Materials and Methods

Case selection criteria—Medical records of dogs examined at All Care Animal Referral Center between January 2001 and December 2004 that underwent MRI because of nasal disease were reviewed. Dogs were included in the study if results of nasal MRI were available and a diagnosis had been obtained by means of histologic examination of nasal biopsy specimens obtained during rhinoscopy or rhinotomy. Dogs were excluded from the study if results of MRI or histologic examination were not available, rhinoscopy had been performed prior to MRI, or nasal disease was not present.

Medical records review—Information obtained from the medical records of dogs included in the study consisted of signalment, history, clinical signs, results of MRI, and results of histologic examination of nasal biopsy specimens. Follow-up information regarding clinical outcome that could be used to help support the histologic diagnosis was obtained through December 2006 by means of telephone conversations with owners and referring veterinarians or by means of follow-up examinations.

Magnetic resonance imaging—Magnetic resonance images were acquired with a 0.5 Tesla MRI unit.^c All dogs were anesthetized during MRI and positioned in sternal recumbency. In general, 3- to 5-mm-thick slices were obtained in the transverse, sagittal, and dorsal planes. Sequences used included fast spin echo T1-weighted scans (repetition time, 600 milliseconds; echo time, 30 milliseconds), proton density weighted scans (repetition time, 2,000 to 2,800 milliseconds; echo time, 17 milliseconds), and T2-weighted scans (repetition time, 2,000 to 2,450 milliseconds; echo time, 30 to 120 milliseconds). Additional T1-weighted images were obtained after administration of gadopentetate dimeglumine^d or gadoversetamide^e (0.1 mmol/kg [0.045 mmol/lb], IV). The acquisition matrix was 192 × 256. All MR images had been evaluated by a single board-certified veterinary radiologist (AR). For the present study, the radiologist's reports were reviewed to determine the presence of a mass effect, frontal sinus involvement, sphenoid sinus involvement, maxillary recess involvement, nasopharyngeal invasion by soft tissue, nasal turbinate destruction, vomer bone lysis, nasal turbinate destruction, paranasal bone destruction, cribriform plate erosion, and whether lesions were unilateral or had extended into the contralateral nasal cavity. Sinus involvement and maxillary recess involvement were characterized as either mucosal thickening with exudate accumulation or a soft tissue mass effect. Soft tissue mass effects were further characterized as either enhanced or minimally or not enhanced following contrast administration. Masses that were enhanced following contrast administration were characterized as having homogenous or heterogeneous enhancement.

Nasal biopsy—In all dogs, nasal biopsy specimens had been obtained during rhinoscopy or rhinotomy. Rhinoscopy was performed with a rigid rhinoscope^f,g immediately after MRI. If clinically indicated, the caudal aspect of the nasopharynx was evaluated with a flexible endoscope or bronchoscope.^h,i If a second biopsy specimen was required, rhinoscopy or rhinotomy was repeated at a later date but MRI was not repeated. Biopsy specimens were fixed in buffered 10% formalin and submitted to a commercial pathology laboratory^j for histologic evaluation by board-certified veterinary pathologists. For the present study, histologic reports were reviewed because tissue samples were not available for review. In instances when histologic examination of a follow-up biopsy specimen yielded a different histologic diagnosis than did examination of the initial specimen, results for the follow-up specimens were used as the final histologic diagnosis.

Statistical analysis—The χ² test was used to determine whether individual MRI features (present vs absent) were significantly associated with histologic diagnosis (neoplasia vs inflammation vs fungal rhinitis). Features that were examined consisted of the presence or absence of a mass effect, nasal turbinate destruction, vomer bone lysis, cribriform plate erosion, paranasal bone destruction, frontal sinus involvement, sphenoid sinus involvement, maxillary recess involvement, nasopharyngeal invasion, and lesion extent (unilateral vs bilateral). For features that were found to be significantly associated with histologic diagnosis, the Fisher exact test was used to determine, for the subpopulation of patients with a mass effect, whether that feature (present vs absent) was significantly associated with a histologic diagnosis of neoplasia (yes vs no). All analyses were performed with standard software,^k and all tests were performed as 2-tailed tests. Values of P < 0.05 were considered significant.

