was not significant. Significant enlargement of the palatine or pharyngeal tonsils was found on the imaging studies in 9 of 16 (56.25%) CIMDL patients. It was associated with small fluid collections within lymphatic tissue in most cases. Similar chronic inflammatory changes of the Waldeyer ring were significantly less common in WG patients (n = 2, 12.5%; p < 0.05). Radiographic signs of otitis media were detected in only 1 CIMDL patient but in 5 WG patients (31.25%). All had different degrees of hearing impairment. In 1 of the WG patients, enhanced tissue within the middle ear was detected which turned out to be granulomatous tissue causing destruction of the ossicle chain. The difference of radiographic middle ear abnormalities did not reach statistical significance. Follow-up imaging studies were obtained in 4 of the CIMDL patients. All of them admitted to continued cocaine abuse. All showed progressive erosion of nasal structures. In 1 patient there was progressive “centrifugal” involvement of the lateral nasal walls and floor with eventual destruction of the entire framework and complete erosion of the hard and soft palate. The central floor of the anterior skull base was also eroded (see Figure 2). Comparison of the imaging abnormalities detected in CIMDL and WG patients corroborated the clinical findings indicating that the severity of tissue destruction is much more severe in CIMDL than in WG patients. Histopathologic findings To determine which histopathologic features are discriminating for CIMDL and WG, we reviewed all nasal biopsy specimens of each patient. Of the 44 biopsy specimens obtained from the CIMDL patients, 19 showed nonspecific changes consisting of fibrosis with mild inflammation or extensive necrosis. The remaining 25 biopsies (57%) showed significant histologic abnormalities (Table 4). In all these biopsies, a dense inflammatory infiltrate of mononuclear cells admixed with neutrophils and eosinophils was present. Nuclear atypia was universally absent. The inflammatory cells frequently encroached upon the wall of venules and arterioles, resulting in variable degrees of narrowing of the lumen. The feature, referred to as “perivenulitis,” was found in 96% of the biopsies (Figure 4A). Microabscesses involving the wall of venules were found in 10 (40%) biopsies (Figure 4B). Obvious leukocytoclastic vasculitis with fibrinoid necrosis was identified in only 7 biopsies (28%) (Figure 4C). Other vascular changes detected in 7 (28%) biopsies consisted of fresh thrombi or sclerotic changes of the vascular wall, likely representing postthrombotic scars (Figure 4D). In 4 samples, the abnormalities involved medium-sized arterioles. It is noteworthy that the leukocytoclastic vasculitis and arteriolar thrombosis coexisted only in a single biopsy. Intra- or extravascular granulomas, scattered giant cells, or deep tissue necrosis with microabscesses were not found in any biopsy of the CIMDL patients.TABLE 4: Histopathologic features in nasal biopsies from 18 patients with CIMDL and 21 patients with WGFig. 4: Histopathologic features of nasal biopsies from cocaine-induced midline destructive lesions (panels A–D) and Wegener granulomatosis (WG) (panels E and F) patients. A. Dense polymorphic perivenular infiltrate, narrowing the lumen without obvious destruction of the vessel wall, referred to as “perivenulitis.”B. The inflammatory infiltrate is particularly dense, and the formation of microabscesses within the vascular wall is recognizable (arrowheads). C. Fibrinoid necrosis and nuclear fragmentation identify leukocytoclastic vasculitis. D. The lumen of a small artery is occluded by scarlike tissue, probably resulting from an organized thrombus. Arrowheads indicate the external elastic lamina. E and F. Histopathologic features only seen in WG lesions are multinucleated giant cells scattered within a dense inflammatory infiltrate (E) and associated with deep tissue necrosis (F). All biopsies were stained with hematoxylin and eosin. (Original magnification for A, 190 ×; B, D, E, F, 95 ×; C, 285 ×).When the histopathologic features of CIMDL patients with and without ANCA were compared, it appeared that occurrence of leukocytoclastic vasculitis was more frequent in the ANCA-positive than ANCA-negative patients (41.6% versus 20%, respectively). However, the difference did not reach statistical significance (p = 0.6). The immunohistochemical composition of the mononuclear inflammatory cells showed a prevalence of CD3+ T cells, with a few CD56+ NK cells, and variable numbers of CD68+ macrophages. CD20+ B lymphocytes were rare or totally