Abstract 

Computed tomography (CT) was performed on 36 dogs with nasal aspergillosis to assess whether this imaging technique can be used to predict the success of a noninvasive intranasal infusion of enilconazole. A CT score based on the severity of the disease was given to each dog, prior to treatment, by dividing the nasal cavities and frontal sinuses into 8 anatomical regions. After therapy, the dogs were classified into 2 response groups (success group: dogs cured after 1 treatment; failure group: dogs needing more than 1 treatment or with treatment failure). No significant relationship on the logistic scale was found between the CT score and the response to treatment. High sensitivity (treatment failures correctly predicted) and specificity (treatment successes correctly predicted) could not be obtained at the same time, whatever the cut-off value chosen. The results of this study suggest that CT cannot predict the therapeutic success of nasal aspergillosis in dogs treated with a 1-hour infusion of enilconazole. However, dogs with a low score seem to be good candidates to respond after 1 treatment.

Mycotic infection is a common cause of chronic nasal disease in the dog (1). Aspergillus fumigatus , an ubiquitous soil saprophyte, is the most common causative agent (1). Effective treatment of this disease is still challenging. Systemic administration of thiabendazole, ketoconazole, or fluconazole resulted in clinical cure in approximately 50% of dogs, and of itraconazole in 70% of dogs (2,3,4,5,6). Topical administration of antifungals (twice daily for 7 to 14 d) by surgically implanted drains into the frontal sinuses, the nasal cavities, or both was effective in 77% to 89% of dogs treated with either clotrimazole or enilconazole (7,8). Recently, a noninvasive technique using nonsurgically placed catheters has been developed to infuse the topical drug into the nasal cavities and frontal sinuses under general anesthesia (8,9). Prior to treatment, the nasal cavities are isolated by occluding the nasopharynx and nares with Foley catheters (8,9). Then, an antifungal is instilled into the nasal cavities and left “in situ” for a period of 1 h (8,9). This technique results in a more complete distribution of the drug and is associated with fewer complications than is a topical treatment through surgically placed catheters (8,9). It was associated with a clinical cure in 47% to 65% of dogs after 1 treatment and 87% after repeated treatments with either clotrimazole or enilconazole (8,9). Currently, several treatment protocols with variations of this technique have been under investigation to improve therapeutic success, tolerance by the animal, and compliance by the owners (9,10,11,12,13,14). Relapse at a later date is very unusual for dogs in which fungal elimination has been confirmed by a reexamination (1). A grading system based on clinical, radiographic, and rhinoscopic examinations has been used to evaluate the severity of the lesions before treatment and the clinical cure after treatment (4).

The technique of computed tomography (CT) is becoming more available in veterinary medicine and is widely used for examination of head disorders, including the evaluation of the nasal cavities and associated structures (15,16,17,18,19,20,21,22,23,24). Computed tomography offers several advantages relative to conventional radiography for examination of the nasal cavities and frontal sinuses: cross-sectional images that eliminate the superimposition of different structures, adjustment of the contrast scale to optimize optical density and discriminate even the fine turbinate structures, multiplanar reconstructions for better evaluation of the cribriform plate, and, with helical CT, the possibility to perform examinations under deep sedation instead of general anesthesia (24,25,26). Not surprisingly, CT has been shown to be more sensitive than radiography in detecting nasal disease and defining of the extension of the lesions (16,19,20). In dogs with nasal aspergillosis the efficacy of the distribution of an antifungal into the nasal cavities and frontal sinuses when using a noninvasive technique has been demonstrated by CT (27).

Computed tomography has been used, in both human and veterinary medicine, in attempts to predict the outcome of multiple disease processes on the basis of an accurate evaluation of the extension of the lesions, particular CT features, scoring or scaling systems, or perfusion studies (28,29,30,31). Mathews et al (8) used CT in 32 dogs with nasal aspergillosis to try to predict the likelihood of a successful treatment by using a scoring system based on the severity of the lesions. A cut-off point was determined at which a sufficiently high sensitivity and specificity could be obtained. They concluded that their scoring system could be used by others to evaluate CT images of dogs with nasal aspergillosis and to inform owners about the likelihood of a favorable response to treatment (8).

The purpose of this study was to evaluate the value of CT to predict the effect of therapy in 36 dogs with nasal aspergillosis treated with a noninvasive infusion of enilconazole 1%.