Results

Medical records of 128 dogs that underwent nasal MRI during the study period were reviewed, and 78 dogs met the criteria for inclusion in the study. For dogs included in the study, median age at the time of MRI was 9 years (range, 7 months to 14 years). There were 23 castrated males, 12 sexually intact males, 25 spayed females, and 18 sexually intact females. Twenty were mixed-breed dogs, and 58 were purebred dogs representing 35 breeds. The most common breeds were Golden Retriever (n = 7), Labrador Retriever (6), German Shepherd Dog (4), Australian Shepherd (3), Shetland Sheepdog (3), Rottweiler (2), Dachshund (2), Cocker Spaniel (2), Border Collie (2), and Bichon Frise (2).

In 75 dogs, nasal biopsy specimens had been obtained during rhinoscopy, and in 3, biopsy specimens had been obtained during rhinotomy. Two of the 3 dogs in which specimens were obtained during rhinotomy had initially undergone rhinoscopy after MRI, and MRI was not repeated prior to rhinotomy in these dogs. In both of these dogs, the histologic diagnosis for initial biopsy specimens was lymphoplasmacytic rhinitis, whereas aspergillosis was diagnosed on the basis of histologic examination of follow-up biopsy specimens. Two dogs underwent rhinoscopy twice; MRI was not repeated prior to the second rhinoscopy procedure in these dogs. In these 2 dogs, results of histologic examination of initial biopsy specimens were consistent with lymphoplasmacytic rhinitis, but results of histologic examination of follow-up biopsy specimens were consistent with chondrosarcoma in 1 dog and aspergillosis in the other.

Thirty-three dogs were classified on the basis of results of histologic examination as having neoplastic disease. Specific conditions included nasal carcinoma (n = 13), squamous cell carcinoma (4), chondrosarcoma (4), adenocarcinoma (3), lymphosarcoma (3; Figure 1), spindle cell sarcoma (2), undifferentiated sarcoma (1), mast cell tumor (1), hemangiosarcoma (1), and multilobular osteochondrosarcoma (1). Seven dogs were classified as having nasal aspergillosis on the basis of results of histologic examination of biopsy specimens in combination with results of serologic testing and fungal culture. In 5 of these 7 dogs, fungal organisms were seen in histologic sections. In one of the dogs with nasal aspergillosis, nasal mites were seen during rhinoscopy. The remaining 38 dogs were classified as having inflammatory nasal disease identified as lymphoplasmacytic rhinitis (n = 30), suppurative rhinitis (3), mixed rhinitis (3), eosinophilic rhinitis (1), or granulation tissue (1). The dog with granulation tissue had an abscess associated with a retained tooth root.

View Full Size

Nasal MR images from a dog with nasal lymphosarcoma. A—T1-weighted images in the dorsal plane obtained before and after administration of contrast medium. Notice the bilateral heterogeneously enhancing soft tissue mass (arrow head), nasal turbinate destruction, and vomer bone lysis (*). B—Proton density weighted and T2-weighted images in the transverse plane. Notice the nasopharyngeal infiltration (curved arrow), frontal sinus infiltration (open arrow), and sphenoid sinus infiltration (closed arrow).

Citation: Journal of the American Veterinary Medical Association 232, 12; 10.2460/javma.232.12.1844

• Download Figure
• Download figure as PowerPoint slide

Lesion extent (unilateral vs bilateral) was not significantly (P = 0.81) associated with histologic diagnosis (Table 1). However, all other MRI features evaluated were significantly associated with histologic diagnosis. In particular, absence of a mass effect was significantly (P < 0.001) associated with inflammatory disease; all 20 dogs without a mass effect on MR images had inflammatory disease, whereas all dogs with neoplasia or fungal rhinitis had a mass effect. The 20 dogs without a mass effect were described as having thickened nasal mucosa with or without exudates. Five of these dogs had nasal turbinate destruction, and 2 also had vomer bone erosion. Only 2 of these 20 dogs had exudate accumulation in the frontal sinus, and only 1 had exudate accumulation in the maxillary recess.

