Criteria for Selection of Cases

Diagnostic reports of lingual biopsy specimens from dogs submitted to the Colorado State University Veterinary Diagnostic Laboratory from January 1995 to October 2004 were collected and reviewed.

Procedures

Data obtained from diagnostic reports included age, sex, and breed of dog and the histologic diagnosis for each lingual biopsy specimen. In cases in which the diagnosis was ambiguous, slides were reviewed by a single pathologist (MMD).

Statistical analysis—Descriptive and comparative statistics were computed by use of commercially available statistical software.^a,b Biopsy results for each dog were allocated to 1 of 4 categories on the basis of morphologic diagnosis (glossitis, neoplasia, other, or normal). Neoplastic lesions were dichotomized according to whether lesions were malignant or benign. For tumors in which malignancy was not clearly implied by the diagnosis (melanoma, mast cell tumor, and plasma cell tumor), subclassification (malignant or benign) was based on reported biological behavior for the respective neoplasm type. When histologic characteristics of lingual biopsy specimens indicated more than 1 diagnosis, the specimen was assigned a diagnosis corresponding to the predominant histologic findings or presumptive primary lesion.

Specimens were stratified by category of diagnosis for initial data exploration and subsequent statistical analysis. Sex was converted to a dichotomous variable irrespective of neuter status. Breeds were recorded as defined by the person that submitted the biopsy specimen. For mixed-breed dogs, the breed considered most representative was used. In cases in which the principal breed characteristics could not be identified, dogs were classified as mongrels. Dogs were categorized as large- or small-breed dogs according to breed standards for weight provided by the American Kennel Club and a cutoff value of 15 kg (33 lb); dogs < 15 kg were considered as small-breed dogs. Mongrel dogs were excluded from this size categorization.

The association between breed and lesion was explored for breeds that have been previously reported to be overrepresented for specific lesions and for breeds that were observed in the study population to be frequently affected with certain lesions. Relative frequencies were compared between the category of diagnosis and categoric variables of breed, size, and sex by use of ORs and 95% CIs. Confidence intervals were considered significant when they did not include the value 1. χ² Tests were used when there were > 2 categories. The mean age of dogs was compared among groups of diagnoses by use of ANOVA and t tests. Values of P < 0.05 were considered significant. The Anderson-Darling normality test was used to verify a normal distribution for age.

Results

During the study, submissions for 1,196 lingual biopsy specimens were identified. These represented 0.8% of all biopsy specimens in dogs received by the diagnostic laboratory during the study period. Neoplasia was the most common diagnosis (n = 646 [54%]), followed by glossitis (397 [33%]) and other (146 [12%]). Tissues from 7 (0.6%) specimens were considered as normal and were excluded from further analysis. Specimens from 115 different breeds were submitted. Thirty-seven percent of biopsy specimens submitted were from Chow Chows, Cocker Spaniels, Golden Retrievers, Labrador Retrievers, and Poodles. Age was normally distributed among all dogs in the study population (P < 0.005). The mean age of dogs was significantly (P < 0.001) different among categories of disease.

Neoplasia was the most frequent morphologic diagnosis, representing 54% of lingual biopsy specimens (Table 1). Twenty-six types of neoplasia were diagnosed; 16 tumors were malignant and 10 were benign. Mean ± SD age of dogs in which lingual neoplasia was diagnosed was 10.0 ± 3.3 years (range, 0.5 to 18 years). Mean age of dogs was significantly (P < 0.001) different among various types of neoplastic lesions. Large-breed dogs were 2.31 times as likely to have a diagnosis of lingual neoplasia, compared with small-breed dogs (95% CI, 1.76 to 3.02).

Of the 646 lingual biopsy specimens in which neoplasia was detected, 64% were malignant. Mean ± SD age of dogs with malignant tumors was 10.5 ± 2.7 years, which was significantly (P < 0.001) higher than dogs with benign tumors (8.9 ± 3.8 years). No significant (OR, 0.92; 95% CI, 0.67 to 1.30) difference in the proportion of malignant tumors was detected between male and female dogs. Within the population of dogs with neoplastic lesions, large-breed dogs were significantly (OR, 3.68; 95% CI, 2.42 to 5.63) more likely to have malignant lingual neoplasia than small-breed dogs. The most frequent malignant neoplasms included melanomas, SCCs, hemangiosarcomas, and fibrosarcomas.

