Recurrence and survival in dogs with excised colorectal polyps: A retrospective study of 58 cases

2.1 Cases

We reviewed the medical records from 7 veterinary centers (Aquivet veterinary hospital [Eysines, France], Canis veterinary hospital [Palma de Mallorca, Spain], Frégis veterinary hospital [Arcueil, France], Oniris veterinary teaching hospital [Nantes, France], OnlyVet veterinary hospital [Saint-Priest, France], and VetAgroSup veterinary teaching hospital [Marcy l'Etoile, France]) for dogs diagnosed with colorectal polyps. Dogs were included if a macroscopically visible colic or rectal mucosal protrusion was identified, excised and submitted to histological examination. When available, histological preparations were reevaluated by 2 boarded certified pathologists. Preparations could be unavailable because of property rights or because they had been discarded for storage reasons. Cases were included between September 1, 2006, and December 31, 2020.

2.2 Data collection

An online questionnaire (Google Form, Googleplex, Mountain View, United States of America) for each included case was completed by an internal medicine board specialist. The collected information included details about age at diagnosis, sex, reproductive status, breed, size (small dogs were defined by a body weight of less than 15 kg and large dogs by a body weight of 15 kg or more), clinical signs at presentation and the duration before polyp removal, physical examination findings, method of diagnosis, number of polyps, macroscopic description of the polyp (diameter, pedunculated versus sessile, and smooth versus irregular), distance of the polyp to the anus, method of removal (pull-out method, endoscopic diathermy, or colectomy) and associated complications, imaging modalities, duration of follow-up, date to first recurrence of polyp, and date and cause of death if applicable.

The overall number of dogs and the number of dogs for each breed that attended the correspondent hospital in the same time period were also recorded at 4 French centers (Aquivet veterinary hospital, Oniris veterinary teaching hospital, OnlyVet veterinary hospital and VetAgroSup veterinary teaching hospital) to assess the overall prevalence of colorectal polyps and association with breeds. The 4 centers contributed by providing 45 cases out of the 58 included (77%). The software of the remaining 3 centers could not give these numbers.

2.3 Histological analysis

The specimens were processed according to standard procedures and sections stained with hematoxylin-eosin-safran (HES). All slides were reviewed by 2 board-certified pathologists (T. Larcher and E. Brisebard) in a blinded fashion according to the RVC. A descriptive sheet (Microsoft Excel 365, Microsoft Corporation, Redmond, United States of America) to collect histopathological data was designed based on a previous publication by Schlemper (Table 1) and used to assign the diagnosis to a category according to the RVC (Table 2).²⁵ In case of disagreement between the 2 pathologists, specimens were re-evaluated to reach a final consensual diagnosis used to identify factors associated with the recurrence of colorectal polyps and survival. Margins were reported to be complete or incomplete, and submucosa was described as present or absent of each sample.

2.4 Statistical analysis

The normality of quantitative variables was assessed using Shapiro-Wilk testing. Normally distributed variables are presented as mean and SD; otherwise, results are presented as median and interquartile range. Categorical variables are presented as percentages. The odds ratios were used to quantify the association between each breed and presence (versus absence) of colorectal polyps. A Fisher's exact test with a Bonferroni-Dunn correction for multiple comparison was used to statistical significance.

We allocated endoscopic diathermy and pull-out methods for polyp removal into the same group mucosectomy.

Because we did not have margins available in all cases, we considered categories 4.2 and above of the RVC (noninvasive carcinoma, suspicious for invasive carcinoma, intramucosal carcinoma, or submucosal invasion of neoplasia) as worse diagnosis likely to benefit from colectomy and therefore considered separately from categories 4.1 and below in statistical analysis for the associations with time to recurrence of polyps or death.

T₀ was the time of first diagnosis of colorectal polyps; T_recurrence was the time of the first recurrence, if applicable; and T_death was the time of (all-cause) death, if applicable. The remission time was the time between T_recurrence and T₀, and the survival time was the time between T_death and T₀. In the survival analysis where the outcome was recurrence of polyps, recurrence-free dogs were censored at the time of the last follow-up appointment or at T_death if they died before the study endpoint (December 31, 2020). In the survival analysis where the outcome was all-cause death, surviving dogs were censored at the time of the last follow-up appointment.

The following explanatory variables were considered for the 2 survival analyses: age at T₀, sex, reproductive status, size of the dog (small versus large dogs), breed (West Highland White Terrier versus non-West Highland White Terrier), basis of the polyp (pedunculated or sessile), largest diameter of the polyp, method of polyp removal (mucosectomy versus colectomy), the histological diagnosis (category 4.2 or above versus category 4.1 or below of the RVC) and the completeness of the margins (complete versus incomplete). We tested these variables for their association with time to recurrence or time to all-cause death using Kaplan-Meier curves and univariable Cox proportional hazard regression models. The variables were included into a multivariate Cox proportional model if the P-value in univariate analysis was ≤.2 by using a backward stepwise selection procedure based on the Akaike information criterion. Interaction terms between all the variables were tested, and they were included in the model if they were found to be significant. The assumption of proportionality of hazards was investigated using the scaled Schoenfeld residuals method. The following variables were not considered for survival analysis because they could participate to the outcome by influencing other already considered variables: duration before polyp removal, distance to the anus. Significant complication associated with the method of polyp removal was not reported and could not be used as a potential prognostic factor. Adjunctive medical treatments were not included either in this analysis because of their high diversity in terms of nature of drugs/diets, duration and combination of treatments.