Because absence of a mass effect could be used to exclude all dogs with neoplasia in this population, analyses were performed on data for the 58 dogs with a mass effect to identify MRI findings that could be used to differentiate dogs with neoplasia from dogs with nonneoplastic conditions (ie, inflammatory or fungal disease; Table 2). In this subpopulation of dogs, vomer bone lysis, cribriform plate erosion, paranasal bone destruction, nasopharyngeal invasion, and sphenoid sinus invasion were significantly associated with histologic diagnosis (neoplastic vs nonneoplastic). Nasal turbinate destruction, frontal sinus invasion, and maxillary recess invasion were not.

In 31 of the 33 dogs with neoplasia, the mass was enhanced following administration of gadolinium. The 2 dogs with masses that were not enhanced or minimally enhanced had a mast cell tumor and a squamous cell carcinoma that primarily affected the nares with minimal extension into the rostral portion of the nasal passages. One dog with chondrosarcoma had a homogeneously enhancing mass involving both nasal passages and maxillary recesses and the right frontal and sphenoid sinuses. The remaining 30 dogs with neoplasia had heterogeneously enhancing masses involving the nasal passages or frontal sinuses.

Eighteen of the 38 dogs with inflammatory rhinitis had a mass effect. In 5 of these dogs, the mass was not enhanced or minimally enhanced following administration of gadolinium, suggesting that masses may have been composed of secretions. Four of these dogs had nasal turbinate destruction, 3 had invasion of the frontal sinus by the mass, and 2 had infiltration of the maxillary recess by the mass. Twelve dogs had heterogeneously enhancing mass effects with exudate in the nasal passages. Eleven of these dogs had 1 or more MRI features that were associated with a histologic diagnosis of neoplasia versus nonneoplastic disease. Two dogs had cribriform plate erosion, 3 had nasopharyngeal invasion by the mass, 8 had vomer bone lysis, 2 had paranasal bone destruction, and 2 had sphenoid sinus invasion. Four dogs, including the 2 with cribriform plate erosion, survived < 3 months, suggesting the possibility of undiagnosed neoplasia. However, 5 dogs survived > 2 years. One dog had a homogenously enhancing mass effect in the nasal passages and nasal turbinate destruction. This dog survived > 3 years with prednisone therapy.

The 7 dogs with aspergillosis all had a mass effect in combination with moderate to severe nasal turbinate destruction that resulted in cavitation of the nasal passages in 6 dogs. Six of the dogs had masses that were heterogeneously enhanced following administration of gadolinium. The remaining dog had a mass involving the right nasal passage and right frontal sinus that was not enhanced following administration of contrast medium. Six of the 7 dogs had infiltration of the frontal sinus with soft tissue and exudate. Five of the 7 dogs had thickened mucosa and a hyperintense rim of tissue in the nasal passage or frontal sinus on T2-weighted and postcontrast T1-weighted images.

In 50 of the 58 dogs with a mass, the mass was enhanced following administration of contrast medium, and in 48 of these 50 dogs, enhancement was classified as heterogeneous, which was considered consistent with exudate intermixed with soft tissue. Histologic diagnosis was not significantly (P = 0.68) associated with type of contrast enhancement (heterogeneous vs homogeneous).

Discussion

Results of the present study suggested that nasal MR image findings can aid in establishing a tentative diagnosis of rhinitis or neoplasia in dogs with nasal disease. The lack of a mass effect on MR images was significantly associated with inflammatory disease. In dogs with a mass effect on MR images, vomer bone lysis, cribriform plate erosion, paranasal bone destruction, sphenoid sinus invasion by a mass, and nasopharyngeal invasion by a mass were significantly associated with a diagnosis of neoplasia.

The population of dogs in the present study was similar to populations reported in previous studies^1,5,6,8,14,a of dogs with nasal disease. Thus, we believe our findings can be generalized to all dogs with chronic nasal disease undergoing MRI.

