S P E C I A L
F E A T U R E
C l i n i c a l
R e v i e w
Is an Isolated TSH Elevation in Chronic Nonthyroidal Illness “Subclinical Hypothyroidism”? Elaine M. Kaptein, Jonathan S. LoPresti, and Matthew J. Kaptein Department of Medicine, University of Southern California, Los Angeles, California 90033
Context: Elevated TSH with normal T4 frequently occurs with chronic kidney, liver, and heart diseases. Whether isolated TSH elevations represent mild thyroid gland failure has not been established. Evidence Acquisition: PubMed was searched for longitudinal studies in chronic heart, liver, or kidney disease documenting persistent isolated TSH elevations or progression to overt hypothyroidism. Evidence Synthesis: Four articles met inclusion criteria. In 16 end-stage renal failure patients, four had isolated TSH elevations. All normalized within 14 months. In 452 systolic heart failure patients, 20 had isolated TSH elevations, five of 20 were persistent, and none progressed to overt hypothyroidism within 6 months. In 207 untreated chronic hepatitis C patients, 12 had isolated TSH elevations and four had increased TSH with reduced free T4; all were female, and 14 had positive antithyroid antibodies. After 1 year, two of 12 developed “clinical hypothyroidism.” In 72 chronic hepatitis C patients, nine females had positive antithyroid antibodies. Two antibody-negative patients had TSH 5– 6 mU/L with reduced free T4. After 1 year, three of four with positive antithyroid antibodies and baseline TSH ⬍ 4 mU/L had elevated TSH with reduced free T4. Conclusions: In chronically ill patients, there is inadequate evidence to determine: 1) that isolated TSH elevations usually persist or progress to overt hypothyroidism; 2) the etiology and clinical significance of isolated TSH elevations; and 3) whether levothyroxine therapy is indicated for persistent isolated TSH elevations. Thus, isolated TSH elevations in chronic renal, cardiac, or liver diseases have not been documented to indicate mild thyroid gland failure. (J Clin Endocrinol Metab 99: 4015–4026, 2014)
C
riteria for diagnosing mild thyroid gland failure (“subclinical hypothyroidism”) in the general population have been arbitrarily applied to patients with chronic nonthyroidal illnesses (1–11). Mild thyroid gland failure is defined as TSH levels persistently between 4.5 and 20 mU/L, with normal T4 estimates, or progression to overt hypothyroidism (12–15). An increased TSH level is a sensitive indicator of mild thyroid gland failure and decreased free T4 negative feedback only if the hypothalamic-pituitary-thyroid-gland-peripheral-tissue axis is intact and serum TSH and free T4 concentrations reflect bioactive TSH and in vivo free T4 (15).
These assumptions may not apply in nonthyroidal illnesses (14 –16). Isolated TSH elevations in chronic nonthyroidal illnesses have been attributed to mild thyroid gland failure without substantiating evidence, and patients have been arbitrarily treated with levothyroxine (LT4) therapy with variable outcomes (5, 8, 9). The TSH to free T4 relationship is altered in ambulatory elderly and the obese, which may provide clues to alterations in chronic nonthyroidal illness. Leptin, dopamine, glucocorticoids, and cytokines can affect TSH independently of free T4 concentrations without affecting thyroid status (17). In many obese patients, TSH is modestly elevated with normal free T4 values (18), likely resulting from
ISSN Print 0021-972X ISSN Online 1945-7197 Printed in U.S.A. Copyright © 2014 by the Endocrine Society Received March 25, 2014. Accepted August 5, 2014. First Published Online August 28, 2014
Abbreviations: BMI, body mass index; CHD, coronary heart disease; CHF, congestive heart failure; ESRD, end-stage renal disease; L-T4, levothyroxine; TBG, T4-binding globulin; TPO, thyroid peroxidase.
doi: 10.1210/jc.2014-1850
J Clin Endocrinol Metab, November 2014, 99(11):4015– 4026
jcem.endojournals.org
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 December 2014. at 11:16 For personal use only. No other uses without permission. . All rights reserved.
4015
4016
Kaptein et al
Table 1.
Increased TSH in Chronic NTI
J Clin Endocrinol Metab, November 2014, 99(11):4015– 4026
Elements of Systematic Review of Elevated TSH in Chronic Nonthyroidal Illnesses
Questions posed regarding patients with chronic nonthyroidal illnesses 1. What evidence indicates that mild thyroid gland failure is present with isolated TSH elevations? a. Do TSH elevations persist or progress over time, with reductions in free T4 values? 2. What is the underlying etiology of the isolated serum TSH elevations? 3. Do elevated in vitro TSH levels represent bioactive TSH? 4. What criteria can be used to define mild thyroid gland failure? 5. What are the clinical associations with isolated TSH elevations? 6. Is there evidence that L-T4 therapy is indicated to treat persistent isolated TSH elevations and has an acceptable risk-tobenefit ratio? 7. What are the criteria to diagnose overt hypothyroidism? a. Are free T4 immunoassays reliable in patients with chronic nonthyroidal illnesses? 8. What are TSH goals when L-T4 treatment is given for overt hypothyroidism with chronic illness? Inclusion criteria Studies of patients with chronic kidney, liver, or heart disease who had: 1. Elevated serum TSH and normal total/free T4 with follow-up for persistent isolated TSH elevation or progression to overt hypothyroidism. 2. Documentation of effects of L-T4 therapy for mild or overt thyroid gland failure on signs and symptoms, or TSH and T4 levels. Exclusion criteria 1. Studies with TSH and T4 follow-up of ⬍3 months. 2. Studies without a T4 or free T4 estimate value. 3. Patients without chronic nonthyroidal illnesses. 4. Patients with acute nonthyroidal illnesses. 5. Patients receiving medications known to induce thyroid gland failure or alter serum TSH and T4 concentrations, including dopamine or somatostatin analogs, phenytoin, carbamazepine, sertraline, iodine, iodine containing medications including amiodarone, lithium, interferon-␣, glucocorticoids, rifampin, and sunitinib (15, 16, 29, 30). 6. Patients with known thyroid disease. 7. Studies where L-T4 therapy was initiated before monitoring for persistence or progression of TSH or T4 measurements.
obesity, not causing it (17–20). The ambulatory elderly have a high prevalence of isolated TSH elevations, not due to mild thyroid gland failure (21–23), but secondary to altered hypothalamic-pituitary-thyroid set point, down-regulation of hepatic and pituitary type 2 deiodinase activity, reduced TSH bioactivity, concurrent nonthyroidal illnesses, medications, or laboratory artifacts (24 –28). We undertook a systematic literature review to address the questions posed in Table 1.
Methods Evidence acquisition We searched PubMed for the following terms: subclinical hypothyroidism, hypothyroidism, TSH, T4, chronic kidney disease, chronic heart disease, chronic heart failure, coronary heart disease, chronic liver disease, longitudinal studies, and L-T4 therapy. The search was limited to reports in English published from 1946 to May 23, 2014, and supplemented by personal files and bibliographies of relevant articles. Longitudinal studies of patients with chronic renal, liver, or cardiac disease were reviewed for data indicating frequency of persistent isolated TSH elevations, progression of TSH elevations with reduced T4 estimates, and clinical associations (Table 1). Response of TSH and T4 values and clinical parameters following L-T4 therapy were assessed.
Results We reviewed 351 articles from the literature search. No randomized controlled trials addressed any of the questions posed. Only four articles described observational studies that had longitudinal data in chronically ill patient cohorts matching our inclusion criteria (6, 31–33). Information in these studies was compared to published data for the general population, ambulatory elderly, and patients with obesity. Definitions of mild thyroid gland failure and overt hypothyroidism in the general population Mild thyroid gland failure is defined as a persistently isolated TSH elevation or a progression to overt hypothyroidism (13–15). To diagnose an isolated TSH elevation as mild thyroid gland failure, the hypothalamic-pituitarythyroid-peripheral-tissue axis must be intact, and other causes of TSH elevations must be excluded, such as nonthyroidal illnesses or medications (15, 16, 34). Symptoms and signs of hypothyroidism are nonspecific and nondiagnostic and should not be used to define mild thyroid gland failure (16, 30). An isolated elevation of TSH with a normal total/free T4 value has been reported in 4 to 10% of the general population not on thyroid hormone therapy (Table 2) (1,
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 December 2014. at 11:16 For personal use only. No other uses without permission. . All rights reserved.
