The data have previously been published in abstract form. Wakeling J, Elliott J, Syme HM. Subclinical hyperthyroidism in cats. Endo Abstr 2007;13:315; and Wakeling J, Elliott J, Syme HM. TSH measurement in cats: a prospective study J Vet Intern Med 22(3):725.
Evaluation of Predictors for the Diagnosis of Hyperthyroidism in Cats
Version of Record online: 30 AUG 2011
Copyright © 2011 by the American College of Veterinary Internal Medicine
Journal of Veterinary Internal Medicine
Volume 25, Issue 5, pages 1057–1065, September/October 2011
How to Cite
Wakeling, J., Elliott, J. and Syme, H. (2011), Evaluation of Predictors for the Diagnosis of Hyperthyroidism in Cats. Journal of Veterinary Internal Medicine, 25: 1057–1065. doi: 10.1111/j.1939-1676.2011.00790.x
- Issue online: 20 SEP 2011
- Version of Record online: 30 AUG 2011
- Manuscript Accepted: 18 JUL 2011
- Manuscript Revised: 24 JUN 2011
- Manuscript Received: 24 OCT 2010
- Feline Research and Petplan Charitable Trust
- Subclinical hyperthyroidism;
- Thyrotropin hormone (TSH);
- Total thyroxine (tT4)
In humans, subclinical hyperthyroidism is diagnosed when serum thyroid hormone concentrations are within the reference range but thyroid stimulating hormone (TSH) concentration is subnormal. In a previous study, a higher prevalence of thyroid nodular disease was found in euthyroid geriatric cats with undetectable TSH (<0.03 ng/mL) compared to those with detectable TSH concentrations, suggesting subclinical hyperthyroidism might also exist in cats.
Euthyroid cats with undetectable TSH concentrations have subclinical hyperthyroidism and may subsequently develop overt signs of hyperthyroidism.
One-hundred four client-owned cats.
In this prospective cohort study, euthyroid geriatric (≥9 years) cats were recruited during routine health checks. Plasma biochemistry was performed at baseline and every 6 months thereafter. Total thyroxine and TSH concentrations were determined annually. Short-term follow-up data (within 14 months of recruitment) were used to detect variables at entry that were predictive of the diagnosis of hyperthyroidism, using univariable analysis followed by multivariable logistic regression analysis. Log rank analysis was used to test the association of initial TSH concentration with diagnosis of hyperthyroidism during the total available follow-up.
Results and Conclusions:
Median (range) follow-up was 26 (0–54) months and annual incidence of hyperthyroidism during the study was 7.4%. Cats that became hyperthyroid within 14 months had higher ALKP activity (P = 0.02) and higher prevalence of goiter (P = .03) at baseline than controls. Cats with undetectable TSH at baseline (29/104; 28%) were significantly (P < .001) more likely to be diagnosed with hyperthyroidism. However, not all cats with undetectable TSH became hyperthyroid during the study.
area under the curve
chronic kidney disease
free (unbound) thyroxine
International Renal Interest Society
lost to follow-up
receiver operating characteristic
Royal Veterinary College
systolic blood pressure
thyroid stimulating hormone
urine specific gravity
Hyperthyroidism is a common disease of older cats and has many similarities to human toxic nodular goiter including insidious onset, presence of hyperplastic or adenomatous nodular changes in the thyroid, and peak incidence in old age. Hyperthyroidism in human patients is usually diagnosed by demonstration of subnormal thyroid stimulating hormone (TSH) concentrations in conjunction with increased free thyroxine (fT4) concentrations. A syndrome of subclinical hyperthyroidism has been identified in human patients who have within-reference-range fT4 and tri-iodothyronine (T3) concentrations but TSH concentrations below the reference range.[2-4]
Elderly human patients have an increased risk of nodular goiter and subclinical hyperthyroidism particularly if they experience chronic mild to moderate iodine deficiency throughout their lifetime. In iodine deficient areas, the prevalence of subclinical hyperthyroidism in older people (55–65 years old) is reported to be 4–15%.[5-7] Elderly patients with subclinical hyperthyroidism have been shown to develop overt hyperthyroidism at a rate of approximately 2–9% per year,[8, 9] and the risk of developing hyperthyroidism increases with age.[3, 4, 9] Patients with large nodular thyroid glands and subnormal serum TSH concentrations may be at particular risk of developing clinical hyperthyroidism when dietary iodine intake is increased, or when given iodine-containing contrast agents.