Melanoma represented 23% of all lingual neoplasms. Mean ± SD age of dogs with lingual melanomas was 11.1 ± 2.3 years (range, 5 to 17 years), which was significantly (P < 0.001) greater than the mean age of dogs with all other types of neoplasms when pooled. There was no significant (OR, 1.25; 95% CI, 0.86 to 1.83) difference in the frequency of lingual melanomas between male and female dogs. The odds of lingual melanoma being diagnosed in Chow Chows and Chinese Shar-Peis were 40.37 (95% CI, 19.95 to 81.70) and 24.43 (95% CI, 2.98 to 200.18) times those for other breeds of dogs with lingual neoplasia, respectively. The odds of lingual melanoma being diagnosed in large-breed dogs were significantly (OR, 6.88; 95% CI, 2.95 to 15.05) greater than small-breed dogs.

Squamous cell carcinoma was the second most common lingual neoplasm, representing 18% of all lingual neoplasms. Mean ± SD age of dogs with SCC was 10.7 ± 2.6 years (range, 5 to 17 years), which was significantly (P = 0.006) greater than the mean age of dogs with other types of lingual neoplasms when pooled. The odds of having lingual SCC were significantly (OR, 2.26; 95% CI, 1.45 to 3.54) greater in female than male dogs. Dogs with lingual SCC were significantly more likely to be Poodles (OR, 4.61; 95% CI, 1.91 to 11.13), Labrador Retrievers (OR, 2.41; 95% CI, 1.46 to 3.98), or Samoyeds (OR, 24.63; 95% CI, 2.85 to 212.92), compared with other breeds with lingual neoplasia. The odds of having lingual SCC were not significantly (OR, 1.21; 95% CI, 0.69 to 2.11) different in large-breed dogs, compared with small-breed dogs.

Mean ± SD age of dogs with lingual hemangiosarcoma was 10.8 ± 2.8 years (range, 4 to 17 years). Border Collies were 11.8 (95% CI, 3.18 to 43.81) times as likely to be diagnosed with lingual hemangiosarcoma as other breeds with lingual neoplasia. There was no significant difference in the frequency of male versus female or small-versus large-breed dogs with lingual hemangiosarcoma.

Mean ± SD age of dogs with lingual fibrosarcoma was 9.4 ± 3.4 years (range, 1 to 15 years). There was no significant (OR, 1.24; 95% CI, 0.63 to 2.45) difference in the frequency of lingual fibrosarcoma between male and female dogs. The odds of having lingual fibrosarcoma were significantly (OR, 3.64; 95% CI, 1.56 to 8.47) greater for Golden Retrievers than other breeds of dogs with lingual neoplasia. There was no significant (OR, 0.97; 95% CI, 0.39 to 2.44) difference in the frequency of fibrosarcoma between large- and small-breed dogs.

The 3 most common benign lingual neoplasms were squamous papilloma, PCT, and granular cell tumor. Thirty-two (46%) squamous papillomas were determined to have microscopic features consistent with a viral etiology. Mean ± SD age of dogs with squamous papillomas was 6.5 ± 4.3 years (range, 0.6 to 18 years), which was significantly (P < 0.001) lower than ages of dogs with other types of lingual neoplasms when pooled.

Mean ± SD age of dogs with lingual PCT was 9.6 ± 3.3 years (range, 7 to 15 years). Dogs with lingual PCTs were 10.80 (95% CI, 4.81 to 24.26) times as likely to be Cocker Spaniels as other breeds with lingual neoplasia. Small-breed dogs were significantly (OR, 4.02; 95% CI, 2.30 to 7.04) more likely to have lingual PCT than large-breed dogs.

Mean ± SD age of dogs with lingual granular cell tumors was 10.4 ± 2.7 years (range, 4 to 17 years). There was no significant (OR, 1.23; 95% CI, 0.65 to 2.12) difference in the frequency of granular cell tumors between male and female dogs. No specific breed patterns were observed; however, small-breed dogs were 4.75 (95% CI, 2.36 to 7.04) times as likely to be affected as large-breed dogs.

