Thyroidectomy in dogs with thyroid tumors: Survival analysis in 144 cases (1994‐2018)

Abstract Background Few studies have assessed predictors of outcome in dogs with thyroid tumors undergoing thyroidectomy. Objective To estimate the survival and identify prognostic factors in dogs with thyroid tumors treated by thyroidectomy. Animals A total of 144 client‐owned dogs with thyroid neoplasia that underwent thyroidectomy. Methods Retrospective study. Data for analysis included hospital attended and year of surgery, signalment, thyroxine concentration, thyroid tumor features (lobe involvement, size, invasiveness, histopathological type), thrombosis, metastasis, additional surgery and therapy, administration of adjuvant chemotherapy. The association of predictors with survival (time from surgery to death) were assessed by calculating cause‐specific hazard ratios (HRcs) and 95% confidence intervals (CI). Causes of death were classified as thyroid‐related or because of other cause. Results Overall median survival time was 802 days (CI95% = 723‐1015 days); 89 dogs (77.4%) survived >500 days. Metastases were identified at admission in 12 (8.3%) dogs and were associated with higher thyroid cancer‐related fatality (HR = 5.83, CI95% = 1.56‐21.78; P = .009). Thrombosis occurred in 40 dogs and was associated with increased risk of death because of other cause (HR = 2.73, CI95% = 1.18‐6.35; P = .019). Nonfollicular carcinoma (HR = 4.17, CI95% = 1.27‐13.69; P = .018) and administration of chemotherapy (HR = 3.45, CI95% = 1.35‐8.82; P = .01) were associated with higher risk of thyroid cancer‐related death. Conclusions and Clinical Importance Dogs with thyroid tumors undergoing thyroidectomy have a long life expectancy. Despite the rare presence of nonfollicular carcinoma and metastases, thyroidectomy should still be considered in some of these dogs.


| INTRODUCTION
Thyroid tumors in dogs have a prevalence of 1-4% among all neoplasms, and represent the most common tumor of the neuroendocrine system. [1][2][3][4][5] Older dogs are affected more frequently, 6,7 and the Beagle, Boxer and Golden Retriever breeds are overrepresented. 2,3,6,8 Unilateral thyroid lobe involvement is more common than bilateral, 6,9,10 and approximately 50-60% of affected dogs are euthyroid, 30-40% are hypothyroid, and 10-20% are hyperthyroid. 5 Carcinomas are the most frequent thyroid tumors in dogs and, among them, 70% arise from follicular cells (ie, follicular thyroid carcinomas) and 30% from parafollicular cells (ie, medullary thyroid carcinomas), both are invasive. Adenomas, on the other hand, are rare in dogs. [3][4][5]11,12 In dogs, follicular thyroid carcinomas are histologically classified as welldifferentiated (ie, follicular, compact, follicular-compact and papillary), poorly differentiated, undifferentiated, and carcinosarcoma (malignant mixed thyroid tumor). 12 At diagnosis, 16-38% of dogs with thyroid tumors have distant metastasis. 4,13 Thyroidectomy represents the treatment of choice for freely movable tumors and sometimes could be indicated also for nonresectable ones, along with external beam radiation or radionuclide therapy, chemotherapy or a combination of these therapies. 10,14,15 There are many purported factors potentially associated with less favorable outcome. Undifferentiated carcinomas have a less favorable outcome, 5,16,17 however there are conflicting reports in the literature regarding local invasiveness and outcome. 7,18,19 Tumor size was a negative prognostic factor that appeared also related to the presence of metastases, in an investigation with specimens obtained at necropsy, 20 but not in other reports with measurements performed at the time of thyroidectomy. 14,16,18 Thrombosis of the tumor and cervical vessels negatively affects the outcome. 7 Higher WHO staging and metastatic disease (local or distant) might not result in a poor outcome. 14,21 A median survival time (MST) of 500-1100 days is reported after thyroidectomy. 7,18,19,22,23 In particular, MST >960 days is documented both in dogs undergoing unilateral and bilateral thyroidectomy, 18,23 and 630 days in dogs with gross vascular invasion treated with unilateral or bilateral thyroidectomy. 19 Bilateral thyroidectomy negatively affects outcome and the MST of these dogs is 365 days. 14 Of dogs with functional tumors undergoing thyroidectomy the MST is 2190 days. 24 No benefit of adjuvant chemotherapy on MST is observed in dogs treated with thyroidectomy alone vs thyroidectomy combined with chemotherapy, irrespective of metastasis. 14 Administration of toceranib phosphate on macroscopic disease to affected dogs, for a minimum of 10 weeks is beneficial in 75% of cases (ie, complete response to stable disease). 25 Conversely, no benefit on MST is observed in dogs treated with thyroidectomy alone vs thyroidectomy combined with chemotherapy, irrespective of metastasis. 14 The aims of the present study were to retrospectively assess survival in dogs with thyroid tumors treated by thyroidectomy, and to identify prognostic factors using competing risk analysis.