Thirty-six dogs were used in this study. All tests made on these dogs were part of the routine clinical examination in dogs with chronic nasal disease and were performed with the owners consent. Fourteen different breeds were represented: rottweiler (n = 9), golden retriever (n = 6), Labrador retriever (n = 5), Belgian shepherd dog (n = 3), Newfoundlander (n = 2), Afghan hound (n = 2), German shepherd dog (n = 2), Doberman pinsher (n = 1), Alaskan malamute (n = 1), basset hound (n = 1), bull terrier (n = 1), pointer (n = 1), Pyrenean shepherd dog (n = 1), and teckel (n = 1). The mean age of the dogs was 4.4 y. There were 19 males and 16 females. Physical, serologic, imaging (radiography and CT), and rhinoscopic examinations were carried out for all dogs. During rhinoscopy, swabs, cytobrush, and biopsies were obtained for culture, cytologic examination, and histologic examination, respectively. Diagnosis of nasal aspergillosis was based on at least 3 positive diagnostic tests, including direct visualization of fungal colonies at rhinoscopy.

Computed tomography was performed with a 4th generation helical CT (Picker 6000 PQ; Picker, Eastlake, Ohio USA). General anesthesia was induced with droperidol and fentanyl (Thalamonal; Janssen-Cilag, Beerse, Belgium), 0.08 mg/kg body weight (BW), IV, and penthotal (Phenobarbital; Abbott, Abbott Park, Illinois USA), 5 to 15 mg/kg BW, IV, and then maintained with halothane 1.5% to 2% (Fluothane; Zeneca, Wilmington, Delaware USA). All dogs were placed in ventral recumbency. Transverse contiguous slices were obtained from the caudal level of the frontal sinuses to the nostrils. Technical settings were 110 kV, 125 mA, pitch 1.5, slice thickness 3 mm. Pre- and postcontrast (700 mgI/kg BW, IV, of a nonionic iodine contrast medium) (Omnipaque 300; Nycomed, Brussels, Belgium) studies were performed. Reformatted dorsal plane images were also obtained. Hard copies were printed with a bone window (window width (WW) 3500 — window level (WL) 500) and a soft-tissue window (WW 340 — WL 25). When necessary, the WW and WL were adjusted on the computer monitor for visualization of other structures.

The CT studies of these 36 dogs were scored retrospectively by a board-certified radiologist (JHS). The same scoring system as described by Mathews et al (8) was used to score the nasal cavities and frontal sinuses. Therefore, each nasal cavity was divided in 4 anatomical regions, for a total of 8 anatomic sites: region I — the nasal turbinates rostral to the maxillary recess, region II — the maxillary turbinates at the level of the maxillary recess, region III — the ethmoid turbinates caudal to the maxillary recess, and region IV — the frontal sinus (Figure 1). Each of the 8 anatomic sites was given a score of 0 to 3 based on the severity of the lesions (0 = no abnormality, 1 = mild turbinate atrophy or fluid accumulation, 2 = moderate disease, 3 = severe disease). For the frontal sinus (region IV), periorbital invasion and frontal bone changes (hyperostosis/lysis/mixed) were also used to evaluate the severity of the disease. The scores were added to provide a total score for each dog. Maximum possible total CT score was 24.

All dogs were treated with a 1-hour infusion of 1% (10 mg/mL) enilconazole (Imaverol; Janssen-Cilag), delivered via nonsurgically placed catheters by using a technique comparable with that described by Mathews et al (21). The dogs were evaluated clinically and rhinoscopically after 3 to 4 wk and the treatment was repeated until clinical and rhinoscopic healing was achieved. The dogs were classified into 2 response groups on the basis of their response to treatment (success group: dogs cured after 1 treatment; failure group: dogs cured after more than 1 treatment or not cured).

A logistic regression analysis, using the CT score as a continuous independent variable and treatment success or failure as a dependent variable, was performed to evaluate the relationship between the CT scores and the probability of failure. Both total score and score of the 4 deepest locations (regions III and IV) were used, multiplying the latter by 2 in order to work on the same scale. A receiver operating characteristic (ROC) curve was derived connecting pairs of sensitivity (percentage of dogs with an unfavorable response to treatment that were predicted to be treatment failures) and specificity (percentage of dogs with a favorable response to treatment that were predicted to do so) for different cut-off values of the CT score (32). Special attention was focused on the results obtained with a CT score < 8, which was defined in a previous study as a cut-off value (8).