In cats, TSH can be measured using the DPC chemiluminescent canine TSH assay[11, 12]; however, the assay has suboptimal sensitivity with TSH concentrations below the limit of quantification of the assay (<0.03 ng/mL) in both hyperthyroid and in some apparently euthyroid cats.[11, 13] Despite this limitation, 2 previous retrospective studies in cats have provided evidence that a period of subclinical hyperthyroidism may exist in this species. In the first retrospective study, TSH concentrations were undetectable in 15 of 16 cats 1–3 years before the diagnosis of hyperthyroidism.2 The second study documented a higher prevalence of histopathologic changes in the thyroid gland consistent with hyperthyroidism in euthyroid geriatric (≥9 years) cats with TSH <0.03 ng/mL compared to euthyroid geriatric cats with TSH ≥0.03 ng/mL. In addition, 3 other studies have documented that nodular hyperplasia and adenomas are found frequently in the thyroid glands of apparently euthyroid cats.[15-17]
The term subclinical hyperthyroidism as used in human medicine is something of a misnomer as this term actually refers to a biochemical diagnosis (ie, low TSH concentration with normal fT4 concentration or other thyroid hormones) and does not actually imply that the patients are completely asymptomatic. In fact, subclinical hyperthyroidism in humans has clinically relevant effects on cardiovascular function[18, 19] and bone density that can be reversed by treatment of the underlying thyroid disease.[21, 22] In particular, there is between a 2- and 5-fold increased risk of developing atrial fibrillation[23-25] and this risk is especially marked in elderly patients. As a result, diagnosis and treatment of subclinical hyperthyroidism in humans often is recommended for patients older than 60 years with TSH concentration <0.1 mIU/L, especially those with, or at risk of, heart disease, osteopenia, osteoporosis, and those with symptoms suggestive of hyperthyroidism. However, some longitudinal studies have[27-29] (and other studies have not[24, 30]) found an increased risk of mortality or adverse cardiovascular events (other than atrial fibrillation) in patients with subclinical hyperthyroidism. The treatment of subclinical hyperthyroidism therefore remains controversial in human patients.
It is possible in cats that physical and biochemical changes associated with hyperthyroidism may also be detectable before confirmation of the diagnosis (ie, increased serum fT4 or tT4 concentration), as is the case in people. Changes such as the presence of goiter or an increased heart rate could act as markers for an increased risk of developing overt hyperthyroidism. Indeed, in 1 previously published study, prophylactic thyroidectomy was performed in apparently euthyroid cats that had goiter, because it was felt that these cats might have an increased risk of developing biochemical and clinical hyperthyroidism. However, a proportion of the cats that underwent surgery merely had cystic changes in their thyroids glands indicating that a more specific method of detecting impending hyperthyroidism is required.
The purpose of the present prospective cohort study was to determine whether euthyroid geriatric (≥9 years) cats with undetectable TSH concentrations (<0.03 ng/mL) are at increased risk for subsequently developing overt hyperthyroidism. A 2nd purpose of the study was to determine whether there are any signs associated with subclinical hyperthyroidism in cats.
The study population was drawn from 2 first opinion hospitals that primarily or exclusively treat pets of clients with low incomes: the Beaumont Animal Hospital (Royal Veterinary College, RVC) and the People's Dispensary for Sick Animals, London, UK. Cats are recruited to the RVC feline research clinic within the 2 hospitals by internal referral.
Inclusion and Exclusion Criteria
Geriatric cats (≥9 years old) were enrolled in this prospective cohort study between November 2004 and December 2006 and were included if they presented to the RVC feline research clinics for routine health screening rather than because the owner considered the cat to be unwell. Cats were not excluded if clinical concerns were identified during the initial consultation or if further testing disclosed underlying health problems such as chronic kidney disease (CKD) or hypertension. Cats that had been previously diagnosed with metabolic, renal, or clinically relevant cardiovascular disease, as well as cats that were on any medication other than routine parasiticides, were excluded. In addition, cats diagnosed with hyperthyroidism on the date of recruitment, or within 3 months of the study start date, also were excluded. Cats were considered to be euthyroid at recruitment if they had a serum tT4 concentration within reference range and they fulfilled the above inclusion criteria for the study.
At recruitment, a full history was obtained and clinical examination performed including systolic blood pressure (SBP) measurement using the Doppler technique as previously described.[32, 33] Blood samples were taken by jugular venipuncture between 9:00 am and 1:00 pm on the date of examination and were placed into heparinized and plain collection tubes. Serum and plasma were separated from the blood cells within 6 hours and packed cell volume and plasma biochemical analysis (including measurement of creatinine, urea, total protein, albumin, potassium, phosphate, chloride, sodium, and calcium concentrations and alkaline phosphatase [ALKP] and alanine transferase activities [ALT]) were performed.2 Serum tT4 and TSH measurements were performed at the RVC laboratory3 using the DPC feline (tT4) and canine (TSH) Immulite chemiluminescent assays. Undetectable serum TSH concentration was defined as <0.03 ng/mL and detectable TSH was defined as TSH ≥0.03 ng/mL throughout the study; 0.03 ng/mL is the limit of quantification of the assay in both dogs and cats. Where required for diagnosis, serum for fT4 measurement was submitted to an external, commercial, laboratory.3 Urine was collected by cystocentesis and urinalysis, including urine specific gravity (USG), dipstick chemistry analysis, urine pH, and microscopic sediment examination, was performed on the day of collection. Urine culture was performed if sediment examination supported the diagnosis of a urinary tract infection. The ethics and welfare committee of the RVC approved the diagnostic protocol and client consent forms. Sample collection was performed with the informed consent of the owner.
