Thyroid hormone abnormalities and outcome in dogs with non-thyroidal illness
The authors of this article were in receipt of a Petsavers Grant to support aspects of their clinical research.
Abstract
Objectives: To document thyroid hormone abnormalities in dogs with non-thyroidal illness and identify markers of prognostic value.
Methods: Circulating total and free thyroxine, total triiodothyronine and thyrotropin concentrations were measured in 196 dogs with non-thyroidal illness. Clinical signs, previous medications and outcome were recorded in each case. Data were analysed to determine endocrine prognostic factors, and to document the prevalence of thyroid hormone abnormalities.
Results: Total triiodothyronine, and total and free thyroxine concentrations were decreased in 75·9, 34·7 and 4·5 per cent of cases, respectively. Dogs which were euthanased had significantly decreased total triiodothyronine, and total and free thyroxine concentrations compared with those which made a full recovery. Total triiodothyronine concentrations were significantly lower in dogs that were euthanased compared with those which made a partial recovery.
Clinical Significance: Thyroid hormone concentrations may be used as prognostic indicators in dogs with non-thyroidal illness. Low triiodothyronine syndrome may be more common in dogs than previously recognised.
Introduction
Despite the large number of referral veterinary institutions, there is little information available concerning clinical outcome of patients and prognostic indicators within these facilities. Although often assumed that certain clinical factors, such as disease type and severity, are associated with a poorer prognosis, this has not been objectively assessed.
Alterations in thyroid hormone concentrations in response to non-thyroidal illness (NTI), often termed the “euthyroid sick syndrome”, is well recognised in both human beings and animals. The most commonly recognised abnormality in human beings is suppression of serum total triiodothyronine (T3) concentration alone, with total thyroxine (T4) values remaining within reference limits. In more severe, frequently life-threatening illness, suppression of both serum total T4 and total T3 concentration occurs. Free T4 concentration is typically maintained within reference limits (Wiersinga 2005).
By contrast, suppression of total T3 concentration alone appears uncommon in dogs with NTI. Concurrent suppression of total T4 and T3 values, or lone suppression of total T4 concentrations, is more frequently reported (Peterson and others 1984, Larsson 1988, Elliot and others 1995, Peterson and others 1997, Kantrowitz and others 2001). Free T4 concentration is less affected by NTI, with subnormal concentrations more common in severely ill animals (Kantrowitz and others 2001).
Most previous canine studies have only documented the prevalence of thyroid hormone abnormalities in NTI, without assessing their prognostic value. Two studies examined the prognostic value of certain thyroid hormone parameters but were confined to animals with either critical illness or parvovirus infection (Elliot and others 1995, Schoeman 2006). A large study assessing the prognostic value of a wide range of thyroid hormone concentrations across a variety of disease types and severities has not yet been performed.
The aim of this study was to investigate a comprehensive range of thyroid-function-related tests in dogs with varying NTI to identify endocrine markers of prognostic value.
Materials and methods
Case material and sample collection
A random selection of canine serum and plasma samples were collected during the routine investigation of clinical cases at the University of Glasgow Veterinary School. Once aliquots of the samples had been removed for their primary analyses, the remainder was frozen and stored until thyroid hormone assays were performed.
Corresponding case details were collected by conducting interviews with relevant clinicians shortly after sample submission. Data included signalment and clinical and historical features including recent known drug therapies. There were a limited number of options for disease category, disease course and outcome from which clinicians were requested to select only one. The following disease categories were provided: cardiovascular, congenital, dermatological, endocrine, gastrointestinal, general medicine, immune-mediated, infectious, neoplastic, neurological, ophthalmological, orthopaedic, respiratory, soft tissue, trauma, urogenital, unknown and other. If more than one category applied to a case, the clinician selected the one considered to be the principal complaint. The eventual outcome of cases was determined by repeated interview with the relevant clinician or by examination of case notes. Four outcomes were possible: died, euthanased, partial recovery and full recovery. Animals within the first two categories died or were euthanased, or were recommended to be euthanased, during the hospitalisation period. The dogs within the latter two categories survived and were discharged from hospital. Partial recovery was defined as recovery without complete cure and/or the requirement for permanent medication to control clinical signs. Animals within the full-recovery category achieved full clinical recovery without the requirement for continuing therapy.