In the present study, the presence of a mass effect on MR images was not specific for a diagnosis of neoplasia because many dogs with nonneoplastic disease had a mass effect. For example, a dog with granulation tissue secondary to a tooth root abscess and all 7 dogs with fungal rhinitis had mass effects. Low specificity of a mass effect may also have been attributable to misclassification of some dogs identified as having inflammatory disease. Neoplastic disease is frequently associated with inflammation, and histologic examination of small or superficial biopsy specimens obtained by means of rhinoscopy may lead to a misdiagnosis of inflammatory disease. Previous studies^6,15–17 have shown incomplete agreement between results of histologic examination of rhinoscopic or endoscopic biopsy specimens and results of histologic examination of surgical biopsy or necropsy specimens. Rhinotomy allows better visualization of lesions and acquisition of larger tissue samples but was performed in only 3 dogs in the present study. Because of the retrospective nature of the present study, follow-up biopsy or necropsy data were typically not available to confirm the accuracy of histologic diagnoses. In instances when neoplasia was suspected because of a mass effect on MR images but histologic examination of a biopsy specimen revealed only inflammatory disease, repeated rhinoscopy or rhinotomy was recommended. However, these follow-up procedures were performed in only a few patients, and many owners elected palliative treatment or euthanasia because of financial considerations or the patient's poor condition. Some of these patients had short survival times, as would be expected with untreated neoplasia or fungal disease. However, some of these patients had long survival times with only nonspecific treatment for inflammatory rhinitis, exceeding expected survival times for dogs with untreated neoplasia.⁸ Whether these patients could have had granulation tissue or benign neoplastic conditions is unknown, as follow-up MRI or biopsy was not performed.

Low specificity of a mass effect may also have been attributable to misinterpretation of secretions with variable protein concentrations and fungal mycetomas as masses. Magnetic resonance imaging has been reported to be superior to CT or radiography for evaluation of soft tissue structures in nasal disease.^2,13,18 However, in humans, fungal mycetomas and inspissated secretions in the sinuses can be misinterpreted.¹⁸ The protein content of sinonasal secretions can significantly affect signal intensity on both T1- and T2-weighted images, causing various combinations of hypointense or hyperintense signals. The signal intensity on T1-weighted images changes from hypointense to hyperintense as the protein concentration of the secretion increases. Conversely, the signal intensity on T2-weighted images changes from hyperintense to hypointense as the protein concentration increases. Once the protein concentration exceeds 28%, the secretions become inspissated and hypointense on both T1- and T2-weighted images.^18,19 Owing to possible signal overlap between secretions and fungal colonies, it is often not possible to differentiate the two. This was observed in a previous study² of dogs with aspergillosis. In the present study, most masses were described as heterogeneous because of variability of signal intensity and enhancement with contrast. This variability could have been because of secretions with variable protein content adjacent to soft tissue.

Finally, differentiation between soft tissue and secretions may have been affected by image quality and technique. Field strength and matrix size may have decreased resolution. Because of the slow speed of image acquisition, there may have been a time delay between contrast administration and image acquisition, resulting in loss of enhancement of soft tissue. Thus, with the equipment and techniques used, it was sometimes difficult to differentiate secretions from soft tissue structures. Acquisition of fluid-attenuation inversion recovery images may have helped differentiate soft tissue structures in the sinuses and evaluate cribriform plate erosion.¹⁸ Fat suppression techniques are often used in human sinonasal imaging to help delineate soft tissue structures in the sinuses and periorbital regions.¹⁹ Whether these techniques would be useful in dogs is unknown because of the lack of fatty tissue in the sinuses of dogs.