The dog's breed, sex, and age, as well as its CT score and number of treatments, are reported in Table 1. From the 36 dogs, there were 20 dogs that were cured after 1 treatment (success group) and 16 dogs that were cured after more than 1 treatment or were considered treatment failures (failure group). There was no statistically significant relationship between CT scores and probability of treatment failure, either for total score (P = 0.12) or for the score of the 4 deepest locations (P = 0.06) (Figure 2). The failure probability increased slightly with the score of the 4 deepest locations as compared with the total score. Dogs with a CT score < 8 were all successes, and it is mainly due to these dogs that there was a positive, although nonsignificant, relationship between the CT score and the failure probability.

The ROC curve demonstrates that high sensitivity and specificity cannot be obtained at the same time, whatever the cut-off value chosen. When using a cut-off value of < 8, the sensitivity is equal to 100% (exact 95% CI: 79% to 100%), but the specificity is low and equal to 30% (exact 95% CI: 12% to 54%). To increase the specificity to 80%, the cut-off CT score value needs to be increased to 16, but then the sensitivity goes below 20%.

All dogs (6/6) that had a score < 8 were cured after 1 treatment. From the dogs with a score ≥ 8, clinical cure was obtained in 46.6% (14/30) after 1 treatment; in 53.4% (16/30), clinical cure required more than 1 treatment or was not obtained.

In the present study, the severity of the lesions as assessed by CT could not be related to treatment failure (more than 1 treatment). These results may be explained by the difficulty in relating the CT features of nasal aspergillosis with the pathophysiology of the disease. The CT score was based on the severity of the lesions, expressed mainly in terms of the amount of abnormal soft tissue and the degree of turbinate destruction. The amount of abnormal soft tissue in a diseased nasal cavity can be evaluated grossly by CT, but, in most cases, CT does not allow the nature of the tissue to be defined, and it is often even difficult to differentiate soft tissue from fluid (21,24). Use of contrast studies may permit differentiation between the mucosa and other soft tissue or fluid, or between necrotic and vascularized soft tissue. However, it has been demonstrated that attenuation measurements are susceptible to a variety of errors in a diseased nasal cavity, due mainly to the presence and sometimes mixing of many complex structures of different physical densities (21,24,33).

In this study, the postcontrast CT studies were not helpful in evaluating the extension of the lesions. On the one hand, a dilution with a large volume of debris, necrotic tissue, exudate, inflammatory tissue, or large granulomas in the nasal cavity or frontal sinus may complicate optimal diffusion of the antifungal medication (21). Moreover, this soft tissue may serve as a growth medium for the fungus (21). However, even in such cases, a single topical infusion may be successful. On the other hand, an empty nasal cavity does not guarantee that the treatment will be effective, as there may be squamous or osseous metaplasia of the mucosa that decreases diffusion of the antifungal medication or “mucosal stripping” with compromised local circulation and immunity that will prevent healing (34,35).

The impact of the degree of turbinate destruction on the outcome is also difficult to predict. A severe turbinate destruction, emptying the nasal cavity, will permit better diffusion of the antifungal medication (21). However, a too severe turbinate destruction may diminish the local immunity and prevent healing (34).

Other CT features that were used to evaluate the severity of the disease in our classification system were invasion of periorbital structures, cribriform plate destruction, and frontal bone changes. Fungal infection of periorbital structures may be a cause of treatment failure, and it has been suggested that the topical therapy should be combined with a systemic antifungal medication in these dogs (7,36). Periorbital involvement was present in 2 dogs in our study. Both dogs were treated successfully by a topical antifungal medication after 1 and 3 treatments, respectively. The effect of the leakage of an antifungal medication into the central nervous system in case of cribriform plate destruction has not been studied in dogs with nasal aspergillosis (21). In the present study, 1 dog that showed localized cribriform plate destruction was clinically cured after 2 topical infusions of 1% enilconazole without clinical evidence of neurological signs. Bony changes are not considered to influence therapeutic success.

The failure to relate the severity of the lesions on CT to treatment failure may also be explained by the inability of CT to give information about potentially important predictors of therapeutic success, such as the environmental status, a bacterial or fungal secondary infection, an impaired immune function, or the resistance of the fungus to the antifungal medication (7,34,36).