doi: 10.1210/jc.2014-1850
2, 12–15, 35). TSH elevations normalize in 50% of these individuals over time (23). TSH levels should be re-evaluated after 3 to 6 months to rule out laboratory error or transient increases (13, 14) and to determine whether progressive increases in TSH levels in association with reduced free T4 estimates have occurred, indicating development of overt hypothyroidism. TSH assays are poorly harmonized among laboratories, and values may be higher due to assay variability rather than to thyroid gland failure (36). In the general population, L-T4 therapy has been recommended for persistent TSH ⬎10 mU/L (16), whereas L-T4 treatment with TSH ⬍ 10 mU/L remains controversial (13–16). Progression to overt hypothyroidism— defined as a persistently increased TSH, usually ⬎10 IU/mL, with a reduced free T4 estimate— has been reported to occur in 2 to 5% of the general population per year (13–15) (Table 2). Those at the highest risk for progression to overt hypothyroidism are elderly, female, with positive anti-thyroid peroxidase (TPO) antibodies, TSH ⬎ 10 mU/L (13– 15, 35) (Table 2), or thyroid hypoechogenicity by ultrasonography (37). Positive anti-TPO antibodies usually indicate autoimmune thyroid disease. Antithyroid antibodies and ultrasound abnormalities do not indicate thyroid gland failure. Elevated TSH values in the elderly Basal TSH levels may be elevated in 3 to 15% of the elderly (21–23, 38) (Table 2). In one study of 5182 elderly patients, 3.3% had isolated TSH elevations at baseline, which persisted in 54% at 6 months, with none having reduced free T4 estimates (38). In another study of 3594 community-dwelling patients over age 65, 12.8% had isolated TSH elevations at baseline (23). After 2 and 4 years, 56% (208 of 369) and 75% (121 or 161) had persistent TSH elevations, respectively, whereas 2% (eight of 369) and 2% (four of 208) had an elevated TSH with a reduced free T4 estimate or TSH ⬎ 20 mU/L, respectively (23). Isolated TSH elevations in community-dwelling elderly have not been shown to contribute to decreased functional capacity (38). The TSH to free T4 relationship is frequently altered in the ambulatory elderly due to changes independent of thyroid gland failure (14, 21, 28, 39). Two cohort studies have shown that aging is associated with an increase in TSH without a fall in free T4 (26, 27). A large cross-sectional study showed that at any given free T4 value within the reference range, TSH was higher in older people than younger people, suggesting that older people will be diagnosed with mild thyroid gland failure more commonly than younger people despite having identical free T4 estimate values (24). Consequently, a diagnosis of mild thy-
jcem.endojournals.org
4017
roid gland failure in the elderly cannot be accurately established using criteria from the general population because increased TSH may be due to causes unrelated to thyroid gland failure. Elevated TSH levels in patients with obesity In 21 097 patients with body mass index (BMI) of ⱖ30 kg/m2, 2.2% had isolated TSH elevations, and 0.7% had elevated TSH levels with reduced free T4 estimates (40) (Table 2). Isolated TSH elevations were reported in 6% of morbidly obese subjects, compared to 4% of lean subjects, with anti-TPO antibodies present in 17% of obese and 8% of lean subjects (18). Elevated leptin levels may have a stimulatory effect on TSH secretion, resulting in increased circulating TSH concentrations (20). In a retrospective study of 86 morbidly obese patients undergoing gastric bypass or banding, isolated TSH elevations were present in 10.5% before surgery and in 0% 6 months after surgery, in association with a mean BMI decrease from 49 to 32 kg/m2 (41). Normalization of TSH with weight loss suggests that obesity per se plays an integral role in TSH regulation, independent of free T4 concentrations (17, 19, 20). Isolated TSH elevations in patients with chronic nonthyroidal illnesses A high frequency of elevated TSH with normal total or free T4 estimates is observed with chronic kidney diseases (1.6 to 28%) (3, 4, 10, 11, 34) and congestive heart failure (CHF) (6 to 16%), unrelated to medications known to induce thyroid gland failure (7, 42) (Table 2). In a small study of patients with chronic hepatic cirrhosis, 38% had elevated TSH by RIA and normal free T4 by tracer equilibrium dialysis (43). Isolated TSH elevations were reported in 0 to 6% with untreated chronic hepatitis C (32, 33, 44, 45). In 134 patients with chronic hepatitis C and 41 with chronic hepatitis B, 3.7% with hepatitis C and 0% with hepatitis B had isolated TSH elevations, whereas TPO antibodies were positive in 20% and 5%, respectively (45). Untreated hepatitis C-positive patients have a 1.6-fold increased risk for thyroid autoimmune disorders, even without cirrhosis, compared to patients with sera negative for hepatitis C virus antibody, those with chronic hepatitis B infection, and healthy subjects (46 – 48). What evidence indicates that mild thyroid gland failure is present with elevated TSH and normal free T4 values in patients with chronic nonthyroidal illnesses? In most studies, TSH and free T4 values were not followed over time, and symptoms and signs to indicate hy-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 December 2014. at 11:16 For personal use only. No other uses without permission. . All rights reserved.
4018
Kaptein et al
Increased TSH in Chronic NTI
J Clin Endocrinol Metab, November 2014, 99(11):4015– 4026
Table 2. Review of TSH Elevations With or Without Free or Total T4 Reductions in the General Population, Elderly, Obese, and With Chronic Nonthyroidal Illnesses First Author, Year (Ref.)
Clinical Characteristics
Study Design
TSH, mU/La
Total T4/Free T4 Assayb
Frequency of Increased TSH/Normal T4 Values
General population Hollowell, 2002 (35)
NHANES III US general population
Cross-sectional, n ⫽ 13 344
⬎4.6
Total T4, 4.5–13.2
4.3%
Surks, 2004 (13)
General population
Extensive literature review
Variable
4 to 8.5%
Lo, 2005 (1)d
NHANES III
Cross-sectional, n ⫽ 14 623
Mild, 4.5–10; severe, ⬎10 ⬎4.5
Total T4 ⬎ 4.5
8%
General population
Extensive literature review
⬎4.5
Variable
4 to 10%
Adult outpatients
Cross-sectional, n ⫽ 3089
⬎4.5
Free T4
9.5%
Varied by study from community-dwelling adults to elderly worldwide General population
Meta-analysis, n ⫽ 55 287, 11 prospective studies Extensive literature review
⬎4.5 to ⬍20
T4 when available, various methods Variable
6.2% TSH, ⬎4.5 to ⬍20 (3.3 to 16.4%) 4 to 20%, (4.3 to 9.5% in US)
Community-dwelling ⬎65 y old, in Cardiovascular Health Study USA Community-dwelling ⬎65 y old, in Cardiovascular Health Study USA Well elderly
Prospective cohort study, n ⫽ 3233; 13 y follow-up Longitudinal study over 4 y, n ⫽ 3594 Review of the literature
⬎4.5 to ⬍20
Free T4
15%
⬎4.5 to ⬍20
Free T4
12.8%, persistent in 56% at 2 y
Free T4, various methods
15%
Pravastatin in the Elderly at Risk (PROSPER) studyf
Prospective, n ⫽ 5182; 6-mo follow-up
⬎Upper reference range, but ⬍10 ⬎4.5
Free T4
3.3%, persistent in 54% at 6 mo
Morbidly obese before and after gastric bypass or banding BMI ⬎ 30 kg/m2 from HUNT study Obese (mean BMI, 43 kg/m2) referred for workup to a tertiary care center
Retrospective, n ⫽ 86
⬎5.5
ND
Cross-sectional, n ⫽ 21 097 Cross-sectional, n ⫽ 165 obese, n ⫽ 118 lean
⬎4.0 ⬎4 to ⬍10
Free T4 Free T4
10.5% pre-op, and 0% at 6 mo post-op 2.2% 6% obese, 4% lean
CKD stage 2– 4
Retrospective, n ⫽ 129 not treated, n ⫽ 180 on L-T4
Free T4
ND
CKD stages 2– 4, treated with L-T4 for 24 mo
Retrospective, n ⫽ 113 before and after L-T4
⬎4.9 to 10, repeated within 3 mo ⬎4.9, repeated within 3 mo
Free T4
ND
ESRD
Literature review
⬎4.5
Total T4/free T4 index
10.5–12.5% ⬎ 4.5, 1% ⬎ 10
Kang, 2008 (3)
ESRD on PD ⬎3 mo (Korea)
Cross-sectional, n ⫽ 51
⬎5
Free T4
27.5%
Ng, 2012 (4)f
ESRD on PD ⬎ 6 mo (Taiwan)
Cross-sectional, n ⫽ 122
⬎4
Free T4
15.6%
ESRD diabetic patients in Germany
Prospective observational multicenter, n ⫽ 1000 Retrospective, n ⫽ 2715
4.1–15
Free T4
1.6%
⬎5.0/5.7 to ⬍10
Not assessed
12.9%
Literature review
4.1–15
Not documented
8.9 to 24.8%
Biondi, 2008 (15) Chonchol, 2008 (2)
d
Rodondi, 2010 (12) Cooper, 2012 (14) Elderly Cappola, 2006 (22) Somwaru, 2012 (23)e Visser, 2013 (21)
Virgini, 2014 (38) Obesity Chikunguwo, 2007 (41) Asvold, 2009 (40) Marzullo, 2010 (18) Chronic kidney disease pre-ESRD Shin, 2012 (5)
Shin, 2013 (8)
ESRD Kaptein, 1996 (34)
Drechsler, 2014 (10) f
⬎4.5; mild, 4.5– 9; severe, ⬎10
Rhee, 2014 (11)
ESRD on PD and hemodialysis (United States) ESRD
Kalk, 1980 (31)
ESRD on hemodialysis
Prospective crosssectional, n ⴝ 16, 14-mo follow-up
>8.0
Free T4 index
25%, all transient
CHF, United States and Europe
Six prospective cohorts, n ⫽ 25 390
4.5–19.9
Free T4, various methods
8.1% (range 5.5–16.2%)
Rhee, 2013 (7)
CHF
⬎4.6
Total T4
4.6%
Frey, 2013 (6)
Systolic HF
Retrospective NHANES III, n ⫽ 14 879 Prospective INH study, 6-mo follow-up, n ⴝ 452
>4.0
Free T4/free T3
4.5%, persistent in 25% (5 of 20) at 6 mo
“Euthyroid” hepatic cirrhosis
Cross-sectional, n ⫽ 23
⬎7.5 mg/dL
38%g
Borzio, 1983 (50)
Liver cirrhosis and chronic hepatitis
⬍5
Tran, 1993 (33)
Chronic Hep C
Cross-sectional, n ⫽ 33 and 22, respectively Prospective, n ⴝ 72, followed for 1 y
Equilibrium dialysis, ⬍1.8 ng/mL Total T4
>4
Free T4
0%
Rhee, 2013 (51)
Cardiac disease Gencer, 2012 (42)
Chronic liver disease Chopra, 1974 (43)
1.8%h
(Continued)
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 December 2014. at 11:16 For personal use only. No other uses without permission. . All rights reserved.