| Experimental design
This is a multi-institutional retrospective case series study. For each dog included in the study, variables retrieved from medical records were collected and used to determine their influence on MST, defined as the time from thyroidectomy to death or administrative censoring (last visit). These variables were: the institution where thyroidectomy was performed and the year of surgery, the age (in years, at the time of surgery), gender, neutering status (ie, intact, spayed and castrated), breed, dog size (ie, small, <10 kg; medium, 10-25 kg; large, >25 kg), presurgical serum thyroxine (T4) concentration (ie, increased, normal or decreased), thyroid lobe involvement (ie, unilateral or bilateral), type of thyroidectomy (ie, unilateral or bilateral), additional surgeries performed (ie, cervical lymphadenectomy and muscular, internal jugular vein resection and parathyroid gland autotransplantation 26 ), tumor major axis size, histopathological tumor type, local invasiveness (determined by palpation characteristics of the ventral cervical area combined with computed tomography examination of the same area in some dogs and histopathological examination in others, or histopathological examination in others), thrombosis, metastatic disease at the time of surgery (determined by computer tomography and histopathological examination), administration of adjuvant chemotherapy, radiation therapy after thyroidectomy, levothyroxine and calcitriol supplementation following surgery. In dogs diagnosed with bilateral thyroid lobes tumor involvement, who undergone bilateral thyroidectomy, at least 1 of the parathyroid glands was transplanted. The parathyroid gland was freed from his thyroid capsule, inserted in a muscular pocket, which allowed revascularization and regaining functionality after few weeks. The effect of each aforementioned variable on survival was studied in relation to the cause of death which was eithers thyroid cancer or other cause.

| Statistical analysis
Competing risk analysis was used in this study. This was because the classical Kaplan-Meier analysis, which censors the observed survival time when a competing event intervenes, violates the assumption of independence between time to event and censoring, yielding an underestimate of the survival probability 27 and overestimating the probability of death. Two competing events were considered, that is, death because of thyroid cancer or death because of another cause. The (cause specific) cumulative incidence function (CIF) estimates the marginal probability for each competing event (ie, the probability of experiencing the event of interest before time t and before the occurrence of a different type of event) as a function of time and was used to estimate the likelihood of the occurrence of each of the 2 causes of death. Gray's test was used to compare cause-specific CIFs in dogs classified according to each variable gathered at surgery. 28 The extended Cox proportional hazards model was used to estimate the cause-specific hazard ratio (HR) for each variable. 29,30 The model accounted for the multicentric nature of the study by allowing different baseline hazards for the 4 clinics. The proportional hazards assumption was evaluated by studying the Schoenfeld residuals. 31 Linearity was assumed based on the analysis of the martingale residuals. After an initial assessment, the effect of neutering was not considered.
In order to preserve the sample size, the random survival forest algorithm for imputing missing data was used. 32 Bivariate associations between variables were investigated by Fisher's exact test and t-test.
Logistic regression was used to model the probability of concurrent metastasis and thrombosis at diagnosis, and odds ratios (ORs) were calculated. All analyses were performed using R with mstate, cmprsk, randomForestSRC and survival packages. 31 Presurgical serum T4 concentration was measured in 103 (71.5%) dogs; 9 (8.7%) cases had abnormally high T4, all with unilateral involvement, 7 (6.8%) cases had abnormally low T4, including 6 cases with unilateral and 1 with bilateral involvement, and 87 (84.5%) cases were euthyroid.
The year of surgery was available for all dogs; surgery was performed from July 18, 1994 to June 20, 2018 (Table 1). Thyroidectomy alone was performed in 109 of the 144 (75%) dogs, while in 35 (24.3%) additional surgery was required, including cervical lymphadenectomy, muscular and internal jugular vein resection, and parathyroid gland autotransplantation. 26 Often, even with marginal resection it was possible to obtain a complete removal of the thyroid tumors, which was considered curative.
Whenever possible, the surgical procedures performed have led to complete removal of the thyroid tumors, even with small margins. In those cases where local invasion of the blood vessels or nerves was noted, resection of the venous structures was performed, while the arterial and nervous structures were subjected to debulking with safeguarding of the latter. This approach has still allowed an excellent local control of the thyroid tumors. At surgery, the mean major axis of the excised lobe was mea- Eighteen (12.4%) dogs received levothyroxine supplementation and 14 (9.7%) calcitriol and calcium following surgery, or calcium following surgery. Of the latter, 12 (85.7%) cases had bilateral thyroid lobes involvement and benefited also from autotransplantation of 1 of the parathyroid glands. Twelve (66.7%) of these 18 dogs were euthyroid before thyroidectomy, 2 (11.1%) had decreased T4, 2 (11.1%) had increased T4, and in the remaining 2 (11.1%) the thyroid status was not recorded before surgery.