For the purposes of this study, drugs capable of suppressing thyroid function were considered to be non-steroidal anti-inflammatory drugs, anticonvulsants, glucocorticoids and trimethoprim-potentiated sulphonamides. Dogs that had received thyroid hormone replacement therapy were excluded from the study.
Sample analysis
Total T4 and total T3 concentrations were measured using commercial radioimmunoassays (MAGIC T4 and MAGIC T3, respectively; Chiron Diagnostics). Free T4 concentration was measured by equilibrium dialysis (Free T4 by Equilibrium Dialysis; Nichols Institute Diagnostics) and thyrotropin (cTSH) concentrations by commercial immunoradiometric assay (Canine TSH IRMA; Diagnostic Products). All assays were validated for use in the dog (Dixon 1999). All standards, samples and controls were run in duplicate and samples with a high duplicate error were rejected and re-assayed.
Data analysis
The ages of the dogs in each of the outcome groups were compared using analysis of variance with post hoc pairwise comparison of means using Newman-Keuls multiple range tests. The Kruskal-Wallis test, followed by the Dunn test, was used to assess differences between thyroid hormone concentrations of dogs with different outcomes. The effect of thyroid suppressive medication on hormone results was evaluated with a Mann-Whitney U test. The effect of disease category and duration of illness on eventual outcome was evaluated by chi-squared analysis. Where chi-squared testing was performed, only categories which had data cells from more than one variable, with five or more cases in each cell, were included in the statistical analysis. P values less than 0·05 were considered statistically significant.
Hormone results less than the limit of detection (LoD) of the assay were assigned a value equal to the LoD. Analyses were performed on an expanded database of the population described by Dixon and Mooney (1999) to determine optimal cut-off values of 14·8 nmol/l (total T4), 6·66 pmol/l (free T4), 0·69 ng/ml (cTSH) and 1·77 nmol/l (total T3) for diagnosing hypothyroidism. Dogs with clinical signs suggestive of hypothyroidism were classified as euthyroid or hypothyroid by TSH stimulation testing. Receiver operating characteristic (ROC) curve and differential positive rate analyses were used to determine the cut-off value with the greatest combined sensitivity and specificity and, therefore, accuracy. The sensitivity, specificity and area under the ROC curve were 0·96, 0·70 and 0·92 for total T4; 0·81, 0·93 and 0·93 for free T4; 0·76, 0·80 and 0·81 for total T3; and 0·79, 0·82 and 0·82 for cTSH concentrations, respectively (Dixon 1999). These cut-off values were used to define values as subnormal (total and free T4, total T3) or elevated (cTSH).
Results
Clinical material
Samples were collected from 196 dogs. The mean age was 6·8 years (range 0·3 to 16·5 years). Eighty-three were female (43 neutered) and 113 male (23 neutered). Only one dog died in the course of the study, during surgery for orthopaedic disease. Of the 39 euthanased cases, it was considered likely that the majority would have ultimately died of their primary disease if euthanasia had not been performed. Results are summarised in Table 1.
| Euthanased | Partial recovery | Full recovery | |
|---|---|---|---|
| Number of dogs (total) | 38 | 81 | 76 |
| Disease course | |||
| Acute | 15 | 8 | 19 |
| Chronic | 23 | 73 | 57 |
| Disease category | |||
| Gastrointestinal disease | 3 | 8 | 12 |
| Neoplasia | 22 | 16 | 7 |
| Orthopaedic disease | 1 | 8 | 18 |
- * Only three disease categories have been included because there were insufficient case numbers within the other groups to allow further analysis of effect on outcome
Only neoplasia (n = 45), gastrointestinal (n = 23) and orthopaedic (n = 28) disease categories had sufficient cases to evaluate effect of disease category on eventual outcome. Dogs with neoplasia were significantly less likely (P<0·01) to make a full recovery but more likely (P<0·05) to make a partial recovery compared with the other two groups. Orthopaedic cases were significantly more likely (P<0·05) to make a full recovery and less likely (P<0·05) to make a partial recovery. Chi-squared analysis of the neoplasia group compared with all other dogs, when considered as a single group, showed a highly significant (P<0·0001) increase in the likelihood of euthanasia in the dogs with neoplasia.