Diagnostic Criteria for Hyperthyroidism and Chronic Kidney Disease
Throughout the study, cats were diagnosed as hyperthyroid based on 1 of 3 criteria: serum tT4 concentrations >55 nmol/L (reference range, 19–55 nmol/L), serum tT4 concentration >40 nmol/L and fT4 >40 pmol/L (reference range, 19–40 pmol/L) in a cat with concurrent illness and compatible clinical signs of hyperthyroidism, or by T3 suppression test.
Chronic kidney disease was diagnosed based on plasma creatinine concentration exceeding the upper limit of the laboratory reference range (177 μmol/L) and low USG (<1.035) and further staged according to the International Renal Interest Society (IRIS) criteria. Hypertension was diagnosed when SBP was >170 mmHg in combination with hypertensive retinopathy or SBP >170 mmHg on at least 2 consecutive office visits.
Some cats diagnosed with hyperthyroidism during this study were treated by thyroidectomy. Histopathology was performed on these surgically removed thyroid glands, and hyperplasia and adenomatous changes were graded on a scale of 1–5 as described previously.
Cats without evidence of underlying health concerns were re-examined every 6 months. Owners of cats with azotemia, urinary tract infection, or hypertension were requested to bring their cats back to the RVC feline research clinic for re-evaluation and appropriate treatment every 1–8 weeks as clinically indicated, with routine re-evaluation every 6–8 weeks. At each follow-up visit, cats received a complete physical examination and SBP measurement, and plasma biochemistry was performed as described for the initial visit. Clients were reminded to bring healthy cats in for 6 monthly check-ups, with at least 2 telephone calls made and 1 letter sent before a client was classified as lost-to-follow-up (LTFU). Clients who failed to present their cat for a 6 month check-up also were sent reminders 12 months after recruitment. Repeat serum thyroid hormone analysis was performed at least every 12 months, or more frequently if historical (eg, weight loss, polyphagia), clinical (eg, goiter, tachycardia, murmur) or biochemical (eg, increased ALT or ALKP activities) signs suggested the possibility of thyroid disease. Unless cats died or were LTFU they were followed for at least 2.5 years.
Short-term (up to 14 months) Follow-up
In the 1st analysis, thyroid status at the first annual re-evaluation was determined and baseline variables were compared between those cats that were and those that were not diagnosed with hyperthyroidism within 14 months of enrollment. To be included in the analysis of baseline risk factors, cats had to either be diagnosed with hyperthyroidism within 14 months (hyperthyroid group) or to have been re-examined and confirmed as euthyroid (nonhyperthyroid group) approximately 1 year (9–14 months) after entry into the study. Euthyroid cats that were not re-examined at least 9 months after recruitment, either because they had died or were not made available for repeat examination and blood testing, were considered LTFU. LTFU cats were included as a separate group in the univariable analysis and were excluded from the final binary logistic regression analysis.
The following baseline data were examined for an association with the diagnosis of hyperthyroidism in the short-term follow-up data analysis: age, breed and sex, client-reported history of weight loss, vomiting, diarrhea, polyphagia, polydipsia, skin disease or behavioral change, and physical examination findings including body weight, presence of a goiter, SBP, heart rate, presence of a murmur, skin disease, moderate to marked dental disease, other diseases (eg, ocular disease, respiratory tract disease) and the following laboratory measurements: tT4, TSH, ALKP, ALT, potassium, phosphate, creatinine, PCV, and USG.
Long-term (up to 4.5 years) Follow-up
In the second analysis, all available follow-up data (2.5–4.5 years) were used to determine whether an undetectable baseline TSH concentration was a significant predictor for the subsequent diagnosis of hyperthyroidism. Time-to-event Kaplan–Meier curves were generated and log rank analysis was performed to compare the risk of diagnosis of hyperthyroidism in cats with detectable or undetectable TSH at baseline. Cats were censored if they reached the study end date without being diagnosed with hyperthyroidism, if they were LTFU, or if they died. All cats that met the criteria for entry to the study at the baseline visit were included in this analysis. The end date of the study was June 2009 giving a maximum possible follow-up of 4.5 years from the study start date in November 2004 and a minimum follow-up of 2.5 years from the end of recruitment in December 2006.
Data were analyzed using computerized statistics software4 and in all cases P < .05 was taken to indicate statistical significance.