doi: 10.1210/jc.2014-1850
Table 2.
jcem.endojournals.org
4019
Continued
Frequency of Anti-TPO Abc ⴙve
Progression to Increased TSH/Low T4 Values
Risk Factors for Progression
Clinical Associations/ Recommendations
Frequency of Increased TSH/Decreased T4 Values
13%
ND
ND
0.3%
Frequent
2–5%/y ND
ND
ND
Elderly, females, TPO Ab ⫹ve, TSH ⬎20 ND
ND
ND
ND
Positive in 60 – 80%
Up to 4%/ y in women with TPO Ab ⫹ve
TPO Ab ⫹ve, iodine intake, TSH ⬎ 10–15
No response to L-T4 if TSH ⬍ 10, repeat TSH every 6 –12 mo 20 to 23% increased TSH with eGFR ⬍ 60 mL/min/1.73 m2 Monitor every 6 –12 mo, L-T4 therapy if TSH ⬎ 10 18% increased TSH with eGFR ⬍ 60 mL/min/1.73 m2 Increased CHD events and CHD mortality for TSH 10 –19.9 Treat if TSH ⬎10 mU/L
ND
55% with increased TSH Frequent
Female, increased age, TPO ⫹ve ⬎ age 65 y, TSH ⬎10, TPO Ab ⫹ve ND
ND
ND
ND
No increase in CHD, CVD, CV
1.6% had TSH ⬎ 20 or TSH ⬎ 4.5/low
death or all-cause death ND
free T4b 0.61% had TSH ⬎ 20 or TSH ⬎ 4.5/ low free T4b ND
4.3%/y
ND
2% at 2 y
TSH ⬎ 10 mU/L
ND
A few % per year
Higher TSH, TPO Ab, goiter
ND
0%
ND
ND ND ND ND ND
Female (64%)
Decreased or no change in mortality, repeat TSH in 3– 6 mo No decreased functional capacity
ND
ND
ND
ND
0%
ND 17% obese, 8% lean
ND ND
ND ND
ND ND
TSH ⬎ 4/low free T4 in 0.7% TSH ⬎ 4/low free T4b in 4% obese (1.8% TSH ⬎ 20), 0% lean
⫹ve 26/81 on L-T4, 0/59 no L-T4
No increase in TSH over 36 mo
None
Less decline in eGFR with L-T4 to TSH WNL over 6 mo
ND
⫹ve in 40%
ND
ND
Less decline in eGFR with L-T4 to TSH WNL in 64%, faster decline in 30%
ND
ND
All TSHs 10 –20 were transient within 1–2 wk ND
ND
Do not treat with L-T4
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Lower LVEF and fractional shortening Longer PD duration, lower BMI, higher EPO doses No increased CV events or allcause mortality over 4 y Higher mortality with increased TSH ND
TSH persistently ⬎ 20/low free T4b in 2.6 –9.5% 0%
TSH ⬎ 20/low total or free T4b in 2.6 –
ND
0%
ND
ND
5.4% 0%
ND
ND
ND
ND
ND
ND
ND
0%
ND
Increased incidence and recurrence of HF with TSH 10 –19.9 Increased mortality risk in those with CHF No impact on survival
ND
ND
ND
ND
0%
3.6%
ND
ND
ND
0%
12.5% all female
3 of 4 with normal basal TSH at 1 y
Females, ⴙve antithyroid antibodies
ND
Increased TSH 5.3 and 5.6 with low free T4b in 2 of9 (22%) at
⫹ve in 4 of 14 with TSH ⬎ 5 ND
0%
ND
ND TSH > 4.0 and decreased free T4b in 0%
baseline (Continued)
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 December 2014. at 11:16 For personal use only. No other uses without permission. . All rights reserved.
4020
Kaptein et al
Table 2.
Increased TSH in Chronic NTI
J Clin Endocrinol Metab, November 2014, 99(11):4015– 4026
Continued
First Author, Year (Ref.)
Clinical Characteristics
Study Design
TSH, mU/La
Total T4/Free T4 Assayb
Frequency of Increased TSH/Normal T4 Values
Marazuela, 1996 (32)
Chronic Hep C
>4
Free T4
5.8%
Fernandez-Soto, 1998 (45) Danilovic, 2013 (44)
Chronic Hep C and B
Prospective, n ⴝ 207, followed for up to 1 y pretreatment Prospective, n ⫽ 134 Hep C, n ⫽ 41 Hep B Cross-sectional, n ⫽ 103; controls n ⫽ 96
⬎4
Free T4
3.7% Hep C 0% Hep B
⬎4.5
Free T4
3.9%
Chronic Hep C
Abbreviations: NHANES III, Third National Health and Nutrition Examination Survey, WNL, within reference range; CKD, chronic kidney disease; ESRD, end-stage renal disease receiving dialysis therapy; PD, peritoneal dialysis; EPO, erythropoietin dose; LVEF, left ventricular ejection fraction; eGFR, estimated glomerular filtration rate using four-variable Modification of Diet in Renal Disease study (MDRD) equation; HF, heart failure; CV, cardiovascular; CVD, cerebrovascular disease; INH, Interdisciplinary Network Heart Failure study; HUNT, Nord-Trondelag Health Study; Hep, hepatitis; Ab, antibody; ⫹ve, positive; ND, no data. a Upper population reference range. Note: Many studies do not discriminate between TSH levels that are slightly (4 –10 mU/L) and more severely (⬎10 mU/L) elevated. b Free T4 estimates by direct analog immunoassays unless otherwise stated. Estimate values are method dependent and may be inappropriately low in patients with reduced serum binding protein concentrations and nonthyroidal illnesses (67, 68) c
Anti-TPO Ab indicates anti-TPO antibodies (indicates autoimmune thyroid disease).
d
Included in review by Rhee et al. (11), but additional information provided.
e
Included in review by Gencer et al. (42), but additional information provided.
f
Elderly population, mean age 75 y, with cardiovascular disease or at high risk of developing cardiovascular disease.
g
Blunted TSH response to exogenous TRH.
h
Basal TSH 14 mU/L with increased TSH response to TRH.
Articles that were prospective studies and satisfied inclusion criteria are in bold print.
pothyroidism were not reported. Very little evidence indicated that mild thyroid gland failure was present in chronically ill patients with elevated TSH but normal free T4 values. Only the four highlighted articles (in bold print) in Table 2 met our inclusion criteria. The first is a prospective cross-sectional study of patients hospitalized for systolic heart failure, of which 20 out of 452 had isolated TSH elevations, which normalized in 15 of 20 after 6 months (6). None developed reduced free T4 levels, and there was no association with overall survival (6). The second is a prospective cross-sectional study of 16 hemodialyzed end-stage renal disease (ESRD) patients, in which four had transient isolated TSH elevations, with none progressing over 14 months of follow-up (31). In a cross-sectional study of ESRD patients, 9% of 306 patients had isolated TSH levels between 5 and 10 mU//L, with 1% of 306 having TSH between 10 and 20 mU/L with normal free T4 index values (49). All TSH values between 10 and 20 mU/L decreased to ⬍ 10 mU/L when repeated within 1 to 2 weeks (49). This was not a longitudinal study, so persistence and progression of TSH elevations were not assessed. In the third prospective study, 72 patients with chronic hepatitis C were followed for 1 year (33). Nine of the 72 had positive antithyroid antibodies with normal TSH values. Two of nine with negative antibodies had TSH values of 5.3 and 5.6 mU/L, respectively, with reduced free T4
estimates. After 1 year, three of four patients with positive antithyroid antibodies and baseline TSH values ⬍4 mU/L demonstrated elevated TSH with reduced free T4 estimates. In the fourth prospective study, of 207 patients with untreated chronic hepatitis C followed for 1 year, 5.8% (all female) had isolated TSH levels, and 1.9% had elevated TSH with reduced free T4 levels at baseline (18). Eighty-eight percent with elevated TSH had positive thyroid antibodies. After 1 year, two of 12 patients with isolated TSH elevations developed “clinical hypothyroidism” (no TSH or T4 values provided). In a study of 33 hospitalized patients with chronic liver diseases, only one had an increased TSH of 14 mU/L by RIA, in association with a high titer of antithyroglobulin antibodies, an exaggerated TSH response to TRH, and a normal total T4 level, consistent with Hashimoto’s thyroiditis and mild thyroid gland failure (50). No follow-up data were provided to determine whether the mild thyroid gland failure was transient or persistent. Thus, elevated TSH levels in most patients with chronic nonthyroidal illnesses may be unrelated to impaired thyroid gland function. Proposed mechanisms include effects of chronic illness per se, concurrent medications, altered TSH set point, or increased immunoactive TSH that is not bioactive (14, 28, 34), as has been postulated for the elderly. As in the elderly and obese, isolated TSH elevations
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 December 2014. at 11:16 For personal use only. No other uses without permission. . All rights reserved.