Dogs with acute disease were significantly (P<0·01) more likely to be euthanased but less likely to make a partial recovery than those with chronic disease. Cases which made a full recovery were significantly (P<0·05) younger than both those which were euthanased and those which made a partial recovery. There was no significant difference in outcome of dogs based on either gender or neutering status.
Thyroid hormones
Of the 196 samples, 62 (31·6 per cent) were collected from dogs which had not recently received any medication, 65 (33·2 per cent) from dogs which had recently received thyroid-suppressing medications and the remaining 69 (35·2 per cent) from dogs which had received other medications. There was no significant difference in eventual outcome of cases between any of these groups.
Results of thyroid hormone analyses are summarised in Table 2, presented both before and after exclusion of cases which had recently received thyroid suppressive therapies. Median total and free T4 values were decreased by approximately 15·5 and 14·0 per cent, respectively, in dogs receiving thyroid suppressive therapy compared with the values from dogs not receiving such therapy. There was no significant effect on circulating total T3 or cTSH concentrations.
| Dogs receiving thyroid suppressive therapies, n=65 | Dogs not receiving thyroid suppressive therapies, n=131 | Combined results, n=196 | |
|---|---|---|---|
| Total T4 (nmol/l) | 17·44 (17·02)e | 20·65 (15·47)f | 20·08 (15·92) |
| Free T4 (pmol/l) | 19·56 (13·24)e | 22·75 (11·79)f | 21·60 (12·72) |
| Total T3 (nmol/l) | 1·28 (0·59) | 1·39 (0·78) | 1·36 (0·75) |
| cTSH (ng/ml) | 0·12 (0·20) | 0·14 (0·22) | 0·12 (0·21) |
- * Results are shown for dogs receiving thyroid suppressive therapies, for dogs not receiving thyroid suppressive therapies and for both groups of dogs combined. Hormone results with a different superscript are statistically different from each other: a-b, P<0·001; c-d, P<0·01; e-f, P<0·05
Including all dogs, total T3, total T4 and free T4 concentrations were decreased in 148 of 195 (75·9 per cent), 68 of 196 (34·7 per cent) and eight of 178 (4·5 per cent) of cases, respectively. In dogs in which both total T4 and T3 were measured (n = 195), total T3 alone was reduced in 82 (42·1 per cent), total T4 alone was reduced in three (1·5 per cent) and both hormones were concurrently reduced in 64 (32·8 per cent).
Results of thyroid hormone analyses categorised according to eventual case outcome are shown in Table 3 and 1-4. Dogs which were euthanased had significantly decreased total T3, total T4 and free T4 concentrations compared with those which made a full recovery, and significantly decreased total T3 concentrations compared with those which made a partial recovery. No other hormone parameter was significantly different between dogs with different outcomes. The dog that died had total T3, total T4 and free T4 concentrations of 1·72 nmol/l, 25·56 nmol/l and 25·41 pmol/l, respectively.
| Euthanased, n=38 | Partial recovery, n=81 | Full recovery, n=76 | |
|---|---|---|---|
| Total T4 (nmol/l) | 12·65 (14·26)c | 18·64 (17·8) | 22·46 (13·33)d |
| Free T4 (pmol/l) | 18·42 (15·30)e | 20·82 (11·30) | 24·40 (11·83)f |
| Total T3 (nmol/l) | 0·99 (0·49)a | 1·41 (0·80)b | 1·46 (0·73)b |
| cTSH (ng/ml) | 0·11 (0·22) | 0·16 (0·23) | 0·09 (0·20) |
- * Hormone results with a different superscript are statistically different from each other: a-b, P<0·001; c-d, P<0·01; e-f, P<0·05
Circulating total triiodothyronine (T3) concentrations in dogs with various diseases, categorised by eventual outcome (n=195). The solid lines represent the median value within each group. The dotted line represents the cut-off value as determined by differential positive rate analysis
Circulating total thyroxine (T4) concentrations in dogs with various diseases, categorised by eventual outcome (n=196). See 1 for further interpretation
Circulating free thyroxine (T4) concentrations in dogs with various diseases, categorised by eventual outcome (n=178). See 1 for further interpretation
Circulating thyrotropin (cTSH) concentrations in dogs with various diseases, categorised by eventual outcome (n=181). See 1 for further interpretation
Of the 148 dogs with subnormal total T3 values, one (0·7 per cent) died, 37 (25·0 per cent) were euthanased, 56 (37·8 per cent) made a partial recovery and 54 (36·5 per cent) made a full recovery. Only one dog with a normal total T3 concentration was euthanased. Dogs with subnormal total T3 values were significantly more likely to be euthanased than dogs with normal values. Of the dogs with subnormal total T3 values which had not received thyroid suppressive medication, total and free T4 values were subnormal in 37 (38·5 per cent) and two (2·1 per cent), respectively.