In the short-term follow-up data analysis, continuous data initially were analyzed nonparametrically using the Mann–Whitney U test (for comparison of hyperthyroid and nonhyperthyroid groups) or Kruskal–Wallis test (for comparison of hyperthyroid, nonhyperthyroid and LTFU groups). TSH concentrations <0.03 ng/mL were assigned an arbitrary value of 0.02 for continuous analyses. Results for continuous data are reported as median [25th, 75th percentiles] (except where otherwise indicated). For all categorical binary variables, a chi-square test was performed and if a significant association was found the unadjusted odds ratio [95% CI] was calculated. Because of small cell sizes, categorical data from only the hyperthyroid and nonhyperthyroid groups were compared and, as there was invariably 1 cell in the 2 × 2 table that had an expected number <5, Fisher's exact method was used. The P-values in the univariable analysis are not corrected for multiple comparisons because data were further tested in the multivariable model to adjust for confounding. Baseline data associated with hyperthyroidism at annual follow-up (P < .2) then were tested in a multivariable model. Multivariable logistic regression was used to evaluate the adjusted effect of each risk factor. Variables were added to the model in a manual forward stepwise manner, starting with those variables found to have the highest significance in the univariable analysis.
Linear regression was used to determine whether there was a relationship between cat age and TSH concentration at baseline. A reference range for TSH was calculated nonparametrically using data derived from the study. The data set included all apparently healthy recruited cats (ie, those that did not have evidence of concurrent disease). The reference range was determined by ordering the data in ascending order and the limits of the reference range were set at 2.5–97.5% of the ranked data.
Receiver operating characteristic (ROC) curve analysis was used to determine the optimum cut-off point for TSH concentration when used as a predictor for the diagnosis of hyperthyroidism at first annual follow-up. The sensitivity and specificity of this cut-off along with the area under the curve (AUC) and 95% confidence intervals are presented.
Using the results of the ROC curve analysis, recruited cats were divided into 2 groups dependent on their baseline TSH concentration. Using all available follow-up for recruited cats, Kaplan–Meier time-to-event analysis was performed using diagnosis of hyperthyroidism as the defining event. Comparison of the time-to-event curves of the 2 groups was performed by log rank analysis.
One hundred ten cats were recruited into the study of which 6 (5.8%) cats were excluded because of a diagnosis of hyperthyroidism either at recruitment (n = 4) or within 3 months (n = 2), giving a prevalence of hyperthyroidism of 6% in recruited geriatric cats that the owners had perceived to be healthy. All excluded cats had tT4 concentration >45 nmol/L (range, 46.8–84.8 nmol/L) and TSH concentration <0.03 ng/mL at recruitment.
The median age at entry into the study of the 103 cats with known age was 12.1, [25th, 75th percentiles; 10.3, 14.0] years. Age was unknown in 1 cat that had been acquired as an adult stray. The predominant breeds were domestic short-haired (n = 70) and long-haired (n = 13) cats; however, the following breeds also were represented: Persian (n = 9), British Short-Haired (n = 4), Burmese (n = 3), Bengal (n = 2), Russian Blue (n = 1), Exotic Short-Haired (n = 1), and British Blue (n = 1). There were 47 male and 56 female cats.
The most common historical and clinical findings at recruitment were moderate to severe dental disease (32/104), chronic infrequent vomiting (including vomiting of hairballs; 30/104), palpable thyroid goiter (27/104), polydipsia (11/104), heart murmur without evidence of congestive heart failure (8/104), skin disease including flea infestation (7/104), upper respiratory tract or ocular disease (chronic clear ocular discharge, sneezing or conjunctivitis; 6/104), and previous history of feline lower urinary tract disease (3/104). After diagnostic testing, 11 cats were found to be mildly azotemic (IRIS stage II), 6 cats had hypertension, and 1 cat had diabetes.
Short-term (up to 14 months) Follow-up
This study examined the association of baseline variables with subsequent diagnosis of hyperthyroidism within 14 months. A total of 19/104 cats were classified as LTFU (of which 3 had died) and the remaining 85 cats were divided into those cats that were diagnosed with hyperthyroidism (hyperthyroid group: n = 11) and those that were not diagnosed with hyperthyroidism (nonhyperthyroid group: n = 74). The median baseline tT4 concentrations (Table 1) were not significantly higher (P = 0.13) in those cats that became hyperthyroid (25.7 [25th, 75th percentiles; 20.6, 30.1] nmol; n = 11) than those that did not become hyperthyroid (21.6 [25th, 75th percentiles; 17.6, 26.9] nmol; n = 74). However, median baseline TSH concentrations were significantly lower (P < 0.001) in the hyperthyroid group (<0.03 [25th, 75th percentiles; <0.03, <0.03] ng/mL) than the nonhyperthyroid group (0.04 [25th, 75th percentiles; 0.03, 0.08] ng/mL). TSH concentrations were not related to age at baseline (R2 = 0.008; P = 0.36).