doi: 10.1210/jc.2014-1850
Table 2.
jcem.endojournals.org
4021
Continued
Frequency of Anti-TPO Abc ⴙve
Progression to Increased TSH/Low T4 Values
Risk Factors for Progression
Clinical Associations/ Recommendations
Frequency of Increased TSH/Decreased T4 Values
6.8%
17% (2 of 12) “clinically” hypothyroid after 1 y
ND
Increased TSH, >10 with low free T4b 1.9% at baseline
20% Hep C, 5% Hep B 0%
ND
Females, ⴙve antithyroid antibodies, older age ND
ND
0%
ND
ND
ND
ND
in chronically ill patients should not be automatically assumed to represent mild thyroid gland failure. There is insufficient evidence to conclude that isolated TSH elevations, which have not been shown to be persistent or to progress to overt hypothyroidism, represent mild thyroid gland failure in chronically ill patients. Thus, the designation “subclinical hypothyroidism” for isolated TSH elevations in chronically ill patients is at this time a misnomer. What are the clinical associations with isolated TSH elevations? Clinical associations with isolated TSH elevations may not relate to mild thyroid gland dysfunction but may reflect severity of nonthyroidal illness or other factors. Isolated TSH elevations in ESRD patients were associated with impaired cardiac function in a small cross-sectional study (3), with higher mortality risk in a retrospective study (51), and with no increase in cardiovascular events or all-cause mortality over 4 years in a large prospective multicenter cohort study (10) (Table 2). In six prospective cohort studies of CHF patients, isolated TSH increases between 10 and 20 mU/L were associated with increased incidence and recurrence of heart failure (42). In a large retrospective study of CHF patients, TSH levels ⬎4.6 mU/L were associated with increased mortality risk (7). What is the underlying etiology of isolated TSH elevations? Do elevated TSH levels represent bioactive TSH? TSH concentrations are currently measured using third-generation immunoassays, which may reflect in-
creased immunoactive but not bioactive TSH, as in the elderly (28). In the future, new technologies for TSH measurements that quantitate only bioactive TSH may help to accurately identify chronically ill patients with mild thyroid gland failure (28). Until then, mild thyroid gland failure cannot be reliably diagnosed in chronically ill patients solely by isolated TSH elevations but require follow-up for persistence and progression to overt hypothyroidism (15) (Figure 1). What criteria can be used to define mild thyroid gland failure in chronic nonthyroidal illnesses? No data were found to address this question. In patients with chronic nonthyroidal illnesses, the hypothalamic-pituitary-thyroid-peripheral-tissue axis and serum thyroid hormone binding are altered, and in vitro TSH and free T4 estimate values may not reflect in vivo bioactive concentrations, altering the TSH to free T4 relationship (Figure 1) (14, 28, 34, 52). The range of TSH values considered to indicate mild thyroid gland failure in otherwise healthy subjects overlaps completely with those associated with nonthyroidal illnesses (Figure 1) (52, 53). TSH response to exogenous TRH has been used to differentiate mild thyroid gland failure from effects of nonthyroidal illness (43, 50), although this is rarely done in clinical practice. Increased TSH response to exogenous TRH indicates thyroid gland failure that may be persistent or transient, as with iodine excess or thyroiditis.
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 December 2014. at 11:16 For personal use only. No other uses without permission. . All rights reserved.
4022
Kaptein et al
Increased TSH in Chronic NTI
Figure 1. Relationships between serum TSH and direct equilibrium dialysis free T4 values in subjects with a normal hypothalamic-pituitarythyroid-peripheral tissue axis and in patients with nonthyroidal illnesses. [Adapted from E. M. Kaptein: Clinical application of free thyroxine determinations. Clin Lab Med. 1993;13:653– 672 (53), and E. M. Kaptein and J. C. Nelson: Serum thyroid hormones and thyroidstimulating hormone. In: Surks MI, ed. Atlas of Clinical Endocrinology, Volume 1: Thyroid Diseases. 1999:15–31 (52).] The relationship of TSH with free T4 by direct dialysis may only be used to diagnose mild (hatched area) and overt primary and central hypothyroidism when the in vitro assays reflect bioactive TSH and free T4 in vivo, and the hypothalamic-pituitary-thyroid-peripheral tissue axis is intact. In chronic nonthyroidal illnesses, when the hypothalamic-pituitary-thyroidperipheral tissue axis is altered and the in vitro assays may not reliably reflect bioactive TSH and free T4 in vivo, the relationship of TSH to free T4 by direct dialysis may be altered.
Isolated TSH elevations alone cannot be assumed to indicate mild thyroid gland failures in patients with chronic nonthyroidal illnesses (Figure 1). However, progressive TSH elevations over time with reduction in free T4 estimate values by methods not affected by altered serum T4 binding may indicate thyroid gland failure, as in the general population (15). Is there evidence that L-T4 therapy is indicated to treat isolated TSH elevations in patients with chronic nonthyroidal illnesses? No data were found to address this question. Recent retrospective studies suggest that L-T4 therapy in chronic kidney disease patients with isolated TSH elevations were associated with a slower decline in renal function over time (5, 8). Patients with dyslipidemia, high anti-TPO antibody titers, and symptoms to suggest hypothyroidism may have received L-T4 therapy, whereas those without these findings may have remained untreated, potentially biasing the findings (5). Because data to indicate benefit in chronic nonthyroidal illnesses with isolated TSH elevations is lacking, L-T4 therapy should probably be reserved for patients with elevated TSH and reduced free T4 values determined by a method not affected by reduced serum T4 binding in vitro.
J Clin Endocrinol Metab, November 2014, 99(11):4015– 4026
Is there evidence that L-T4 therapy for isolated TSH elevations has an acceptable risk-to-benefit ratio in the general population or the elderly? The benefit of L-T4 therapy in the general population with isolated TSH elevations remains controversial (13– 15, 30, 54 –57). Pre-emptive L-T4 therapy has not been shown to prevent progression to overt hypothyroidism, consistently improve symptoms and signs of hypothyroidism or quality of life, or reduce cardiovascular events or mortality (30). Data for the increased risk of harm from subclinical hyperthyroidism are stronger than data for potential benefit from treatment of isolated TSH elevations in the general population (58 – 61). Several lines of evidence indicate that in older people, isolated elevation in TSH need not be routinely treated. In the elderly, treatment for TSH levels ⬍10 mU/L has not been shown to be beneficial (62) and does not improve cognitive function (63). A large observational study of the United Kingdom General Practitioner Research Database found that treatment with L-T4 was associated with fewer ischemic heart disease events and reduced all-cause mortality during an 8-year period in 40 to 70 year olds with isolated TSH elevations of 5–10 mU/L, but not in those ⬎70 years of age (64). In the Cardiovascular Health Study, which included subjects ⱖ65 years old, isolated TSH elevations between 4.5 and 20 mU/L were not associated with increased coronary heart disease (CHD), cerebrovascular disease, cardiovascular risk, or all-cause mortality over 13 years of follow-up, and only patients with TSH levels ⱖ10 mU/L were more likely to progress to overt hypothyroidism (22, 23, 39). In an elderly subgroup from the Cardiovascular Health Study, TSH levels increased 13% during 13 years of follow-up but were not associated with increased mortality (26). In a meta-analysis from 11 prospective studies of 55 287 adult patients, many of whom were elderly, increased CHD events and CHD mortality were significantly associated only with isolated TSH elevations between 10 and 20 mU/L (12). A reduction of all-cause mortality was observed with isolated TSH elevations between 5 and 10 mU/L in a large retrospective cohort study in Denmark over 5.5 years of follow-up (59), whereas incident ischemic heart disease events and related mortality were increased with isolated baseline TSH 6 –15 mU/L in the Whickham survey cohort after 20 years of follow-up (65). Differences in outcomes may relate to the length of follow-up or characteristics of the study population. In the Denmark study, the risk of major adverse cardiovascular events was increased with overt and subclinical hyperthyroidism over a median of 5.5 years
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 December 2014. at 11:16 For personal use only. No other uses without permission. . All rights reserved.
