Serum thyroxine and thyrotropin concentrations decrease with severity of nonthyroidal illness in cats and predict 30‐day survival outcome

Abstract Background In cats, nonthyroidal illness affects serum thyroid hormone concentrations. Serum thyroxine (T4) and triiodothyronine (T3) concentrations commonly decrease, whereas free T4 (fT4) concentrations vary unpredictably. Limited information exists regarding effects on serum thyrotropin (thyroid‐stimulating hormone [TSH]) concentrations in cats with nonthyroidal illness syndrome (NTIS). Objectives To characterize alterations in thyroid function that develop in cats with NTIS and to correlate these alterations with severity and outcome of the nonthyroidal illness. Animals Two hundred and twenty‐two cats with NTIS and 380 clinically normal cats of similar age and sex. Methods Prospective, cross‐sectional study. All cats had serum T4, T3, free T4, and TSH concentrations measured. Cats were grouped based on illness severity and 30‐day survival. Results Cats with NTIS had lower serum T4 and T3 concentrations than did normal cats (P < .001). Serum fT4 and TSH concentrations did not differ between groups. Serum T4, T3, and fT4 concentrations progressively decreased with increasing disease severity (P < .001). The 56 cats that died had lower T4, T3, and TSH concentrations than did the 166 survivors, with no difference in fT4 concentration. Multivariable logistic regression modeling indicated that serum T4 and TSH concentrations both predicted survival (P < .02). Conclusions and Clinical Importance Cats with NTIS commonly develop low serum T4, T3, and TSH concentrations, the prevalence and extent of which increases with disease severity. Clinicians should consider evaluating thyroid function in cats with severe NTIS, because doing so could help determine probability of successful treatment responses before investing considerable time, effort, and finances in addressing the underlying disease.


| INTRODUCTION
In humans, a variety of acute and chronic illnesses can alter the results of commonly used thyroid hormone function tests, such as serum total thyroxine (T 4 ), free thyroxine (fT 4 ), triiodothyronine (T 3 ), and thyrotropin (thyroid-stimulating hormone [TSH]) concentrations. [1][2][3] This condition, known as the nonthyroidal illness syndrome (NTIS, previously termed "sick euthyroid syndrome"), is not a primary thyroid disorder but instead results from changes in secretion of TSH, as well as altered secretion, transport, metabolism, tissue uptake, and action of the thyroid hormones. [3][4][5][6] A likely adaptive response to the systemic illness, NTIS attempts to decrease peripheral tissue energy expenditure and minimize metabolic demands during the stress of the illness. [3][4][5] Nonthyroidal illness can have marked effects on thyroid function tests. Human patients, especially those with severe or critical illness, commonly develop low serum T 4 and T 3 concentrations. [1][2][3]7,8 Similarly, several nonthyroidal illnesses suppress serum T 4 and T 3 to low concentrations in dogs. [9][10][11][12] In both humans and dogs, serum fT 4 concentrations, when measured by equilibrium dialysis, usually remain within the reference interval. [1][2][3]8 Most human patients with NTIS initially have normal serum TSH concentrations, but many will develop low TSH concentrations, especially those with severe illness. [1][2][3]8,13 Approximately 10% to 15% of human patients will develop high serum TSH concentrations, particularly during the recovery phase of their illness. 1,8,[14][15][16] Similarly, dogs with NTIS usually maintain normal serum TSH concentrations, but occasionally have high serum TSH concentrations. 9,10,17 In both human and dogs with NTIS, the finding of low serum T 4 or fT 4 concentrations, together with high TSH concentrations, complicates evaluation of thyroid function and increases the risk for misdiagnosis of primary hypothyroidism. [1][2][3]11,18 In both humans 3,19-23 and dogs 10,12 with NTIS, development of low serum T 4 and T 3 concentrations increases the likelihood of death, a finding that might be useful as a prognostic indicator. Furthermore, in both humans and dogs, finding of low serum TSH concentrations has predicted mortality. 12,16,22,[24][25][26] Few studies have examined the relationship between thyroid function and mortality in cats with NTIS. [27][28][29][30] In 2 studies that examined cats with a variety of nonthyroidal diseases, 27,28 cats that died or were euthanized had lower serum T 4 concentrations than did cats that survived, suggesting that serum T 4 concentrations may also be indicative of survival outcome. Similarly, a recent study of cats with panleukopenia reported that low serum T 4 concentrations were associated with poor outcome. 30 To our knowledge, no study has evaluated if serum T 3 or TSH concentrations can help predict survival outcome in cats with NTIS.
As in humans and dogs, recent studies have demonstrated the utility of serum TSH concentrations for diagnosing cats with iatrogenic and naturally occurring hypothyroidism. 31 [1][2][3]11,18 We sought to better determine the effect of nonthyroidal illness on commonly used serum pituitary-thyroid function tests (T 4 , T 3 , fT 4 , TSH) in cats. Furthermore, we sought to determine the effect of severity of illness and disease category on serum thyroid hormone and TSH concentrations, as well as to examine whether abnormalities in any of these hormones could predict patient outcome and survival.