There were 397 cases of glossitis, representing 33% of lingual lesions diagnosed during the study period (Table 1). In most cases, the cause of glossitis was not evident. Inflammatory lesions occasionally centered around intralesional foreign material, including plant (n = 20), hair (10), suture (4), fragments of bone (3), or unidentifiable material (3). Intralesional bacteria were identified in 26 specimens; 22 of these also had accompanying neutrophilic or ulcerative inflammation. Intralesional spiralshaped bacteria, fungal organisms, or viral inclusions were not identified in any lesion. Concurrent vasculitis was described in only 3 cases of ulcerative glossitis.

Mean ± age of dogs with inflammatory lingual lesions was 8.8 ± 3.6 years (range, 0.5 to 18 years). The mean age of dogs varied significantly (P = 0.034) by inflammatory lesion. There was no significant (P = 0.61) difference in the frequency of inflammatory lingual lesions between sexes. A diagnosis of inflammatory lingual lesion was 2.41 (95% CI, 1.83 to 3.16) times as likely in small-breed dogs as in large-breed dogs. Of small-breed dogs, the odds of having inflammatory lingual lesions were significantly greater in Shih Tzus (OR, 3.95; 95% CI, 1.47 to 10.60), Dachshunds (OR, 4.46; 95% CI, 2.01 to 9.89), and Malteses (OR, 6.72; 95% CI, 2.68 to 16.87), compared with all dogs in the study population. Ten of 19 cases of glossitis diagnosed in Maltese dogs were classified as ulcerative.

Noninflammatory, nonneoplastic lesions accounted for 12% of lingual lesions in the study. The mean age of dogs within this classification was significantly (P < 0.0001) different among lesions. Dogs with calcinosis circumscripta, which was the most frequently observed lesion within the “other” classification (n = 51), had low mean ± SD age (1.6 ± 1.7 years; range, 0.25 to 7 years), with 82% of affected dogs being < 2 years old. There was a significant (OR, 1.93; 95% CI, 1.09 to 3.44) difference in the frequency of calcinosis circumscripta between male and female dogs. Rottweilers (OR, 8.37; 95% CI, 3.16 to 22.21) and German Shepherd Dogs (OR, 4.08; 95% CI, 1.90 to 8.80) were significantly more likely to have calcinosis circumscripta than other breeds of dogs in the study population. The odds of a large-breed dog having calcinosis circumscripta were 2.55 (95% CI, 1.17 to 5.72) times those in small-breed dogs.

Discussion

In the study reported here, neoplasia was the most common type of lingual lesion in dogs. Among companion animal species, oral cancer most frequently affects dogs.³ Breeds reportedly at increased risk for developing oral neoplasia include the German Shepherd Dog, Poodle, German Shorthaired Pointer, Weimaraner, Golden Retriever, Boxer, Cocker Spaniel, and mixedbreed dogs.^3,5 Reportedly, the most common neoplasms of the oral cavity in dogs are SCC, melanoma, and fibrosarcoma.^1–4 The same tumors, as well as granular cell tumor, are frequently reported to affect the tongue in dogs.^4,6–8 Our findings are in agreement with the frequency of oral neoplasia reported in those studies; however, results of the study reported here indicated that, in the tongue, PCT and hemangiosarcoma are also common diagnoses. Lingual mast cell tumors, hemangiosarcomas, lipomas, leiomyosarcomas, and hemangiomas have been reported.^6,9 Although rhabdomyomas and rhabdomyosarcomas reportedly affect the tongue in dogs,^10–12 neoplasms of skeletal muscle origin were not detected in our study population. This observation may emphasize the rare occurrence of such tumors, the difficulty in collecting a biopsy specimen from neoplasms that emerge deep within lingual musculature, or obscurity in accurate histologic differentiation of these tumors from granular cell tumors based on similar cytomorphologic features. Granular cell tumor may represent a nonspecific diagnosis that includes neoplasms of different cellular origins with similar histologic appearances, possibly including tumors of skeletal muscle origin.^10,13,14

Oral neoplasms, especially melanomas, are reportedly more frequent in males than females⁴; however, in the study reported here, a significant difference in the frequency of male or female dogs with lingual neoplasia was not detected. Results of our study indicated that lingual masses in large-breed or aged (mean, 10 years old) dogs were more likely to be malignant than benign or inflammatory. Dogs in which malignant tumors were diagnosed were significantly older than those with benign tumors and were more likely to be large-breed dogs; however, there was no sex predilection.