|Variable||Data at Entry into Study|
|Nonhyperthyroid Group||n||Hyperthyroid Group||n||Pa||LTFU||n||Pb|
|Age (years)||12.2 [9.9, 14.0]||73||12.1 [10.8, 13.7]||11||.51||11.7 [11.0, 14.7]||19||.75|
|Breed (n, % pure breed)||17 (23%)||74||0 (0%)||11||.11||4 (21%)||19||NC|
|Sex (n, % female)||41 (55%)||74||3 (27%)||11||.11||12 (67%)||18||NC|
|Weight (kg)||4.7 [3.9, 5.4]||73||4.5 [3.9, 6.0]||11||.61||4.0 [3.6, 5.3]||19||.34|
|Vomiting (n, % yes)||15 (20%)||74||5 (46%)||11||.12||10 (53%)||19||NC|
|Goiter (n, % yes)||16 (22%)||74||6 (55%)||11||.03||5 (26%)||19||NC|
|Murmur (n, % yes)||2 (3%)||74||2 (18%)||11||.08||4 (21%)||19||NC|
|HR (bpm)||180 [172, 196]||73||184 [150, 220]||10||.96||180 [176, 210]||19||.86|
|SBP (mmHg)||140 [125, 155]||73||140 [121, 165]||11||.68||144 [128, 172]||19||.67|
|tT4 (nmol/L)||21.6 [17.6, 26.9]||74||25.7 [20.6, 30.1]||11||.13||19.0 [15.2, 24.4]||19||.11|
|TSH (ng/dL)||0.04 [0.03, 0.08]||74||<0.03 [<0.03, <0.03]||11||<.001||0.04 [<0.03, 0.06]||19||.001|
|Creatinine (μmol/L)||137 [121, 160]||74||145 [132, 176]||11||.23||127 [114, 142]||19||.11|
|USG||1.052 [1.034, 1.060]||68||1.050 [1.039, 1.057]||10||.52||1.042 [1.028, 1.060]||15||.58|
|ALKP (IU/L)||26 [20, 33]||74||33 [27, 61]||11||.02||28 [23, 41]||19||.04|
|ALT (IU/L)||54 [45, 74]||74||57 [50, 82]||11||.41||58 [46, 84]||19||.58|
|Potassium (mmol/L)||3.9 [3.7, 4.2]||74||4.00 [3.5, 4.3]||11||.93||4.0 [3.8, 4.5]||19||.49|
|Phosphate (mmol/L)||1.3 [1.2, 1.5]||71||1.4 [1.2, 1.4]||10||.70||1.4 [1.3, 1.6]||19||.75|
|PCV (%)||38 [35, 40]||73||38 [35, 42]||11||.69||36 [33, 40]||18||.42|
Table 1 shows details of the baseline variables examined for an association with the diagnosis of hyperthyroidism within 14 months, with cats that were LTFU analyzed separately from the hyperthyroid and nonhyperthyroid groups. Cats in the hyperthyroid group were significantly more likely than nonhyperthyroid group cats to have a goiter at baseline (OR 3.5, 95% CI [1.0–12.7]; P = .03), more likely to have a murmur (P = .08), less likely to be purebred (P = .11), more likely to be male (P = .11), more likely to have a history of vomiting (P = .12), and had higher ALKP activity (P = .02). These variables were tested in the multivariable model along with tT4 and TSH concentrations. TSH concentration was treated as a binary variable with cats divided into 2 groups with baseline TSH <0.03 ng/mL or with TSH ≥0.03 ng/ml. Only the binary variable TSH concentration was an independent variable predictive for the diagnosis of hyperthyroidism within 14 months with an unadjusted odds ratio of 39 (OR 39, 95% CI 5–331; P < .001).
Cats with undetectable TSH concentration (<0.03 ng/mL) at baseline (n = 29) were significantly more likely than cats with detectable TSH concentration (n = 75) to be diagnosed with hyperthyroidism within 14 months (Table 2; P < .001). Of the 25 cats with undetectable TSH concentration at baseline and at least 9 months of follow-up, 40% (10/25) were diagnosed with hyperthyroidism within 14 months whereas only 1.4% (1/75) cats with detectable TSH concentration at baseline became hyperthyroid during the same period. The results of the ROC curve analysis indicated that TSH concentration was a significant predictor for the diagnosis of hyperthyroidism and showed that the optimum cut-off TSH concentration was <0.03 ng/mL (AUC 0.83, 95% CI 0.70–0.97; P < .001). The sensitivity and specificity (with 95% CI) for this cut-off were 91 (62–98)% and 80 (70–87)%, respectively.
|Baseline TSH (ng/mL)||Thyroid Status 9–14 Months Following Recruitment|
|Hyperthyroid TSH <0.03 (ng/mL)||Euthyroid TSH <0.03 (ng/mL)||Euthyroid TSH ≥0.03 (ng/mL)||LTFU|
The derived reference range for feline TSH concentrations using the DPC Immulite chemiluminescent canine TSH assay in the healthy cats ≥9 years (n = 90) was 0−0.15 ng/mL.