| Study design and selection of cats
We enrolled 2 groups of client-owned cats for this prospective crosssectional study, which included cats with nonthyroidal illness and clinically normal cats. Cats with a history of hyperthyroidism were excluded.
Ethical approval for the study was obtained from our institution's animal use and care committee, and blood collection was performed after informed owner consent.

| Clinically normal, euthyroid cats
We recruited 380 clinically normal cats as controls, as well as to establish institutional reference intervals for serum T 4 , T 3 , fT 4 , and TSH concentrations. These cats were considered healthy based on an unremarkable client history, physical examination (ie, none had palpable thyroid nodules or showed signs of hypothyroidism 34 ), and routine laboratory testing (ie, CBC, serum biochemistry profile, and urinalysis).

| Cats with NTIS
Two hundred and twenty-two cats were diagnosed with NTIS on the basis of results of history, physical examination, laboratory testing (eg, CBC, serum biochemistry profile, urinalysis, FeLV, and feline immunodeficiency virus status), and, variably, as required by the primary disease process, imaging (eg, radiography, ultrasonography, computerized tomography, or magnetic resonance imaging), and cytology or histologic examination. All cats were considered to be euthyroid on the basis of results of history, physical examination (ie, none had palpable thyroid nodules or showed signs of hypothyroidism 34 ), and diagnostic tests that established a specific diagnosis of nonthyroidal disease. None of these cats had received medications within the 2-week period before blood sampling that might affect serum thyroid hormone concentrations (eg, nonsteroidal anti-inflammatory agents, sulfonamides, phenobarbital, tricyclic antidepressants, glucocorticoids), and none had received methimazole or thyroid hormone replacement. [37][38][39][40] Cats with NTIS were allocated to 3 groups based on disease severity (ie, mild, moderate, and severe). This judgment was made by the clinician who examined the cat, in consultation with the primary author (M.E. Peterson), and was based on a number of factors, including the cats' clinical signs, results of laboratory testing, duration of illness, need for hospitalization, response to treatment, and survival. In terms of hospitalization requirement and duration, we allocated cats to the mild disease group if the clinician believed that the cat could be treated as an outpatient. We allocated cats to the moderate disease group if the clinician recommended brief hospitalization (regardless of the owner's permission to hospitalize the cat). We allocated cats to the severe disease group if the clinician recommended intensive hospital care, whether or not the owner accepted these recommendations.
The 222 cats also were divided into 10 groups based on their primary category of disease (ie, cardiac, dermatologic, endocrine, gastrointestinal, hepatic, infectious, neoplastic, neurologic, respiratory, and urologic/renal disease). In cats that suffered from >1 disease, the selected category was based on the primary or most severe issue, as determined both by the clinician examining the cat and primary author (M.E. Peterson). Finally, these cats also were classified according to 30-day survival outcome (ie, alive or dead within 30 days of serum thyroid hormone testing).

| Assays for thyroid hormone and thyrotropin (TSH) concentrations
Serum concentrations of total T 4, total T 3 , fT 4 by dialysis, and TSH were determined by assays validated for use in cats as previously described. 41 The sensitivity (ie, limit of quantification) of the each assay was 6.5 nmol/L for T 4 , 0.55 nmol/L for T 3 , 5 pmol/L for fT 4 , and 0.03 ng/mL for TSH. 41 For the T 3 and TSH assays, analytic sensitivity was not low enough to distinguish low-normal from low concentrations (ie, many clinically normal cats have undetectable serum T 3 and TSH concentrations when measured by these assays). 41 All blood samples for hormone assays were centrifuged within 1 hour after collection; serum was separated and stored at ≤4 C until assayed by a commercial laboratory (Antech Diagnostics, Lake Success, New York) the next day.