Statistical analyses were performed using R (R Core Team [2021]. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/). P-values <.05 indicated statistically significant differences.

3.1 Epidemiological findings

The mean age of the dogs was 7.8 years (3.3). The reproductive status was known in 56 dogs. The group included 27 intact males (48%), 13 spayed females (23%), 12 castrated males (21%), and 4 intact females (7%). They were composed of 34/58 small dogs (59%) and 24 large dogs (41%). The odds ratio results for the identification of breeds associated with the presence of colorectal polyps are summarized in Table 3. After Bonferroni-Dunn correction, West Highland White Terriers were significantly overrepresented among dogs with colorectal polyps compared to dogs without colorectal polyps (OR: 20.0; 95% confidence interval [CI]: 7.5-52.1; corrected P < .001).

Twelve dogs suffered from comorbidities: leishmaniasis was reported in 3/58 dogs (5%), atopy and myxomatous mitral valve disease in 2/58 (3%) dogs each, steroid-responsive chronic enteropathy, urethral incompetence, chronic kidney disease, exocrine pancreatic insufficiency, and recurrent anal sacs impaction in 1/58 (2%) dogs each.

3.2 Clinical findings

Owners' concerns and clinical findings at physical examination are summarized in Figure 1. Hematochezia was described in 49/58 dogs (84%), and polyps were detectable (visible or palpated) at physical examination in 46/58 dogs (79%).

3.3 Imaging findings

Only 46/58 (79%) files contained information on the possible performance of imaging studies. Among these, imaging investigation was performed in 34/46 dogs (74%) and included abdominal ultrasonography, thoracic radiographs and abdominal and thoracic computed tomography in 33/34 (97%), 9/34 (26%) and 1/34 dogs (2.9%), respectively.

On the single computed tomography study that was performed, no sign for submucosal invasion nor local, regional or distant metastasis was found.

Polyp was visible on ultrasound examination in only 1/34 dogs (5.9%). Invasion of the submucosa was observed in 1 (2.9%) which proved to be a carcinoma on histological analysis.

A primary, single lung mass was found in 1 dog (2.9%) with benign polyp. A diagnosis of a lung carcinoma was finally reached based on radiographic aspect (lobar opacification of the left caudal lung lobe), cytological examination (consistent with anaplastic carcinoma), and follow-up (died 910 days after polyp removal).

Equivocal images for distant metastasis were found in 4/34 dogs (12%) using abdominal ultrasonography. One dog had colic lymphadenopathy (hypoechoic and enlarged lymph nodes). The dog was diagnosed with carcinoma invading the submucosa on histopathology and died of pulmonary metastasis 1 year after polyp removal. A second dog had nonspecific hypoechoic liver nodules and was diagnosed with carcinoma invading the submucosa on histopathology. It died because of mandibular osteosarcoma 990 days after polyp removal. The third dog had hypoechoic and enlarged caudal colic, ileocolic and mesenteric lymph nodes. Polyp was not available for histological reassessment but primary diagnosis was adenoma. The dog was not followed up. The fourth dog had hypoechoic splenic nodules; polyp was not available for histological reassessment but primary diagnosis was adenocarcinoma; the rectal polyp recurred 90 days after surgical removal and the dog was then lost at follow-up.

3.4 Polyp description and colonoscopic findings

A colonoscopy was performed in 47/58 dogs (81%) and a single polyp was found in 44/47 dogs (94%) and 2 polyps in 3/47 dogs (6%). Macroscopic evidence of associated colitis (mucosal thickening, irregularity, hyperemia, ulceration, or a combination of these findings) was documented in 21/47 dogs (45%).

The median diameter of polyps was 10 mm (6.0-20.0 mm), and the median distance of the polyps to the anus was 3.0 cm (1.0-5.0 cm). The basis of the polyps was pedunculated in 29/49 dogs (59%) and sessile in 20/49 dogs (41%). The macroscopic aspect of the polyps was smooth in 48/58 dogs (83%) and irregular in 10/58 dogs (17%). Similar findings (basis and macroscopic appearance) were reported in dogs with 2 polyps.

3.5 Method of polyp removal and adjunctive medical therapy

The polyps were removed using the pull-out method in 45/58 dogs (78%), endoscopic diathermy resection in 7/58 dogs (12%), and colectomy in 6/58 dogs (10%).