In contrast to results of other studies^6,7 in which SCC was the most frequent lingual tumor diagnosed in dogs, melanoma was the most frequent lingual tumor diagnosed in our case series. Fifty-six percent of melanocytic tumors in dogs reportedly involve the oral cavity.¹⁵ Melanoma is the most common malignant neoplasm of the oral cavity of dogs^4,5; however, only 5% to 8% of oral melanomas involve the tongue.^16,17 Results of other studies^2–4 indicate that male dogs have an increased risk of developing oral melanoma; however, a sex predilection for development of lingual melanomas was not detected in our study. Cocker Spaniels, Chow Chows, Golden Retrievers, Miniature Poodles, and Gordon Setters reportedly have a higher risk for developing oral melanomas.^1,2,16 Many of those breeds have pigmented oral mucosa. In the study reported here, lingual melanoma was more common in large-breed dogs, and the odds of lingual melanoma being diagnosed in Chow Chows and Chinese Shar-Peis were markedly increased. This finding is in contrast to findings of 1 study⁴ indicating that small-breed dogs are at increased risk for developing oral melanomas. Oral melanomas are invariably considered malignant. The prognosis for dogs with oral melanomas is reportedly not related to location within the oral cavity and histologic features.^17,18 In our study, histologic criteria for malignancy in 18 lingual melanomas were described as being low to nonexistent. Further study is needed to determine whether well-differentiated lingual melanomas such as these have the same biological behavior as other oral melanomas.

Squamous cell carcinoma was the second most common neoplastic lesion of the tongue detected in the case series reported here. Fifteen percent of nontonsillar oral SSC is reportedly localized to the tongue.¹⁹ Results of several studies^6,19,20 evaluating oral or lingual SCC in dogs did not indicate breed or sex predilections; however, in our study, lingual SCC was significantly more common in female dogs, Poodles, Labrador Retrievers, and Samoyeds.

Other malignant lingual tumors detected in our study included hemangiosarcoma, fibrosarcoma, mast cell tumor, and lymphoma. The frequency of lingual hemangiosarcoma (9% of malignant lingual tumors) was not expected because hemangiosarcoma has been reported to account for only 1% of oral malignancies in dogs.³ In dogs, only 2% of hemangiosarcomas are reported to involve the oral cavity, and 1% of nonvisceral hemangiosarcomas involve the tongue.^21,22 In 1 retrospective study,²² breeds commonly affected with nonvisceral hemangiosarcoma included Golden Retrievers, German Shepherd Dogs, Italian Greyhounds, and Beagles. In contrast, only Border Collies had a significantly higher risk for developing lingual hemangiosarcoma in our study. Reportedly, Beagles develop lingual and concurrent cutaneous hemangiosarcomas.²³ Whether the lingual tumor was a primary mass or represented a component of metastatic or multicentric disease could not be determined in our study.

In the study reported here, Golden Retreivers were most likely to have lingual fibrosarcoma. Large-breed dogs have been reported to be overrepresented with oral fibrosarcoma⁴; however, a significant difference in the occurrence of lingual fibrosarcomas between largeand small-breed dogs was not detected in our study. Male dogs reportedly have an increased risk for developing oral fibrosarcomas³; however, a sex predilection was not detected in our case series.

Dogs with benign lingual neoplasms were younger than those with malignant lingual lesions; however, this finding may have been influenced by the young mean age of dogs with squamous papilloma. The most common benign lingual tumors included squamous papilloma, PCT, and granular cell tumor. Although the mouth is a common site for development of extramedullary PCT, the frequency with which oral PCTs are restricted to the tongue is not known.^24,25 Cocker Spaniels are frequently affected with mucocutaneous PCTs.²⁴ Results of the case series reported here indicated that small-breed dogs in general and Cocker Spaniels were more likely to have lingual PCTs. Reportedly, male dogs are more frequently affected with extramedullary PCTs²⁵; however, a sex predilection for development of lingual PCTs was not detected in our study. Small-breed dogs were more likely to have a diagnosis of lingual granular cell tumors, compared with large-breed dogs; however, no specific breed or sex predispositions were detected.