| Data and statistical analyses
Data were assessed for normality using the D'Agostino-Pearson test and by visual inspection of graphical plots. 42 Data were not normally distributed; therefore, all analyses were performed using nonparametric tests.
Undetectable serum TSH concentration was defined as <0.03 ng/mL and all undetectable serum TSH concentrations were assigned an arbitrary value of 0.02 ng/mL for continuous data analysis, as previously described. 41 Similarly, all undetectable serum T 4 concentrations (<6.5 nmol/L) were assigned an arbitrary value of 3.5 nmol/L, whereas all undetectable serum T 3 concentrations (<0.55 nmol/L) were assigned an arbitrary value of 0.45 nmol/L for continuous data analysis.
We used data from our 380 clinically normal cats to establish our institutional reference intervals for serum concentrations of T 4 , T 3 , fT 4 , and TSH using a nonparametric method to identify the central 95th percentile interval (ie, 2.5 through 97.5th percentile range). 43,44 Table 1 shows our reference intervals with 90% confidence intervals (CIs) for the thyroid hormones determined using this method.
Results for continuous data (eg, serum thyroid hormone and TSH concentrations) are expressed as median (25th-75th percentile) and represented graphically as boxplots (Tukey method). 45 Results for qualitative data are expressed as ratio (breed, sex) or number (%) of cats. Continuous variables were compared between 2 groups by use of the Mann-Whitney U test and for ≥3 groups by the Kruskal-Wallis test, followed by the Dunn multiple comparisons test. 46,47 Categorical variables were compared among groups using the Chi-square test.
To evaluate the predictive value of serum thyroid hormone and TSH concentrations on survival, we performed logistic regression using the 30-day survival outcome as the dependent variable, and the serum T 4 , T 3 , fT4, and TSH concentrations as independent variables. 48,49 For this analysis, we entered serum T 4 and fT4 concentrations as continuous variables and serum T 3 and TSH concentrations as dichotomous (binary) variables (data coded 0 for undetectable concentrations; 1 for detectable concentrations). The significance of each explanatory variable was tested using the Wald test. Results of the model are reported in terms of adjusted odds ratios with 95% confidence intervals (95% CIs) for each explanatory variable. To evaluate the model's ability to discriminate between groups, we calculated the area under the receiver operating characteristic (ROC) curve. We also generated a classification table to compare the observed and predicted survival outcome and determine the percentage of cases correctly classified using the logistic regression model. 48 T A B L E 1 Reference intervals for total thyroxine (T 4 ), total triiodothyronine (T 3 ), free T4 by dialysis, and TSH established in 380 clinically normal cats

| Serum fT 4 concentrations
Serum fT 4 concentrations in the cats with NTIS (median = 27 pmol/L) did not differ from concentrations in the clinically normal cats T A B L E 2 Serum thyroxine (T 4 ), triiodothyronine (T 3 ), free T 4 , and TSH in 222 cats with NTIS grouped into 10 categories of disease and severity of illness

| Serum TSH concentrations
Serum TSH concentrations in the cats with NTIS did not differ from those in the clinically normal cats (median for both groups = 0.05 ng/mL; P = .93; Figure 1D). Serum TSH concentrations were slightly high in 7  (Table 2).

| Serum T 4 concentrations
The 68 cats with severe disease had lower serum total T 4 concentrations (median = 9.7 nmol/L) than either the 72 cats with moderate disease (20.6 nmol/L, P < .001) or the 82 cats with mild disease (27 nmol/L; P < .001; Figure 2A). Cats with moderate disease had lower serum T 4 concentrations than did the cats with mild disease (P < .001; Figure 2A) Figure 1 for key. NTIS, nonthyroidal illness syndrome; TSH, thyroidstimulating hormone

| Serum T 3 concentrations
The cats with severe disease had lower serum T 3 concentrations (median = 0.46 nmol/L) than did cats with mild disease (0.54 nmol/L; P < .001), and cats with moderate disease had lower serum T 3 concentrations (0.46 nmol/L) than did those with mild disease (P < .001; Figure 2B). Serum T 3 concentrations did not differ between cats with moderate and severe disease (P = .26; Figure 2B). Fifty of the 82 cats (61%) with mild disease, 57 of the 72 (79%) with moderate disease, and 57 of the 68 (84%) cats with severe disease had undetectable serum T 3 concentrations (P = .03).