doi: 10.1210/jc.2014-1850
(59). In a population-based study of mortality in individuals ⱖ 60 years old and not on L-T4 therapy, a single low TSH concentration was associated with increased mortality, particularly due to circulatory and cardiovascular disease (60). In elderly patients from the Cardiovascular Health Study, higher free T4 levels were associated with death (26). The commonest cause of a sustained reduction in TSH concentration in the general population is overtreatment with L-T4 (30). A retrospective cohort study of 52 298 patients from the United Kingdom was followed for up to 5 years after initiating L-T4 therapy (58). Thirty-one percent had L-T4 initiated with an isolated TSH elevation between 4 –10 mU/L, no prior cardiovascular risk factors, or classic hypothyroid symptoms. L-T4 was continued long term in 90%, with up to 6% having suppressed TSH levels and 10% with values between 0.1 and 0.5 mU/L during follow-up. These patients may have been overtreated according to recent guidelines (16). Prior studies indicate that 15 to 25% of patients taking L-T4 are overtreated and develop a low TSH level, and overtreatment may be associated with an increased risk of fractures and atrial fibrillation (58). These findings raise concern about the adverse effect of L-T4 overtreatment of mild TSH elevations in advanced age or with cardiovascular risk factors.
jcem.endojournals.org
4023
Is there evidence that L-T4 therapy for isolated TSH elevations has an acceptable risk-to-benefit ratio in chronically ill patients? No data were found to address this question. A major concern with arbitrary treatment of isolated TSH elevations in patients with chronic nonthyroidal illnesses is overtreatment without demonstrated benefit. The risk of L-T4 therapy for isolated TSH elevations in patients with heart disease or chronic kidney disease who are at risk for concurrent cardiac disease is unintentional induction of iatrogenic subclinical hyperthyroidism with the inherent complications (15). It would appear prudent to recommend that patients with isolated TSH elevations be followed without L-T4 therapy.
What are the criteria to diagnose overt hypothyroidism in chronically ill patients? Are free T4 immunoassays reliable in patients with chronic nonthyroidal illnesses? Free T4 values estimated by immunoassays may be reduced with chronic illness due to decreases in serum binding protein concentrations unrelated to hypothyroidism. In a study employing nine different free T4 immunoassays in euthyroid hemodialyzed ESRD patients with normal TSH values, low free T4 values were observed in 4.5 to 59% (66). The presence and magnitude of biases dependent on serum T4 binding capacity for these free T4 assays were demonstrated by a decrease in measured free T4 values for all imSuspect hypothyroidism in munoassays with serum dilution chronic illness (66). Because free T4 immunoassays Persistent TSH elevation may provide spuriously low free T4 estimates in patients with reduced se10 to 20 mU/L >20 mU/L 4.5 to
F E A T U R E
C l i n i c a l
R e v i e w
Is an Isolated TSH Elevation in Chronic Nonthyroidal Illness “Subclinical Hypothyroidism”? Elaine M. Kaptein, Jonathan S. LoPresti, and Matthew J. Kaptein Department of Medicine, University of Southern California, Los Angeles, California 90033
Context: Elevated TSH with normal T4 frequently occurs with chronic kidney, liver, and heart diseases. Whether isolated TSH elevations represent mild thyroid gland failure has not been established. Evidence Acquisition: PubMed was searched for longitudinal studies in chronic heart, liver, or kidney disease documenting persistent isolated TSH elevations or progression to overt hypothyroidism. Evidence Synthesis: Four articles met inclusion criteria. In 16 end-stage renal failure patients, four had isolated TSH elevations. All normalized within 14 months. In 452 systolic heart failure patients, 20 had isolated TSH elevations, five of 20 were persistent, and none progressed to overt hypothyroidism within 6 months. In 207 untreated chronic hepatitis C patients, 12 had isolated TSH elevations and four had increased TSH with reduced free T4; all were female, and 14 had positive antithyroid antibodies. After 1 year, two of 12 developed “clinical hypothyroidism.” In 72 chronic hepatitis C patients, nine females had positive antithyroid antibodies. Two antibody-negative patients had TSH 5– 6 mU/L with reduced free T4. After 1 year, three of four with positive antithyroid antibodies and baseline TSH ⬍ 4 mU/L had elevated TSH with reduced free T4. Conclusions: In chronically ill patients, there is inadequate evidence to determine: 1) that isolated TSH elevations usually persist or progress to overt hypothyroidism; 2) the etiology and clinical significance of isolated TSH elevations; and 3) whether levothyroxine therapy is indicated for persistent isolated TSH elevations. Thus, isolated TSH elevations in chronic renal, cardiac, or liver diseases have not been documented to indicate mild thyroid gland failure. (J Clin Endocrinol Metab 99: 4015–4026, 2014)
C
riteria for diagnosing mild thyroid gland failure (“subclinical hypothyroidism”) in the general population have been arbitrarily applied to patients with chronic nonthyroidal illnesses (1–11). Mild thyroid gland failure is defined as TSH levels persistently between 4.5 and 20 mU/L, with normal T4 estimates, or progression to overt hypothyroidism (12–15). An increased TSH level is a sensitive indicator of mild thyroid gland failure and decreased free T4 negative feedback only if the hypothalamic-pituitary-thyroid-gland-peripheral-tissue axis is intact and serum TSH and free T4 concentrations reflect bioactive TSH and in vivo free T4 (15).
These assumptions may not apply in nonthyroidal illnesses (14 –16). Isolated TSH elevations in chronic nonthyroidal illnesses have been attributed to mild thyroid gland failure without substantiating evidence, and patients have been arbitrarily treated with levothyroxine (LT4) therapy with variable outcomes (5, 8, 9). The TSH to free T4 relationship is altered in ambulatory elderly and the obese, which may provide clues to alterations in chronic nonthyroidal illness. Leptin, dopamine, glucocorticoids, and cytokines can affect TSH independently of free T4 concentrations without affecting thyroid status (17). In many obese patients, TSH is modestly elevated with normal free T4 values (18), likely resulting from
ISSN Print 0021-972X ISSN Online 1945-7197 Printed in U.S.A. Copyright © 2014 by the Endocrine Society Received March 25, 2014. Accepted August 5, 2014. First Published Online August 28, 2014
Abbreviations: BMI, body mass index; CHD, coronary heart disease; CHF, congestive heart failure; ESRD, end-stage renal disease; L-T4, levothyroxine; TBG, T4-binding globulin; TPO, thyroid peroxidase.
doi: 10.1210/jc.2014-1850
J Clin Endocrinol Metab, November 2014, 99(11):4015– 4026
jcem.endojournals.org
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 December 2014. at 11:16 For personal use only. No other uses without permission. . All rights reserved.
4015
4016
Kaptein et al
Table 1.
Increased TSH in Chronic NTI
J Clin Endocrinol Metab, November 2014, 99(11):4015– 4026
Elements of Systematic Review of Elevated TSH in Chronic Nonthyroidal Illnesses
Questions posed regarding patients with chronic nonthyroidal illnesses 1. What evidence indicates that mild thyroid gland failure is present with isolated TSH elevations? a. Do TSH elevations persist or progress over time, with reductions in free T4 values? 2. What is the underlying etiology of the isolated serum TSH elevations? 3. Do elevated in vitro TSH levels represent bioactive TSH? 4. What criteria can be used to define mild thyroid gland failure? 5. What are the clinical associations with isolated TSH elevations? 6. Is there evidence that L-T4 therapy is indicated to treat persistent isolated TSH elevations and has an acceptable risk-tobenefit ratio? 7. What are the criteria to diagnose overt hypothyroidism? a. Are free T4 immunoassays reliable in patients with chronic nonthyroidal illnesses? 8. What are TSH goals when L-T4 treatment is given for overt hypothyroidism with chronic illness? Inclusion criteria Studies of patients with chronic kidney, liver, or heart disease who had: 1. Elevated serum TSH and normal total/free T4 with follow-up for persistent isolated TSH elevation or progression to overt hypothyroidism. 2. Documentation of effects of L-T4 therapy for mild or overt thyroid gland failure on signs and symptoms, or TSH and T4 levels. Exclusion criteria 1. Studies with TSH and T4 follow-up of ⬍3 months. 2. Studies without a T4 or free T4 estimate value. 3. Patients without chronic nonthyroidal illnesses. 4. Patients with acute nonthyroidal illnesses. 5. Patients receiving medications known to induce thyroid gland failure or alter serum TSH and T4 concentrations, including dopamine or somatostatin analogs, phenytoin, carbamazepine, sertraline, iodine, iodine containing medications including amiodarone, lithium, interferon-␣, glucocorticoids, rifampin, and sunitinib (15, 16, 29, 30). 6. Patients with known thyroid disease. 7. Studies where L-T4 therapy was initiated before monitoring for persistence or progression of TSH or T4 measurements.
obesity, not causing it (17–20). The ambulatory elderly have a high prevalence of isolated TSH elevations, not due to mild thyroid gland failure (21–23), but secondary to altered hypothalamic-pituitary-thyroid set point, down-regulation of hepatic and pituitary type 2 deiodinase activity, reduced TSH bioactivity, concurrent nonthyroidal illnesses, medications, or laboratory artifacts (24 –28). We undertook a systematic literature review to address the questions posed in Table 1.