Adjunctive medical treatments included metronidazole (21/58 dogs, duration range: 5-60 days, dose range: 12.5-20 mg/kg BID), amoxicillin/clavulanate (18/58 dogs, duration range: 5-21 days, dose range: 12.5-20 mg/kg BID), lactulose (8/58 dogs, duration range: 2-30 days, dose range: 0.2 mL/kg BID-0.3 mL/kg TID), tramadol (7/58 dogs, duration range: 5-10 days, dose range: 3-4 mg/kg BID), prednisolone (5/58 dogs, duration range: 10-20 days, initial dose range: 0.5-1 mg/kg/day, then tapered), highly-digestible diet (5/58 dogs, duration range: 7-30 days), firocoxib (4/58 dogs, duration range: 10-60 days, dose: 5 mg/kg SID), meloxicam (2/58 dogs, duration: 5 days, dose: 0.1 mg/kg SID), cefalexin (2/58 dogs, duration range: 10-15 days, dose: 15 mg/kg BID), robenacoxib (1/58 dogs, duration: 3 days, dose: 1 mg/kg SID), piroxicam (1/58 dogs, duration: 30 days, dose: 0.3 mg/kg SID), cimicoxib (1/58 dogs, duration: 21 days, dose: 2.5 mg/kg SID), mesalamine (1/58 dogs, duration: 10 days, dose: 10 mg/kg BID), tylosin (1/58 dogs, duration: 10 days, dose: 10 mg/kg TID) and/or hydrolyzed diet (1/58 dogs, duration: 60 days, dose).

3.6 Histologic findings

Margins were available for assessment in 25/32 samples (78%). Submucosa was present in 25/32 (78%) samples. Based on the final histological report, polyps were classified as mucosal low-grade neoplasia (category 3) in 3/32 samples (9%), high-grade adenoma/dysplasia (category 4-1) in 2/32 samples (6%), noninvasive carcinoma (category 4-2) in 4/32 samples (13%), carcinoma with suspected invasion (category 4-3) in 4/32 samples (13%), intramucosal carcinoma (category 4-4) in 12/32 samples (38%), and neoplasia with submucosal invasion (category 5) in 7/32 samples (22%). For the dog with samples from time of first presentation and from the recurrence, polyps were classified as category 5 and 4-4, respectively.

3.7 Outcome

Median follow-up duration was 142.5 (30-530) days.

3.7.2 Survival

Fourteen dogs were dead at the time of data collection. The estimated median survival time was 1640 days (Figure 2). The cause of death was related to the polyp in 4/14 dogs (29%), with suture dehiscence after colectomy at recurrence in 1 dog and distant metastasis in 3 dogs. Four out of 14 dogs (29%) died of unknown cause while in 6/14 dogs (43%) the cause of death was unrelated to the polyp (1 dog in each of the following categories: congestive heart failure, car accident, presumed abdominal mesothelioma, mandibular osteosarcoma, hemangiosarcoma, and status epilepticus).

The results from the Cox proportional hazard regression model for identifying risk factors of death are presented in Table 5. Because none of the dogs in the group of category 4.1 or less of the histological diagnosis variable and none of the dogs in the group of incomplete margins of the sample died, an error resulted when attempting the P-value calculation using the Wald test. Among the other variables, the method of polyp removal was the only 1 to be eligible for multivariate analysis using the Wald test. We therefore used Kaplan-Meier curves for comparison of the histological diagnosis (category 4.1 or less versus category 4.2 or above), the margins of the sample (complete versus incomplete) and the method of polyp removal (endoscopic diathermy or pull-out versus colectomy). Time to death was not significantly different between categories of histological diagnosis or between categories of completeness of margins of samples (histological diagnosis: P logrank = .2, median survival time in group category 4.2 or above: 990 days, 95% IC: 990-NA; margins: P logrank = .05; median survival time in group complete margins: 990 days, 95% IC: 990-NA). Dogs that experienced colectomy had a higher probability of death during follow-up than those that underwent endoscopic diathermy or pull-out (P = .05, Figure 3). Two dogs died in the colectomy group: the first dog died only 2 months after the polyp removal because of a car accident. The second dog died 1 year after polyp removal because of distant metastasis of invasive carcinoma.

TABLE 5. Result of Cox proportional hazard regression model to identify predictors of death.

Explanatory variable (reference)	Univariable model hazard ratio (95% CI)	Univariable model P-value	Multivariable model hazard ratio (95% CI)	Multivariable model P-value
Age (continuous)	1.1 (0.9-1.3)	.31
Sex (male)	1.6 (0.5-6.0)	.46
Reproductive status (neutered)	1.5 (0.5-4.5)	.53
West Highland White Terrier (yes)	0.7 (0.09-5.9)	.75
Basis of the polyp (sessile)	1.6 (0.4-6.1)	.51
Diameter of (largest) polyp (continuous)	1.02 (0.99-1.05)	.15
Method of polyp removal (pull-out or endoscopic diathermy)	0.19 (0.04-1.1)	.06	0.19 (0.04-1.1)	.06

• Note: Variables with P ≤ .2 in univariable model were included in initial multivariate model. An error occurred in calculation of the hazard ratio and associated P-value using the Wald test for the histological diagnosis and completeness of margins as potential explanatory variables because no event occurred in the group of category 4.2 or lower and in the group of incomplete margins.