| Risk of death during study period
The observational period ranged between 7 and 3240 days (median

| Cause-specific CIF
The estimated effect of metastasis on cause-specific CIFs is shown in Figure 2 and Table 3 Table 4).
The presence of thrombosis was associated with a higher risk of death because of other causes (CIF = 0.03 at 30 days, CIF = 0.62 at 2000 days; P = .044), but did not affect risk of death because of thyroid tumor (Figure 4, Table 5).
Administration of adjuvant chemotherapy was associated with an increase in risk of death because of thyroid cancer (CIF = 0.00 at 30 days, CIF = 0.16 at 365 days; P < .001) but not because of other causes ( Figure 5, Table 6).

| Cause-specific hazard ratios
The cause-specific hazard ratios for thyroid cancer (HR TC ) and for other cause of death (HR OC ) are shown in Table 7 and represent the F I G U R E 1 Kaplan-Meier estimate of overall survival (ie, regardless of the death cause) and cumulative incidence (mortality) functions for each of the 2 outcomes (death by thyroid cancer and death by other cause).
T A B L E 2 Cumulative incidence (mortality) and Kaplan-Meier survival at various time points in dogs with thyroid tumor treated by thyroidectomy.
Cumulative mortality (%)  Regarding the survival of affected dogs, the results of this series are similar to previous investigations that indicated an MST ranging between 500 and 1100 days. 11,19,22,23,36 Based on the cumulative mortality assessment, in the present study the risk of death because of thyroid cancer during the first year after thyroidectomy was approximately 10%, and at 5 years approximately 25%. Not unexpectedly, the latter was significantly lower than the risk of death because F I G U R E 4 CIFs for each of the 2 outcomes for subjects with or without thrombosis at the time of surgery.
T A B L E 5 Cause-specific cumulative mortality by presence or absence of thrombosis, at various time points after surgery. In this case series, metastasis at diagnosis was a major risk factor of death in dogs with thyroid cancer, and survival was 35% lower in those with tumor spread. However, although the presence of metastasis at diagnosis was associated with a poor outcome, 6 dogs with metastasis lived longer than 500 days. Hence, thyroidectomy could be taken into consideration in certain cases with metastatic disease, despite the pros and cons that need to be clearly discussed with the owner.
Notably, in a study the presence of metastases in 15 dogs undergoing thyroidectomy did not negatively affect the outcome. 14 Recently, in an investigation into the prognosis of canine thyroid carcinoma based on computed tomography assisted staging, the presence of metastatic disease had no influence on MST, although a trend F I G U R E 5 CIFs for each of the 2 outcomes for subjects with or without postoperative chemotherapy.
T A B L E 6 Cause-specific cumulative mortality by administration of postoperative chemotherapy, at various time points after surgery. As expected, in the present series higher age 3  Thrombosis was also identified as a cause of death unrelated to thyroid cancer. In humans with thyroid carcinoma tumor, thrombosis is associated with an increased chance of death because of thyroid cancer but not to all other causes. 43 The underlying explanation in dogs remains unclear.
In our study, an association between histological tumor type and metastasis was observed. In turn, metastasis was associated with thrombosis, thyroid tumor size, and bilateral tumor lobe involvement.
Thus, not unexpectedly, tumors that spread to other sites are frequently those that have the ability to grow larger and invade the vessels. 20 In addition, similarly to a previous study where larger thyroid tumors were more locally invasive in dogs, 7 we found that tumor thrombosis was associated with bilateral lobe involvement and a longer tumor major axis. Therefore, it could be suspected that in the pathophysiology of thyroid tumors in dogs, metastasis, thrombosis, and size could be connected, but further investigation is needed in order to verify if this association is consistent.
With regard to the year of surgery, earlier years were associated with an increased risk of death because of other cause. This might be T A B L E 7 Cause-specific hazard ratios and 95% confidence interval for each covariate based on the extended Cox Proportional Hazard model calculated for the 2 competing causes of death. because of improved veterinary care in the last decade, possibly increasing the overall life expectancy of the most recent cases. Moreover, the use of the latest generation surgical instruments allows the closure of the thyroid vascularization in a much safer way in humans.
LigaSure, seems to generate less thermal damage to the treated tissues. 44 In veterinary medicine, their use is associated with a reduction in surgery time but does not seem to influence the complication rate and postsurgical hospitalization time. 45 None of the remaining risk factors assessed in this investigation, The present investigation has some limitations. Because of its retrospective nature, some data were incomplete. There was also no standardized adjuvant chemotherapy protocol used to treat the affected dogs. The histopathological evaluation of the excised thyroid tumors was performed by different pathologists and samples could not be collected to review the diagnosis and add further testing.
In conclusion, dogs with thyroid tumors undergoing thyroidectomy have a long life expectancy. Undifferentiated carcinomas and metastases are major risk factors for thyroid cancer-related death.
Although metastases have a guarded outcome, extended survival is observed in certain dogs. Despite the rare presence of nonfollicular carcinoma and metastases, thyroidectomy should still be considered in some of these dogs. Similar to prior investigations, adjuvant chemotherapy did not prove to be beneficial in affected dogs. Unlike some studies, but confirming others, thyroid tumor size, invasiveness and bilateral thyroidectomy did not affect outcomes. The underlying reasons for thrombosis being a risk factor for death unrelated to thyroid cancer are unclear.