Glossitis represented the second most common lingual lesion, comprising 33% of lingual biopsy specimens. Infectious causes of glossitis in dogs include canine calicivirus, canine distemper virus, Pasteurella multocida, Candida spp, and various opportunistic bacteria; little is known about the frequency of their occurrence.²⁶ In the study reported here, the cause of glossitis was usually not evident histologically, regardless of lesion characteristics. Intralesional bacteria were reported in 7% of the cases of glossitis and usually involved ulceration or suppurative inflammation. Although pathogenic bacteria may induce glossitis, commensal bacteria often colonize oral lesions; therefore, the role of bacteria in the development of these lesions is not clear. Inflammation was centered around foreign material, including bone, hair, suture, and plant material in 10% of the cases of glossitis. Intralesional spiral-shaped bacteria, cytoplasmic inclusion bodies consistent with canine distemper virus, or fungal organisms were not detected in any cases of glossitis, which may imply that these etiologies infrequently affect the tongue or that recognition of these etiologies via standard histologic examination is not reliable. Glossitis may also be incited by trauma, bite wounds, migrating foreign bodies, electrical burns, and caustic agents,²⁶ and uremia, niacin deficiency, and diabetes mellitus may predispose dogs to lingual ulceration²⁷; however, these causes or predisposing conditions cannot be detected histologically.

Immune-mediated diseases including eosinophilic granuloma, pemphigus vulgaris, and discoid lupus erythematosus may also affect the tongue,^27–30 although these conditions appear to rarely affect the tongue in dogs. Eosinophilic granuloma, a lesion usually restricted to the tongue of dogs, represented 3% of the cases of glossitis. Lichenoid or interface inflammation, suggestive of an immune-mediated condition such as lupoid dermatoses, bullous pemphigoid, or pemphigus vulgaris, was detected in only 5 biopsy specimens.

The mean age of dogs with inflammatory lingual lesions was younger than dogs with lingual neoplasia. Glossitis was diagnosed in small-breed dogs more than twice as often as large-breed dogs. Small-breed dogs specifically identified at risk for development of glossitis included Shih Tzu, Maltese, and Dachsund. Most cases of glossitis in Maltese dogs were characterized as ulcerative. The Maltese dog has a propensity for developing ulcerative stomatitis, which has been described to involve the tongue and has been suggested to involve an abnormal immune response to dental calculus.³¹

Calcinosis circumscripta was the most common nonneoplastic, noninflammatory lingual lesion detected in the case series reported here. Most dogs with lingual calcinosis circumscripta were < 2 years old. German Shepherd Dogs and Labrador Retrievers were significantly more likely to have calcinosis circumscripta, compared with the remainder of the study population. This finding is consistent with results of another study³² indicating that young and large-breed dogs, especially German Shepherd Dogs, Rottweilers, and Labrador Retrievers, were overrepresented for development of calcinosis circumscripta. Calcinosis circumscripta was detected significantly more often in male dogs in our study, a finding not previously reported. Many of the remaining lesions classified as “other” were degenerative or associated with trauma or wound healing and included hematoma, fibrosis, necrosis, and granulation tissue.

Lingual lesions from dogs are an uncommon submission to diagnostic laboratories, as evidenced by the small proportion of lingual biopsy specimens submitted to the veterinary diagnostic laboratory. Findings described herein may be geographically biased and thus may not be representative of other regions. Because the breed distribution of dogs in the United States is poorly characterized, true breed predispositions to specific diseases are difficult to determine. Infrequent samples from certain breeds and diagnoses reduced the statistical power of our study and resulted in wide and non-significant CIs for clinically noteworthy associations. Further investigation into the association between certain lingual neoplasms and multicentric or metastatic disease, and the prognostic implications for various lingual lesions, is warranted.