Long-term (up to 4.5 years) Follow-up
At the end of the study period (June 2009), recruited cats had been followed for a median of 784 [25th, 75th percentiles; 363, 1092] days (26 months). Cats were censored for the Kaplan–Meier analysis either because they died (n = 21), were LTFU (n = 24), or reached the study end-point without being diagnosed with hyperthyroidism (n = 42). In most cases, the cause of death or euthanasia was not confirmed but was attributed to gastrointestinal disease (n = 4), CKD (n = 4), nasal mass (n = 2), respiratory distress (n = 2), incontinence or bladder neoplasia (n = 2), cerebrovascular accident (n = 1), and autoimmune disease (n = 1), or the cause of death was unknown in 5 cases.
In total, 17/104 cats were diagnosed with hyperthyroidism after recruitment including 11 cats diagnosed within 14 months that were included in the short-term follow-up data analysis and an additional 6 cats that were diagnosed during the subsequent period of follow-up. In 13 cats, hyperthyroidism was diagnosed by documentation of an increased tT4 concentration, 3 cats were diagnosed by measurement of tT4 and fT4 and 1 cat was diagnosed by T3 suppression test. Using the median follow-up of 26 months, an annual incidence of hyperthyroidism of 7.4% per year was calculated in this population.
All 17 cats diagnosed with hyperthyroidism during the study period had multiple clinical signs consistent with this diagnosis including weight loss (15/17), vomiting (10/17), diarrhea (1/17), polyphagia (3/17), polydipsia (7/17), tachycardia (heart rate ≥200 bpm; 5/17), heart murmur (8/17), and above reference range activity of ALT, ALKP, or both (11/17). Of these 17 cats, 5 cats had been diagnosed with CKD at baseline and 3 cats had been diagnosed with and treated for hypertension at the time of recruitment (n = 1) or between recruitment and diagnosis of hyperthyroidism (n = 2). Five hyperthyroid cats had thyroidectomy performed and had evidence of adenomatous hyperplasia or adenoma on histopathology, with a median histopathologic grade of 5 [range 3–5].
Of the 17 cats that became hyperthyroid, 13 cats had undetectable TSH concentration (<0.03 ng/ml) at baseline. In the 4 cats diagnosed with hyperthyroidism that had detectable TSH concentration at baseline, TSH had become undetectable within 12 months of recruitment and subsequent diagnosis of hyperthyroidism occurred within 6–28 months of the documentation of undetectable TSH.
Results of Kaplan–Meier time-to-event analysis yielded a median estimated time to diagnosis of hyperthyroidism of 959 days [95% CI, 212–1705 days] for cats with TSH <0.03 ng/mL at baseline. Cats with undetectable TSH had a significantly higher risk of developing hyperthyroidism during follow-up than did cats with detectable TSH concentration (Fig 1; P < .001).
This prospective study provides evidence that an undetectable TSH concentration (< 0.03 ng/mL) in euthyroid cats is associated with an increased risk for the subsequent diagnosis of hyperthyroidism. Overall, 17/104 cats became hyperthyroid during the study period. The large proportion of cats diagnosed with hyperthyroidism in this prospective study (annual incidence of 7.4%) confirms that hyperthyroidism is very common among older cats in London, UK. These results are similar to estimated annual incidence figures for cats ≥9 years calculated retrospectively from the Beaumont Animal Hospital database during the same period (2004–2007; 9.0%).
The criteria used for the diagnosis of hyperthyroidism after recruitment to this study were relatively stringent. However, recently published diagnostic criteria suggest that a lower cut-off for tT4 concentration (approximately 30 nmol/L) could have been used to trigger the measurement of fT4 in cats with concurrent illness that were clinically suspected to be hyperthyroid. In this study, cats with concurrent illness only were considered to be hyperthyroid if the tT4 concentration was > 40 nmol in association with a fT4 concentration of > 40 pmol/L. Conversely, it could be argued that the criteria for exclusion of cats at recruitment, with regard to the diagnosis of hyperthyroidism, were not sufficiently stringent to exclude cats with occult or mild hyperthyroidism. However, only 2 of 17 cats that became hyperthyroid during the study had an initial tT4 concentration between 30–40 nmol/L and none had a tT4 concentration ≥40 nmol/L at recruitment, therefore the application of more stringent exclusion criteria at recruitment or during follow-up would be unlikely to have changed the results of the study.