The results of the present study are in contradiction with those obtained by Mathews et al (8). In the present study, no significant relationship could be found between CT score and treatment failure (more than 1 treatment needed), while in the previous study, a cut-off point of 8 allowed good classification with respect to sensitivity and specificity. With a cut-off point of 8, a high sensitivity (100%) was obtained in the present study compared with the Mathews study (71% and 78%) (8). However, the specificity (30%) was very low compared with that obtained by Mathews et al (79% and 93%) (8). Thus, using a cut-off point < 8, only 6 of the 20 dogs that needed only 1 treatment were predicted to do so. On the basis of the present study, it appears that the choice of a cut-off < 8 optimizes the sensitivity, while neglecting the specificity. This means that the owner of a dog with a score < 8 can be told that his dog will probably need only 1 treatment. For dogs with a score ≥ 8, no information can be given to the owner, as almost half the dogs (46%) needed only 1 treatment.

The difference in results between the 2 studies could be attributed to differences in the use of the scoring system, differences in the treatment methods, the results, or both. The CT criteria used to score the nasal cavities in each region from 0 to 3, based on turbinate atrophy and fluid accumulation, are not objectively measurable, thus preventing an excellent agreement between reviewers from being obtained, as was also observed in a previous study on CT of chronic nasal disease (24). In the study of Mathews et al (8), significant differences between reviewers were found in 3 of the 8 anatomic regions, but these differences did not have an impact on the cut-off point. Thus, it cannot be excluded that the observer in our study assigned higher scores overall. However, even the use of other cut-off points did not permit good sensitivity and specificity to be obtained at the same time in our study.

Differences between the treatment methods include the choice of the antifungal (enilconazole instead of clotrimazole), intranasal debridement prior to treatment, and placement of catheters in the frontal sinuses. According to the literature, the results obtained with enilconazole are comparable with those obtained with clotrimazole (8,9). A large amount of soft tissue is present in the nasal passages, in approximately 50% of the dogs with nasal aspergillosis, and around the sinonasal ostium, in approximately 70% (24). The efficacy of an endoscopic curretage could not be evaluated objectively in the present study, as the CT examinations were performed only before rhinoscopic debridment of the nasal cavity and frontal sinuses. However, when using enilconazole that is active in its vapor phase for distances up to 10 mm, the importance of endoscopic debridement and placement of drains in the frontal sinus may contribute to a better distribution of the drug, particularly in the mucosa (37). Endoscopic placement of catheters in the frontal sinus has been associated with an increased success rate (12). However, the treatment results in our study are comparable with those obtained by Mathews et al (8). Consequently, the treatment method cannot be responsible for the difference in results observed between the 2 studies.

A limitation of the present study is that only 6 dogs had a score < 8. In all these dogs, this score was associated with lesions restricted to the rostral half of the nasal cavity, unilaterally in 3 dogs and bilaterally in 3 dogs, while no dog had a frontal sinus infection without a concurrent abnormality in the nasal cavity. Lesions restricted to the nasal passages are encountered in approximately 25% of the dogs with nasal aspergillosis (24). All these dogs were correctly predicted to respond after 1 treatment. Dogs with a score < 8, therefore, seem to be good candidates to respond after 1 treatment, but this has to be confirmed on a larger number of dogs with a low CT score. Except for the dogs with a score < 8, use of a larger study will probably not modify the conclusions. It cannot be excluded that a statistically significant relationship between CT score and the probability of treatment failure may be found. However, it probably would have little clinical relevance. Firstly, it would only be a weak relationship mainly due to the dogs with a CT score lower than 8 that were all treatment successes. Secondly, for the current data, the specificity was only equal to 33%. This value might change slightly but not dramatically with a larger sample size due to random variation, but the overall conclusions would not actually change.

The results of this study showed no significant relationship between severity of the lesions, as assessed by CT, and results of therapy in 36 dogs treated with a noninvasive intranasal infusion of 1% enilconazole. Potential reasons are the difficulty in relating the CT features of nasal aspergillosis to the pathophysiology of the disease, and the inability of CT to give information about some important predictors of treatment outcome. Consequently, the view that it is possible on the basis of CT only, to predict treatment outcome, appears to be too simplistic.

Address all correspondence and reprint requests to Dr. J.H. Saunders