In the present study, the presence of a mass was a sensitive method of identifying neoplasia, in that all 33 dogs with neoplasia had a mass, but was not specific, in that 25 of 58 (43%) dogs with a mass effect did not have a histologic diagnosis of neoplasia. Conversely, the absence of a mass was a specific method for identifying inflammatory disease, in that all 20 dogs without a mass had inflammatory disease, but was not sensitive, in that only 20 of 38 (53%) dogs with inflammatory disease did not have a mass. Thus, results of the present study suggested that absence of a mass on MR images could be used to exclude neoplasia as the underlying cause of nasal disease. Nevertheless, by excluding neoplasia in dogs without a mass on MR images, it is possible that some cases of neoplasia could be missed, especially if the neoplastic process were infiltrative rather than proliferative. However, even the dogs with lymphoma and hemangiosarcoma had large mass effects in the present study.

In the subpopulation of dogs with a mass effect on MR images in the present study, vomer bone lysis, cribriform plate erosion, paranasal bone destruction, sphenoid sinus invasion, and nasopharyngeal invasion were all significantly associated with whether dogs had neoplastic or nonneoplastic conditions. Many of these features have previously been associated with neoplasia in studies^6,14,15 of CT and nasal endoscopy. To the authors' knowledge, however, the importance of sphenoid sinus involvement in nasal disease in dogs has not been reported previously. Also, none of these features were specific for neoplasia, in that all of these features could be identified in at least some dogs with nonneoplastic disease. Importantly, vomer bone lysis was identified in 3 of the 7 dogs with aspergillosis in the present study, which was similar to the proportion of dogs with aspergillosis and vomer bone lysis in a previous report.² Vomer bone lysis was also present in 10 of the 38 dogs with inflammatory rhinitis in the present study, 2 of which did not have a mass effect. This differs from results of a previous study⁶ in which no dogs with inflammatory rhinitis had vomer bone lysis on CT images. Vomer bone lysis has been reported in cats with inflammatory rhinitis.^20,21

For dogs with a mass effect in the present study, frontal sinus invasion and maxillary recess invasion by the mass were not useful predictors of histologic diagnosis (neoplastic vs nonneoplastic). Frontal sinus involvement has been reported to be present in 42% of dogs with lymphoplasmacytic rhinitis, 80% of dogs with aspergillosis, and > 90% of dogs with malignant neoplasia.^1,2,5,22 Maxillary recess invasion on MR images has been reported in 92% of dogs with neoplasia,¹ but to the authors' knowledge, its frequency in inflammatory disease has not been evaluated.

For all dogs in the present study, lesion extent (unilateral vs bilateral) was also not found to be significantly associated with histologic diagnosis (neoplastic vs inflammatory vs fungal). This was not surprising, in that unilateral lesions have been reported in 24% of dogs with lymphoplasmacytic rhinitis, 40% of dogs with aspergillosis, and 50% of dogs with neoplasia.^1,2,5

Nasal turbinate destruction was a common feature in dogs with both neoplastic disease and nonneoplastic disease. Extent of nasal turbinate destruction was not routinely categorized in MRI reports used in the present study, although it can provide some diagnostic information. For instance, all 7 dogs with nasal aspergillosis in the present study were reported to have moderate to severe turbinate destruction, often resulting in cavitation of 1 or both nasal passages. Similar findings have been reported previously.² Nasal turbinate destruction has also been seen on CT images from 70% of dogs with lymphoplasmacytic rhinitis, with most dogs having mild to moderate destruction.⁵ Because of the overlap with these findings, moderate to severe turbinate destruction with cavitation of the nasal passages in association with nonspecific biopsy findings should warrant further diagnostic testing to help differentiate fungal from inflammatory rhinitis. The presence of a hyperintense rim of tissue or mucosa in the nasal cavity or frontal sinus on T2-weighted and postcontrast T1-weighted images may also increase the likelihood of fungal rhinitis, as previously reported.²

The retrospective nature of the present study posed several limitations. Surgical biopsy and necropsy data that could be used to corroborate the histologic diagnoses were available for only a few patients. Additionally, tissue blocks and slides were not available for review by a single pathologist. Histologic examinations of nasal biopsy specimens were performed by several pathologists who likely differed with regard to how likely they were to offer a definitive diagnosis, which may have affected our results.