The population tested was selected by examining cats that presented for routine health screening but cats that were found to have concurrent illness at the time of recruitment were not excluded. This is because restriction of the recruited population to those cats without detectable concurrent disease would have inhibited recruitment in this age group. Such a restriction also may have resulted in a population not representative of cats at risk of developing hyperthyroidism, as hyperthyroid cats commonly have concurrent disease, especially CKD and hypertension. By selecting cats presenting for routine health screening we hoped to avoid recruiting cats with moderate to severe concurrent disease that might have led to occult hyperthyroidism. However, owners vary considerably in their appreciation of health in geriatric cats as exemplified by the fact that a number of hyperthyroid cats and a cat with diabetes were recruited.
The unadjusted odds ratio that an undetectable baseline TSH concentration in an apparently euthyroid cat will result in a diagnosis of hyperthyroidism within 14 months was very high (39). However, this OR had a very broad 95% CI (4–323) suggesting a low confidence in the strength of the effect. This low confidence is likely related to the small sample size; however, even the low end of the range (OR of 4) demonstrates a strong risk that a euthyroid cat with TSH concentration <0.03 ng/mL will become hyperthyroid.
The DPC chemiluminescent canine TSH assay used in these studies has a suboptimal sensitivity when used for the measurement of feline TSH. Almost one third of recruited cats had undetectable TSH concentration at baseline and only 13/25 (52%) of followed cats with TSH concentration < 0.03 ng/mL at baseline were diagnosed with hyperthyroidism during the study period. Despite this, the specificity of a single undetectable TSH concentration as a predictor for the diagnosis of hyperthyroidism within 14 months in recruited cats was 80%. Some cats with persistently undetectable TSH concentration took longer than 14 months to become hyperthyroid and this may be an indicator of a period of subclinical hyperthyroidism that varies in length from cat to cat. As discussed above, cats with concurrent disease were not excluded from the study. There is some evidence that the decrease in tT4 concentration in sick euthyroid people might in some cases be mediated by decreased TSH concentrations, especially in the critically ill. However, a previous study in cats did not demonstrate a difference in TSH concentrations between euthyroid cats with CKD and euthyroid geriatric cats.
In humans, subclinical hyperthyroidism is defined as a subnormal TSH concentration in a patient with normal tT4 or fT4 concentration. In our study, however, some cats that were not diagnosed with hyperthyroidism had TSH concentrations that seemed to fluctuate in and out of the detectable TSH range over time and these cats may not have had any derangement of their thyroid function. The limited sensitivity of the DPC canine TSH assay and the demonstration that some euthyroid cats have intermittently undetectable TSH concentration means that it is not possible to assert that all euthyroid cats with a single undetectable TSH concentration have subclinical hyperthyroidism as hypothesized for this study. It is possible that cats with persistently undetectable TSH concentration may be more likely to have subclinical hyperthyroidism than cats with a single undetectable TSH concentration. It would therefore be helpful to determine the prognostic value of repeated TSH measurements perhaps in 2 samples spaced 1 month apart as this may improve the specificity of the test as a predictor for the diagnosis of hyperthyroidism. Finally, it should be borne in mind that not every cat with subclinical hyperthyroidism, however we define it, will eventually become clinically hyperthyroid. Indeed, this is the case in humans where subclinical hyperthyroidism is thought to persist for decades in some patients with nodular thyroid disease.
Although TSH secretion is known in humans[39, 40] and other species[41, 42] to have a marked circadian rhythm, there are limited data regarding the existence of a circadian rhythm of TSH secretion in cats. A pilot study in 6 cats, from which TSH concentration was measured every 2 hours for 24 hours, did not detect a predictable circadian rhythm of TSH secretion, although measured TSH concentrations did fluctuate in some cats. All blood samples in the present study were drawn at a similar time of day (mornings) and therefore the effects of any circadian rhythm on TSH concentration, if it exists in cats, should have been minimal.
Baseline tT4 concentrations were not significantly different between cats that were diagnosed with hyperthyroidism within 14 months and those that were not. However, cats with tT4 concentration >35 nmol/L and TSH concentration <0.03 ng/mL at any point in the study usually were diagnosed with hyperthyroidism within 3–9 months. In addition, cats in this study with a tT4 concentration >35 nmoL/L and a detectable TSH concentration (≥0.03 ng/mL) did not become hyperthyroid including 1 obese cat with an initial tT4 concentration of 44.6 nmol/L and TSH concentration of 0.11 ng/mL at baseline that was followed for more than 2 years. It has been previously demonstrated that obesity in healthy young cats increases fT4 concentration without decreasing TSH concentration, and this phenomenon could be related to a thyroid resistance syndrome linked to obesity in people.