absent (data not shown). All biopsies were negative for infectious organisms as well as for birefringent foreign material. Immunohistochemical and in situ hybridization studies aimed at the detection of Epstein-Barr virus antigen and RNA were also negative (data not shown). The histopathologic changes observed in biopsies obtained from the WG patients are also summarized in Table 4. Significant histologic abnormalities were noticed in 22 biopsies (76%). A dense inflammatory cell infiltrate, with cell composition, distribution, and immunohistochemical characteristics similar to those found in the CIMDL group was detected. Perivenulitis was recognizable in all biopsies (100%). Other vascular changes consisted of microabscesses in the vascular wall (50%), leukocytoclastic vasculitis with fibrinoid necrosis (64%), and fresh or organized thrombi (23%). Pathognomonic histopathologic features (15,22) such as extravascular multinucleated giant cells (Figure 4E) or granulomas and microabscesses with deeply located necrosis (Figure 4F), were detected in 41% and 86% of biopsies, respectively. In summary, biopsies with nonspecific changes were more frequent in CIMDL (44%) than in WG patients (24%), but the difference was not statistically significant. Microabscesses in the vascular wall and perivenulitis were observed with similar frequencies in both groups. Leukocytoclastic vasculitis and fibrinoid necrosis appeared to be more frequent in WG (p = 0.02). However, when the data analysis was based on the occurrence of the lesion in individual patients rather than individual biopsies, no difference was detectable: it occurred in 6 of 18 CIMDL and in 9 of 21 WG patients (p = 0.11). In contrast, extravascular changes consisting of stromal granulomas with giant cells, microabscesses, and deeply located necrosis were features exclusively encountered in WG (p < 0.001). Discussion The abuse of cocaine may cause complications that can affect any organ system (16). Nasal septal perforation was first reported in 1912 (52). Despite the widespread abuse of cocaine, reports of CIMDL are scarce, and its incidence remains unclear. To our knowledge, only 24 cases with adequate clinical and pathologic documentation have been reported to date (4,6,8,17,20,31,38,39,43,49,58,60,62,63,66,75). Only 5 of the reports (4,17,31,62,66) include ANCA test results, most of which were negative. The 2 reports of positive ANCA test results do not include target antigen specifications (4,66). Based on its high specificity for the small vessel vasculitides including WG, a positive ANCA test result was thought to be of differential diagnostic value in this setting (62). In contrast, most of our patients presenting with CIMDL had ANCA. Because of this association they were at risk to be misdiagnosed as having WG. Indeed, some were initially treated with cytotoxic agents to no avail. Therefore, the identification of clinical, radiographic, and histopathologic features that allow a distinction between the 2 entities is important. Clinical and radiographic differentiation of CIMDL and WG Comparing the clinical and radiographic findings in patients with CIMDL with those of WG patients clearly indicates that the degree and localization of inflammatory changes differ substantially between the 2 groups. Furthermore, centrifugal progression of the necrosis from the nasal septum toward the nasal walls was apparent in CIMDL patients. In addition to the ubiquitous nasal septum perforation, most CIMDL patients had erosion of 1 (75%) or both (56.25%) inferior turbinates. Destruction of the middle turbinates (62.5%) occurred only in patients who also had inferior turbinate involvement. Among 18 cases reported in the literature with detailed imaging data (1,2,4,8,9,17,31,38,49,60,62,66,74,75), 1 or both inferior turbinates were affected in all patients, and 1 or both middle turbinates were involved in 89%. The difference may reflect referral and reporting bias. Further destruction of the midfacial architecture involving the lateral nasal wall (medial maxillary sinus wall 31.25%, lamina papyracea 6.25%) or nasal floor structures (hard palate 25%, soft palate 6.25%) was always associated with involvement of inferior and middle turbinates. The notion of centrifugal progression is corroborated by the strong correlation between the extent of septal perforation and the number of nasal structures eroded. In contrast, erosion of structures of the midline was rare among WG patients (12.5%) and was limited to the nasal septum. The localization of mucosal inflammation detected on imaging studies may provide fu