Methods Evidence acquisition We searched PubMed for the following terms: subclinical hypothyroidism, hypothyroidism, TSH, T4, chronic kidney disease, chronic heart disease, chronic heart failure, coronary heart disease, chronic liver disease, longitudinal studies, and L-T4 therapy. The search was limited to reports in English published from 1946 to May 23, 2014, and supplemented by personal files and bibliographies of relevant articles. Longitudinal studies of patients with chronic renal, liver, or cardiac disease were reviewed for data indicating frequency of persistent isolated TSH elevations, progression of TSH elevations with reduced T4 estimates, and clinical associations (Table 1). Response of TSH and T4 values and clinical parameters following L-T4 therapy were assessed.
Results We reviewed 351 articles from the literature search. No randomized controlled trials addressed any of the questions posed. Only four articles described observational studies that had longitudinal data in chronically ill patient cohorts matching our inclusion criteria (6, 31–33). Information in these studies was compared to published data for the general population, ambulatory elderly, and patients with obesity. Definitions of mild thyroid gland failure and overt hypothyroidism in the general population Mild thyroid gland failure is defined as a persistently isolated TSH elevation or a progression to overt hypothyroidism (13–15). To diagnose an isolated TSH elevation as mild thyroid gland failure, the hypothalamic-pituitarythyroid-peripheral-tissue axis must be intact, and other causes of TSH elevations must be excluded, such as nonthyroidal illnesses or medications (15, 16, 34). Symptoms and signs of hypothyroidism are nonspecific and nondiagnostic and should not be used to define mild thyroid gland failure (16, 30). An isolated elevation of TSH with a normal total/free T4 value has been reported in 4 to 10% of the general population not on thyroid hormone therapy (Table 2) (1,
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 December 2014. at 11:16 For personal use only. No other uses without permission. . All rights reserved.
doi: 10.1210/jc.2014-1850
2, 12–15, 35). TSH elevations normalize in 50% of these individuals over time (23). TSH levels should be re-evaluated after 3 to 6 months to rule out laboratory error or transient increases (13, 14) and to determine whether progressive increases in TSH levels in association with reduced free T4 estimates have occurred, indicating development of overt hypothyroidism. TSH assays are poorly harmonized among laboratories, and values may be higher due to assay variability rather than to thyroid gland failure (36). In the general population, L-T4 therapy has been recommended for persistent TSH ⬎10 mU/L (16), whereas L-T4 treatment with TSH ⬍ 10 mU/L remains controversial (13–16). Progression to overt hypothyroidism— defined as a persistently increased TSH, usually ⬎10 IU/mL, with a reduced free T4 estimate— has been reported to occur in 2 to 5% of the general population per year (13–15) (Table 2). Those at the highest risk for progression to overt hypothyroidism are elderly, female, with positive anti-thyroid peroxidase (TPO) antibodies, TSH ⬎ 10 mU/L (13– 15, 35) (Table 2), or thyroid hypoechogenicity by ultrasonography (37). Positive anti-TPO antibodies usually indicate autoimmune thyroid disease. Antithyroid antibodies and ultrasound abnormalities do not indicate thyroid gland failure. Elevated TSH values in the elderly Basal TSH levels may be elevated in 3 to 15% of the elderly (21–23, 38) (Table 2). In one study of 5182 elderly patients, 3.3% had isolated TSH elevations at baseline, which persisted in 54% at 6 months, with none having reduced free T4 estimates (38). In another study of 3594 community-dwelling patients over age 65, 12.8% had isolated TSH elevations at baseline (23). After 2 and 4 years, 56% (208 of 369) and 75% (121 or 161) had persistent TSH elevations, respectively, whereas 2% (eight of 369) and 2% (four of 208) had an elevated TSH with a reduced free T4 estimate or TSH ⬎ 20 mU/L, respectively (23). Isolated TSH elevations in community-dwelling elderly have not been shown to contribute to decreased functional capacity (38). The TSH to free T4 relationship is frequently altered in the ambulatory elderly due to changes independent of thyroid gland failure (14, 21, 28, 39). Two cohort studies have shown that aging is associated with an increase in TSH without a fall in free T4 (26, 27). A large cross-sectional study showed that at any given free T4 value within the reference range, TSH was higher in older people than younger people, suggesting that older people will be diagnosed with mild thyroid gland failure more commonly than younger people despite having identical free T4 estimate values (24). Consequently, a diagnosis of mild thy-
jcem.endojournals.org
4017
roid gland failure in the elderly cannot be accurately established using criteria from the general population because increased TSH may be due to causes unrelated to thyroid gland failure. Elevated TSH levels in patients with obesity In 21 097 patients with body mass index (BMI) of ⱖ30 kg/m2, 2.2% had isolated TSH elevations, and 0.7% had elevated TSH levels with reduced free T4 estimates (40) (Table 2). Isolated TSH elevations were reported in 6% of morbidly obese subjects, compared to 4% of lean subjects, with anti-TPO antibodies present in 17% of obese and 8% of lean subjects (18). Elevated leptin levels may have a stimulatory effect on TSH secretion, resulting in increased circulating TSH concentrations (20). In a retrospective study of 86 morbidly obese patients undergoing gastric bypass or banding, isolated TSH elevations were present in 10.5% before surgery and in 0% 6 months after surgery, in association with a mean BMI decrease from 49 to 32 kg/m2 (41). Normalization of TSH with weight loss suggests that obesity per se plays an integral role in TSH regulation, independent of free T4 concentrations (17, 19, 20). Isolated TSH elevations in patients with chronic nonthyroidal illnesses A high frequency of elevated TSH with normal total or free T4 estimates is observed with chronic kidney diseases (1.6 to 28%) (3, 4, 10, 11, 34) and congestive heart failure (CHF) (6 to 16%), unrelated to medications known to induce thyroid gland failure (7, 42) (Table 2). In a small study of patients with chronic hepatic cirrhosis, 38% had elevated TSH by RIA and normal free T4 by tracer equilibrium dialysis (43). Isolated TSH elevations were reported in 0 to 6% with untreated chronic hepatitis C (32, 33, 44, 45). In 134 patients with chronic hepatitis C and 41 with chronic hepatitis B, 3.7% with hepatitis C and 0% with hepatitis B had isolated TSH elevations, whereas TPO antibodies were positive in 20% and 5%, respectively (45). Untreated hepatitis C-positive patients have a 1.6-fold increased risk for thyroid autoimmune disorders, even without cirrhosis, compared to patients with sera negative for hepatitis C virus antibody, those with chronic hepatitis B infection, and healthy subjects (46 – 48). What evidence indicates that mild thyroid gland failure is present with elevated TSH and normal free T4 values in patients with chronic nonthyroidal illnesses? In most studies, TSH and free T4 values were not followed over time, and symptoms and signs to indicate hy-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 December 2014. at 11:16 For personal use only. No other uses without permission. . All rights reserved.
4018
Kaptein et al
Increased TSH in Chronic NTI
J Clin Endocrinol Metab, November 2014, 99(11):4015– 4026
Table 2. Review of TSH Elevations With or Without Free or Total T4 Reductions in the General Population, Elderly, Obese, and With Chronic Nonthyroidal Illnesses First Author, Year (Ref.)