Because of the effect of thyroid suppressive medication on total and free T4, only cases which had not received such drugs were included in further analysis of these parameters. Of the 39 cases with low total T4 concentration which had not received thyroid suppressive medications, 13 (33·3 per cent) were euthanased, 16 (41·0 per cent) made a partial recovery and 10 (25·6 per cent) made a full recovery. These dogs were significantly more likely to be euthanased than dogs with normal total T4 results. Free T4 values were estimated in 36 of the 39 cases with subnormal total T4 and were subnormal in one (2·8 per cent). Total T3 was measured in all 39 of these cases and was decreased in 38 (97·4 per cent).
In total, two dogs that had not received thyroid suppressive medications had decreased circulating free T4 concentrations and both of these cases were ultimately euthanased. Circulating free T4 values were less than 10 pmol/l in nine dogs. Three of these cases were euthanased, three made a partial recovery and three made a full recovery.
Circulating cTSH concentrations were not significantly different between dogs with different outcomes (Fig 4). Circulating cTSH results were increased in six of 181 (3·3 per cent) samples. The total T4 concentration was concurrently decreased in two cases. Both of these cases had systemic neoplasia. All six dogs with elevated cTSH results had reference-range free T4 values.
Discussion
The results of this study provide extensive data on the association of both clinical and endocrinological factors with disease outcome and provide an insight into their use as prognostic indicators in dogs.
The clinical factors identified as prognostic indicators are largely predictable and confirm many of the assumptions commonly made regarding prognosis. Acute cases were significantly more likely to be euthanased. However, this may have reflected more rapid referral of animals suffering from severe disease. Younger dogs were more likely to make a full recovery than older dogs. While this may reflect an improved ability of the younger animal to cope during systemic illness, it is likely that these results are influenced by the nature of diseases affecting animals of different ages. For example, results may be influenced by a higher prevalence of neoplastic compared with orthopaedic disease in older animals. The complexity of collecting data from clinical cases is recognised. For instance, the euthanased category is potentially influenced by owner request and financial constraints and some patients may have had more than one disease process concurrently. However, the majority of euthanased cases were judged likely to die of their primary disease if euthanasia was not performed, and clinicians were requested to select the category considered to be the principal complaint. Interviews with relevant clinicians at the time of sampling avoided potential unnecessary bias if case records alone were retrospectively analysed.
The effect of thyroid suppressive medication on circulating total and free T4 concentrations largely supports the findings of previous studies (Woltz and others 1983, Torres and others 1991, Panciera and Post 1992, Daminet and others 1999, Gaskill and others 1999, Kantrowitz and others 1999, Daminet and Ferguson 2003, Panciera and others 2006). However, there was no significant effect of these drugs on total T3 or cTSH concentrations which is in contrast to previous reports (Rosychuk 1982, Woltz and others 1983, Daminet and Ferguson 2003). This may reflect variation in drug type, dose and duration of therapy which were not evaluated in the present study.
In contrast to previous reports, the most common thyroid hormone abnormality encountered was a reduction in total T3 alone. Concurrent reduction of both total T4 and total T3 was also common, but the prevalence of subnormal total T4 alone was very low. These results are broadly similar to those described in human beings with NTI (Wiersinga 2005, De Groot 2006, Peeters and others 2006). This reduced prevalence of subnormal total T4 may indicate that as in human beings, T4 suppression is associated with more severe or advanced illness. In support of this was the finding that the frequency of subnormal total T3 in cases with depressed total T4 values was nearly 100 per cent, whereas only approximately 40 per cent of dogs with subnormal total T3 values had a concurrent decrease in total T4 concentration.