In this study, only 1 (1/75) cat that had a detectable TSH concentration at baseline was diagnosed with hyperthyroidism within 14 months, and all 17 cats diagnosed with hyperthyroidism had undetectable TSH concentration before diagnosis. In 1 study, TSH concentrations had 100% sensitivity when used for the diagnosis of occult hyperthyroidism in cats. If cats invariably have a TSH concentration < 0.03 ng/mL both before and at the time of diagnosis of hyperthyroidism, this suggests that a detectable TSH concentration in a geriatric cat could be used to rule out a diagnosis of hyperthyroidism. Therefore, the results of this study provide support for the use of TSH concentration as a diagnostic tool in cats especially in combination with tT4 concentration. In the future, should a more sensitive feline TSH assay become available that can reliably differentiate between cats with subclinical or overt hyperthyroidism and euthyroid cats, the measurement of feline TSH may become the test of choice for the diagnosis of thyroid disease, as it has in the field of human medicine.
The presence of goiter at baseline was significantly associated with the diagnosis of hyperthyroidism at 1 year although goiter was only noted in 6 of 11 cats. Different veterinarians examine cats from week to week at the RVC feline research clinics and a standard palpation technique was not used for assessment of goiter. Therefore, the recording of a goiter at baseline in the clinical record was likely to be not as consistent or precise as it would have been had it been assessed by a single observer using a standard technique.
There is debate in human endocrinology regarding the physiological effects of subclinical hyperthyroidism and the indications for treatment. In the first (short-term) data analysis, baseline ALKP activity was higher in the hyperthyroid group than in the nonhyperthyroid group, albeit in most cases the ALKP activity still was well within the reference range. This suggests that subtle changes may be present in cats before the diagnosis of overt hyperthyroidism. However, none of the other signs associated with hyperthyroidism in cats, such as baseline weight, tachycardia, hypertension or presence of a murmur, were found to differ between the hyperthyroid and nonhyperthyroid groups. Unlike in human patients, atrial fibrillation is not associated with hyperthyroidism in cats but sinus tachycardia is a common finding, particularly in cases with markedly increased tT4 concentrations. Therefore, lack of evidence of tachycardia in our study is of particular note.
Further limitations of this study include the lack of a gold standard for the diagnosis of hyperthyroidism and the relatively small number of cats that were recruited. In studies of humans with subclinical hyperthyroidism, thousands of people are often recruited.[24, 25, 44, 45] Therefore in comparison, the power of this study to detect differences between groups is very low, especially when the risk factors do not have a strong effect.
In conclusion, in this study, an undetectable TSH concentration in a euthyroid geriatric (≥9 years) cat identified increased risk for subsequent diagnosis of hyperthyroidism. Conversely, a TSH concentration of ≥0.03 ng/mL suggests that a cat is less likely to develop clinical hyperthyroidism within the subsequent year. Therefore, the measurement of TSH concentration in healthy geriatric cats could be used as a biomarker or screening test to determine the risk of a future diagnosis of hyperthyroidism.
Wakeling J, Elliott J, Syme HS. Does subclinical hyperthyroidism exist in cats? J Vet Intern Med 2006;20(3):726 (abstract)
Idexx Laboratories, Wetherby, UK
Royal Veterinary College Diagnostic Laboratories, North Mymms, Hatfield, UK
SPSS 10.0 for Windows. SPSS Inc, Chicago, Illinois
The authors thank all the staff and clients of the People's Dispensary for Sick Animals and the Beaumont Animal Hospital; all the staff at the Royal Veterinary College feline research group but especially Timothy Lee Williams MRCVS, Dr Rosanne Jepson PhD, MRCVS and Kim Souttar RVN.
Work supported in part by Beryll Evetts and Robert Luff Fellowship in Feline Research and Petplan Charitable Trust.
- 6High incidence of multinodular toxic goiter in the elderly population in a low iodine intake area vs high incidence of Graves disease in the young in a high iodine intake area - Comparative surveys of thyrotoxicosis epidemiology in East Jutland Denmark and Iceland. J Intern Med 1991; 229:415–420., , , et al.
- 14Adenomatous hyperplasia of the thyroid gland is related to TSH concentration in cats. J Vet Intern Med 2007;20(6):1522. abstract., , , et al.
- 17Thyroid adenomatous hyperplasia in euthyroid cats. J Vet Intern Med 1999;13:242. abstract., , , et al.
- 35Staging chronic kidney disease. In: Elliott J, Grauer FG, ed. BSAVA Manual of Canine and Feline Nephrology and Urology, 2nd ed. India: Replika Press Pvt Ltd; 2007:159–166..
- 37The etiopathogenesis of feline hyperthyroidism. PhD Thesis. Royal Veterinary College. London, UK: London University; 2008:242..
- 38Regulation of thyrotropin secretion. In: Braverman LE, Utiger RD, ed. Werner & Ingbar's The Thyroid: A Fundamental and Clinical Text. Philadelphia, PA: Lippincott Williams & Williams; 2005:197–213..
- 46Modern Epidemiology, Second ed. Philadelphia PA: Lippincott Williams & Wilkins; 1998., .