Clinical Characteristics
Study Design
TSH, mU/La
Total T4/Free T4 Assayb
Frequency of Increased TSH/Normal T4 Values
General population Hollowell, 2002 (35)
NHANES III US general population
Cross-sectional, n ⫽ 13 344
⬎4.6
Total T4, 4.5–13.2
4.3%
Surks, 2004 (13)
General population
Extensive literature review
Variable
4 to 8.5%
Lo, 2005 (1)d
NHANES III
Cross-sectional, n ⫽ 14 623
Mild, 4.5–10; severe, ⬎10 ⬎4.5
Total T4 ⬎ 4.5
8%
General population
Extensive literature review
⬎4.5
Variable
4 to 10%
Adult outpatients
Cross-sectional, n ⫽ 3089
⬎4.5
Free T4
9.5%
Varied by study from community-dwelling adults to elderly worldwide General population
Meta-analysis, n ⫽ 55 287, 11 prospective studies Extensive literature review
⬎4.5 to ⬍20
T4 when available, various methods Variable
6.2% TSH, ⬎4.5 to ⬍20 (3.3 to 16.4%) 4 to 20%, (4.3 to 9.5% in US)
Community-dwelling ⬎65 y old, in Cardiovascular Health Study USA Community-dwelling ⬎65 y old, in Cardiovascular Health Study USA Well elderly
Prospective cohort study, n ⫽ 3233; 13 y follow-up Longitudinal study over 4 y, n ⫽ 3594 Review of the literature
⬎4.5 to ⬍20
Free T4
15%
⬎4.5 to ⬍20
Free T4
12.8%, persistent in 56% at 2 y
Free T4, various methods
15%
Pravastatin in the Elderly at Risk (PROSPER) studyf
Prospective, n ⫽ 5182; 6-mo follow-up
⬎Upper reference range, but ⬍10 ⬎4.5
Free T4
3.3%, persistent in 54% at 6 mo
Morbidly obese before and after gastric bypass or banding BMI ⬎ 30 kg/m2 from HUNT study Obese (mean BMI, 43 kg/m2) referred for workup to a tertiary care center
Retrospective, n ⫽ 86
⬎5.5
ND
Cross-sectional, n ⫽ 21 097 Cross-sectional, n ⫽ 165 obese, n ⫽ 118 lean
⬎4.0 ⬎4 to ⬍10
Free T4 Free T4
10.5% pre-op, and 0% at 6 mo post-op 2.2% 6% obese, 4% lean
CKD stage 2– 4
Retrospective, n ⫽ 129 not treated, n ⫽ 180 on L-T4
Free T4
ND
CKD stages 2– 4, treated with L-T4 for 24 mo
Retrospective, n ⫽ 113 before and after L-T4
⬎4.9 to 10, repeated within 3 mo ⬎4.9, repeated within 3 mo
Free T4
ND
ESRD
Literature review
⬎4.5
Total T4/free T4 index
10.5–12.5% ⬎ 4.5, 1% ⬎ 10
Kang, 2008 (3)
ESRD on PD ⬎3 mo (Korea)
Cross-sectional, n ⫽ 51
⬎5
Free T4
27.5%
Ng, 2012 (4)f
ESRD on PD ⬎ 6 mo (Taiwan)
Cross-sectional, n ⫽ 122
⬎4
Free T4
15.6%
ESRD diabetic patients in Germany
Prospective observational multicenter, n ⫽ 1000 Retrospective, n ⫽ 2715
4.1–15
Free T4
1.6%
⬎5.0/5.7 to ⬍10
Not assessed
12.9%
Literature review
4.1–15
Not documented
8.9 to 24.8%
Biondi, 2008 (15) Chonchol, 2008 (2)
d
Rodondi, 2010 (12) Cooper, 2012 (14) Elderly Cappola, 2006 (22) Somwaru, 2012 (23)e Visser, 2013 (21)
Virgini, 2014 (38) Obesity Chikunguwo, 2007 (41) Asvold, 2009 (40) Marzullo, 2010 (18) Chronic kidney disease pre-ESRD Shin, 2012 (5)
Shin, 2013 (8)
ESRD Kaptein, 1996 (34)
Drechsler, 2014 (10) f
⬎4.5; mild, 4.5– 9; severe, ⬎10
Rhee, 2014 (11)
ESRD on PD and hemodialysis (United States) ESRD
Kalk, 1980 (31)
ESRD on hemodialysis
Prospective crosssectional, n ⴝ 16, 14-mo follow-up
>8.0
Free T4 index
25%, all transient
CHF, United States and Europe
Six prospective cohorts, n ⫽ 25 390
4.5–19.9
Free T4, various methods
8.1% (range 5.5–16.2%)
Rhee, 2013 (7)
CHF
⬎4.6
Total T4
4.6%
Frey, 2013 (6)
Systolic HF
Retrospective NHANES III, n ⫽ 14 879 Prospective INH study, 6-mo follow-up, n ⴝ 452
>4.0
Free T4/free T3
4.5%, persistent in 25% (5 of 20) at 6 mo
“Euthyroid” hepatic cirrhosis
Cross-sectional, n ⫽ 23
⬎7.5 mg/dL
38%g
Borzio, 1983 (50)
Liver cirrhosis and chronic hepatitis
⬍5
Tran, 1993 (33)
Chronic Hep C
Cross-sectional, n ⫽ 33 and 22, respectively Prospective, n ⴝ 72, followed for 1 y
Equilibrium dialysis, ⬍1.8 ng/mL Total T4
>4
Free T4
0%
Rhee, 2013 (51)
Cardiac disease Gencer, 2012 (42)
Chronic liver disease Chopra, 1974 (43)
1.8%h
(Continued)
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 December 2014. at 11:16 For personal use only. No other uses without permission. . All rights reserved.
doi: 10.1210/jc.2014-1850
Table 2.
jcem.endojournals.org
4019
Continued
Frequency of Anti-TPO Abc ⴙve
Progression to Increased TSH/Low T4 Values
Risk Factors for Progression
Clinical Associations/ Recommendations
Frequency of Increased TSH/Decreased T4 Values
13%
ND
ND
0.3%
Frequent
2–5%/y ND
ND
ND
Elderly, females, TPO Ab ⫹ve, TSH ⬎20 ND
ND
ND
ND
Positive in 60 – 80%
Up to 4%/ y in women with TPO Ab ⫹ve
TPO Ab ⫹ve, iodine intake, TSH ⬎ 10–15
No response to L-T4 if TSH ⬍ 10, repeat TSH every 6 –12 mo 20 to 23% increased TSH with eGFR ⬍ 60 mL/min/1.73 m2 Monitor every 6 –12 mo, L-T4 therapy if TSH ⬎ 10 18% increased TSH with eGFR ⬍ 60 mL/min/1.73 m2 Increased CHD events and CHD mortality for TSH 10 –19.9 Treat if TSH ⬎10 mU/L
ND
55% with increased TSH Frequent
Female, increased age, TPO ⫹ve ⬎ age 65 y, TSH ⬎10, TPO Ab ⫹ve ND
ND
ND
ND
No increase in CHD, CVD, CV
1.6% had TSH ⬎ 20 or TSH ⬎ 4.5/low
death or all-cause death ND
free T4b 0.61% had TSH ⬎ 20 or TSH ⬎ 4.5/ low free T4b ND
4.3%/y
ND
2% at 2 y
TSH ⬎ 10 mU/L
ND
A few % per year
Higher TSH, TPO Ab, goiter
ND
0%
ND
ND ND ND ND ND
Female (64%)
Decreased or no change in mortality, repeat TSH in 3– 6 mo No decreased functional capacity
ND
ND
ND
ND
0%
ND 17% obese, 8% lean
ND ND
ND ND
ND ND
TSH ⬎ 4/low free T4 in 0.7% TSH ⬎ 4/low free T4b in 4% obese (1.8% TSH ⬎ 20), 0% lean
⫹ve 26/81 on L-T4, 0/59 no L-T4
No increase in TSH over 36 mo
None
Less decline in eGFR with L-T4 to TSH WNL over 6 mo
ND
⫹ve in 40%
ND
ND
Less decline in eGFR with L-T4 to TSH WNL in 64%, faster decline in 30%
ND
ND
All TSHs 10 –20 were transient within 1–2 wk ND
ND
Do not treat with L-T4
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Lower LVEF and fractional shortening Longer PD duration, lower BMI, higher EPO doses No increased CV events or allcause mortality over 4 y Higher mortality with increased TSH ND
TSH persistently ⬎ 20/low free T4b in 2.6 –9.5% 0%
TSH ⬎ 20/low total or free T4b in 2.6 –
ND
0%
ND
ND
5.4% 0%
ND
ND
ND
ND
ND
ND
ND
0%
ND
Increased incidence and recurrence of HF with TSH 10 –19.9 Increased mortality risk in those with CHF No impact on survival
ND
ND
ND
ND
0%
3.6%
ND
ND
ND
0%
12.5% all female
3 of 4 with normal basal TSH at 1 y
Females, ⴙve antithyroid antibodies
ND
Increased TSH 5.3 and 5.6 with low free T4b in 2 of9 (22%) at
⫹ve in 4 of 14 with TSH ⬎ 5 ND
0%
ND
ND TSH > 4.0 and decreased free T4b in 0%
baseline (Continued)
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 December 2014. at 11:16 For personal use only. No other uses without permission. . All rights reserved.
4020
Kaptein et al
Table 2.
Increased TSH in Chronic NTI
J Clin Endocrinol Metab, November 2014, 99(11):4015– 4026
Continued
First Author, Year (Ref.)