| Serum fT 4 concentrations
Cats with severe disease had lower serum fT 4 concentrations (median = 20.7 pmol/L) than did cats with either moderate (29 pmol/L; P < 0.01) or mild disease (29 pmol/L; P < .001). Serum fT 4 concentrations did not differ between the cats with mild or moderate disease (P = .94; Figure 2C).

| Serum TSH concentrations
Serum TSH concentrations did not differ among the cats with mild (median = 0.05 ng/mL), moderate (0.06 ng/mL), and severe (0.04 ng/mL) nonthyroidal illness (P > .05, Figure 2D). However, when the mild and moderate groups were combined, serum TSH concentrations were higher in these 154 cats compared with the 68 cats with severe disease (P = .02).

| Serum thyroid hormone and TSH concentrations in cats with NTIS cats separated into 10 categories of disease
The cats with endocrine disease (all suffering from either severe diabetes or ketoacidotic diabetes mellitus) had lower serum concentrations of both T 4 and fT 4 (median = 10.9 nmol/L and 20 pmol/L, respectively) than did the cats in the other groups (  Cats that died also had lower serum TSH concentrations (0.02 ng/mL) than did cats that remained alive (0.05 ng/mL; Figure 3D; P = .04). were excluded from the model. The final logistic regression model for 30-day mortality in our cats included serum T 4 , T 3 , fT 4 , and TSH concentrations (Table 3).
Of the 4 variables, serum T 4 concentration showed the highest level of significance for predicting outcome (Table 3, Figure 4), with serum TSH concentration playing a lesser role in this model. The area under the ROC curve for this model was 0.789 (Table 3).
A classification matrix showing the distribution of our cats according to the observed and predicted outcome was derived from the model ( Table 4). The test specificity for this model was high

| DISCUSSION
Our results indicate that cats with nonthyroidal illness commonly develop low or undetectable serum concentrations of T 4 , T 3 , and TSH, the prevalence and extent of which increases with disease severity.
As previously reported in cats with NTIS, [27][28][29]   Our study had several limitations. First, we did not definitively rule out concurrent hyperthyroidism in our cats with NTIS with thyroid biopsy or thyroid scintigraphy. Only a few cats that recovered from their NTIS underwent follow-up serum thyroid hormone testing, because such long-term follow-up was not part of our study design. Although all of our cats with NTIS were considered euthyroid on the basis of history and physical examination (ie, none had clinical signs of hyperthyroidism or palpable thyroid nodules), the reported prevalence of palpable thyroid nodules in proven hyperthyroid cats ranges from 79% 55 to 98%, 56,57 so it is possible that we missed thyroid nodules in a few cats. It is also possible that an occasional cat had ectopic thyroid tissue that would not be identified by palpation.
Although 13 cats had high serum fT 4 concentrations, which could indicate hyperthyroidism, none showed consistent clinical signs. None of the NTIS survivors became clinically hyperthyroid after resolution of their NTIS.
Therefore, the probability of having a large cohort of occult hyperthyroid cats, with thyroid disease being masked by NTIS, is, in our opinion, small.
Several cats with NTIS had multiple comorbidities, which made it difficult to categorize these cats into 1 of the 10 disease groups. In these cases, we allocated the cats to a particular disease group based on most important or severe disease, as determined both by the clinician examining the cat and primary author (M.E. Peterson). However, ultimately, disease group did not appear to be as important as severity of illness in determining the proportion of cats with suppressed serum T 4 or TSH concentration, and therefore, predicting survival outcome.
The third limitation of our study concerns the poor analytic sensitivity of the commercial TSH assay. The assay has a lower limit of detection which is high enough to include both normal and low concentrations. Therefore, we could not differentiate low-normal concentrations from truly low concentrations (ie, many of our clinically normal cats had undetectable serum TSH concentrations, similar to the cats with NTIS). A more sensitive, feline-specific TSH assay, which could differentiate truly low serum TSH from low-normal TSH concentrations, would help determine the value of assessing TSH when prognosticating about survival outcome in cats with NTIS.
In conclusion, our results indicate that cats with NTIS commonly develop low serum T 4 , T 3 , and TSH concentrations, the prevalence and extent of which increase with disease severity. In addition, we found that lower serum T 4 and undetectable TSH concentrations both were associated with mortality and can be used to help predict survival outcome in cats with NTIS.

ACKNOWLEDGMENTS
No funding was received for this study. We thank John Loftus and John Lucy for help in collecting the samples. Preliminary results of this study were presented as an oral abstract at the 2015 ACVIM Forum in Indianapolis, Indiana.