Previous studies in dogs have reported that total T3 is less consistently decreased in NTI than total T4. Studies using a wide range and severity of NTI have documented subnormal total T3 concentrations in 5·9 to 16·1 per cent of cases (Larsson 1988, Kantrowitz and others 2001). A higher prevalence of subnormal total T3 has been documented in animals with hyperadrenocorticism and critical illness (52 and 56 per cent, respectively) (Peterson and others 1984, Elliot and others 1995), but these studies are unlikely to be representative of a broader range of NTI as both hyperadrenocorticism and severe illness are specifically more likely to be associated with reduced T3 concentrations (Peterson and others 1984, Kantrowitz and others 2001).
One explanation for the higher prevalence of subnormal total T3 values in this study relates to the reference ranges used to identify “normal” from “abnormal” values. Commonly, previous studies compared hormone results with those obtained from a putatively healthy group of dogs (Larsson 1988, Peterson and others 1997, Kantrowitz and others 2001). In the present study, the cut-off value used to classify results as abnormal was the concentration which most reliably distinguished dogs presenting with clinical signs suggestive of hypothyroidism as truly hypothyroid or euthyroid. While both approaches are justifiable, the advantage of the analyses used in the present study are that the identification of subnormal values in the dogs with NTI is genuinely separating those dogs based on the severity of their illness rather than simply the presence or absence of illness as occurs in most other studies. This was considered to be a more appropriate approach to this problem and typifies the complexities involved in selecting an appropriate control group.
Of all the parameters of thyroid function evaluated, total T3 concentrations were most strongly correlated with eventual outcome, consistent with the report by Elliot and others (1995). Conversely, the identification of a normal total T3 value was highly predictive of survival although like all the parameters studied, this could not differentiate between the likelihood of a partial or full recovery.
Euthanasia was significantly more likely in dogs with subnormal total T4 concentrations compared with dogs with normal values, a finding similar to reports in other species (Slag and others 1981, Mooney and others 1996). This correlation between suppressed total T4 concentration and decreased likelihood of survival was not identified by Elliot and others (1995) but was recently described in dogs with parvovirus infection (Schoeman 2006).
Analysis of the likelihood of euthanasia in dogs with subnormal free T4 values could not be evaluated because of an inadequate number of cases. However, both cases in which subnormal free T4 concentration was detected were ultimately euthanased suggesting that a subnormal free T4 concentration may be a very poor prognostic indicator in dogs with NTI. Further studies with additional cases are required.
The fact that circulating cTSH concentrations were not correlated to severity of illness is not surprising and confirms previous reports (Kantrowitz and others 2001). In human beings, TSH concentrations are within reference range in most patients with NTI, but reduced concentrations have been reported in critical illness, usually with concurrently reduced total T4 concentrations (Wehmann and others 1985). While similar reductions are likely to occur in canine patients, the sensitivity of the current canine assay does not allow differentiation between normal and reduced TSH values. A few dogs had elevated cTSH concentrations, possibly reflecting recovery from NTI as occurs in human beings (Hamblin and others 1986). Serial analyses would be required to evaluate these cases further.
Undoubtedly, some dogs in the present study may have had hypothyroidism. However, the overall prevalence of hypothyroidism is low, and none of the dogs exhibited features of hypothyroidism that would have warranted its investigation. In addition, all dogs with elevated cTSH concentrations had reference-range free T4 concentrations, and all dogs with reduced free T4 concentrations had reference-range cTSH concentrations, making hypothyroidism unlikely (Dixon and Mooney 1999).
In conclusion, this study has shown that thyroid hormone concentrations may be used as prognostic factors in dogs with NTI. Low T3 syndrome may be more common in dogs than previously recognised and provides the most accurate means of assessing prognosis, particularly as it is least affected by a variety of previously administered thyroid suppressive medications.
Acknowledgements
This study formed part of a larger study of the investigation of hypothyroidism in dogs and was part funded by a BSAVA Petsavers award, for which the authors are extremely grateful.