Clinical Characteristics
Study Design
TSH, mU/La
Total T4/Free T4 Assayb
Frequency of Increased TSH/Normal T4 Values
Marazuela, 1996 (32)
Chronic Hep C
>4
Free T4
5.8%
Fernandez-Soto, 1998 (45) Danilovic, 2013 (44)
Chronic Hep C and B
Prospective, n ⴝ 207, followed for up to 1 y pretreatment Prospective, n ⫽ 134 Hep C, n ⫽ 41 Hep B Cross-sectional, n ⫽ 103; controls n ⫽ 96
⬎4
Free T4
3.7% Hep C 0% Hep B
⬎4.5
Free T4
3.9%
Chronic Hep C
Abbreviations: NHANES III, Third National Health and Nutrition Examination Survey, WNL, within reference range; CKD, chronic kidney disease; ESRD, end-stage renal disease receiving dialysis therapy; PD, peritoneal dialysis; EPO, erythropoietin dose; LVEF, left ventricular ejection fraction; eGFR, estimated glomerular filtration rate using four-variable Modification of Diet in Renal Disease study (MDRD) equation; HF, heart failure; CV, cardiovascular; CVD, cerebrovascular disease; INH, Interdisciplinary Network Heart Failure study; HUNT, Nord-Trondelag Health Study; Hep, hepatitis; Ab, antibody; ⫹ve, positive; ND, no data. a Upper population reference range. Note: Many studies do not discriminate between TSH levels that are slightly (4 –10 mU/L) and more severely (⬎10 mU/L) elevated. b Free T4 estimates by direct analog immunoassays unless otherwise stated. Estimate values are method dependent and may be inappropriately low in patients with reduced serum binding protein concentrations and nonthyroidal illnesses (67, 68) c
Anti-TPO Ab indicates anti-TPO antibodies (indicates autoimmune thyroid disease).
d
Included in review by Rhee et al. (11), but additional information provided.
e
Included in review by Gencer et al. (42), but additional information provided.
f
Elderly population, mean age 75 y, with cardiovascular disease or at high risk of developing cardiovascular disease.
g
Blunted TSH response to exogenous TRH.
h
Basal TSH 14 mU/L with increased TSH response to TRH.
Articles that were prospective studies and satisfied inclusion criteria are in bold print.
pothyroidism were not reported. Very little evidence indicated that mild thyroid gland failure was present in chronically ill patients with elevated TSH but normal free T4 values. Only the four highlighted articles (in bold print) in Table 2 met our inclusion criteria. The first is a prospective cross-sectional study of patients hospitalized for systolic heart failure, of which 20 out of 452 had isolated TSH elevations, which normalized in 15 of 20 after 6 months (6). None developed reduced free T4 levels, and there was no association with overall survival (6). The second is a prospective cross-sectional study of 16 hemodialyzed end-stage renal disease (ESRD) patients, in which four had transient isolated TSH elevations, with none progressing over 14 months of follow-up (31). In a cross-sectional study of ESRD patients, 9% of 306 patients had isolated TSH levels between 5 and 10 mU//L, with 1% of 306 having TSH between 10 and 20 mU/L with normal free T4 index values (49). All TSH values between 10 and 20 mU/L decreased to ⬍ 10 mU/L when repeated within 1 to 2 weeks (49). This was not a longitudinal study, so persistence and progression of TSH elevations were not assessed. In the third prospective study, 72 patients with chronic hepatitis C were followed for 1 year (33). Nine of the 72 had positive antithyroid antibodies with normal TSH values. Two of nine with negative antibodies had TSH values of 5.3 and 5.6 mU/L, respectively, with reduced free T4
estimates. After 1 year, three of four patients with positive antithyroid antibodies and baseline TSH values ⬍4 mU/L demonstrated elevated TSH with reduced free T4 estimates. In the fourth prospective study, of 207 patients with untreated chronic hepatitis C followed for 1 year, 5.8% (all female) had isolated TSH levels, and 1.9% had elevated TSH with reduced free T4 levels at baseline (18). Eighty-eight percent with elevated TSH had positive thyroid antibodies. After 1 year, two of 12 patients with isolated TSH elevations developed “clinical hypothyroidism” (no TSH or T4 values provided). In a study of 33 hospitalized patients with chronic liver diseases, only one had an increased TSH of 14 mU/L by RIA, in association with a high titer of antithyroglobulin antibodies, an exaggerated TSH response to TRH, and a normal total T4 level, consistent with Hashimoto’s thyroiditis and mild thyroid gland failure (50). No follow-up data were provided to determine whether the mild thyroid gland failure was transient or persistent. Thus, elevated TSH levels in most patients with chronic nonthyroidal illnesses may be unrelated to impaired thyroid gland function. Proposed mechanisms include effects of chronic illness per se, concurrent medications, altered TSH set point, or increased immunoactive TSH that is not bioactive (14, 28, 34), as has been postulated for the elderly. As in the elderly and obese, isolated TSH elevations
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 December 2014. at 11:16 For personal use only. No other uses without permission. . All rights reserved.
doi: 10.1210/jc.2014-1850
Table 2.
jcem.endojournals.org
4021
Continued
Frequency of Anti-TPO Abc ⴙve
Progression to Increased TSH/Low T4 Values
Risk Factors for Progression
Clinical Associations/ Recommendations
Frequency of Increased TSH/Decreased T4 Values
6.8%
17% (2 of 12) “clinically” hypothyroid after 1 y
ND
Increased TSH, >10 with low free T4b 1.9% at baseline
20% Hep C, 5% Hep B 0%
ND
Females, ⴙve antithyroid antibodies, older age ND
ND
0%
ND
ND
ND
ND
in chronically ill patients should not be automatically assumed to represent mild thyroid gland failure. There is insufficient evidence to conclude that isolated TSH elevations, which have not been shown to be persistent or to progress to overt hypothyroidism, represent mild thyroid gland failure in chronically ill patients. Thus, the designation “subclinical hypothyroidism” for isolated TSH elevations in chronically ill patients is at this time a misnomer. What are the clinical associations with isolated TSH elevations? Clinical associations with isolated TSH elevations may not relate to mild thyroid gland dysfunction but may reflect severity of nonthyroidal illness or other factors. Isolated TSH elevations in ESRD patients were associated with impaired cardiac function in a small cross-sectional study (3), with higher mortality risk in a retrospective study (51), and with no increase in cardiovascular events or all-cause mortality over 4 years in a large prospective multicenter cohort study (10) (Table 2). In six prospective cohort studies of CHF patients, isolated TSH increases between 10 and 20 mU/L were associated with increased incidence and recurrence of heart failure (42). In a large retrospective study of CHF patients, TSH levels ⬎4.6 mU/L were associated with increased mortality risk (7). What is the underlying etiology of isolated TSH elevations? Do elevated TSH levels represent bioactive TSH? TSH concentrations are currently measured using third-generation immunoassays, which may reflect in-
creased immunoactive but not bioactive TSH, as in the elderly (28). In the future, new technologies for TSH measurements that quantitate only bioactive TSH may help to accurately identify chronically ill patients with mild thyroid gland failure (28). Until then, mild thyroid gland failure cannot be reliably diagnosed in chronically ill patients solely by isolated TSH elevations but require follow-up for persistence and progression to overt hypothyroidism (15) (Figure 1). What criteria can be used to define mild thyroid gland failure in chronic nonthyroidal illnesses? No data were found to address this question. In patients with chronic nonthyroidal illnesses, the hypothalamic-pituitary-thyroid-peripheral-tissue axis and serum thyroid hormone binding are altered, and in vitro TSH and free T4 estimate values may not reflect in vivo bioactive concentrations, altering the TSH to free T4 relationship (Figure 1) (14, 28, 34, 52). The range of TSH values considered to indicate mild thyroid gland failure in otherwise healthy subjects overlaps completely with those associated with nonthyroidal illnesses (Figure 1) (52, 53). TSH response to exogenous TRH has been used to differentiate mild thyroid gland failure from effects of nonthyroidal illness (43, 50), although this is rarely done in clinical practice. Increased TSH response to exogenous TRH indicates thyroid gland failure that may be persistent or transient, as with iodine excess or thyroiditis.
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 December 2014. at 11:16 For personal use only. No other uses without permission. . All rights reserved.
4022
Kaptein et al
Increased TSH in Chronic NTI
Figure 1. Relationships between serum TSH and direct equilibrium dialysis free T4 values in subjects with a normal hypothalamic-pituitarythyroid-peripheral tissue axis and in patients with nonthyroidal illnesses. [Adapted from E. M. Kaptein: Clinical application of free thyroxine determinations. Clin Lab Med. 1993;13:653– 672 (53), and E. M. Kaptein and J. C. Nelson: Serum thyroid hormones and thyroidstimulating hormone. In: Surks MI, ed. Atlas of Clinical Endocrinology, Volume 1: Thyroid Diseases. 1999:15–31 (52).] The relationship of TSH with free T4 by direct dialysis may only be used to diagnose mild (hatched area) and overt primary and central hypothyroidism when the in vitro assays reflect bioactive TSH and free T4 in vivo, and the hypothalamic-pituitary-thyroid-peripheral tissue axis is intact. In chronic nonthyroidal illnesses, when the hypothalamic-pituitary-thyroidperipheral tissue axis is altered and the in vitro assays may not reliably reflect bioactive TSH and free T4 in vivo, the relationship of TSH to free T4 by direct dialysis may be altered.
Isolated TSH elevations alone cannot be assumed to indicate mild thyroid gland failures in patients with chronic nonthyroidal illnesses (Figure 1). However, progressive TSH elevations over time with reduction in free T4 estimate values by methods not affected by altered serum T4 binding may indicate thyroid gland failure, as in the general population (15). Is there evidence that L-T4 therapy is indicated to treat isolated TSH elevations in patients with chronic nonthyroidal illnesses? No data were found to address this question. Recent retrospective studies suggest that L-T4 therapy in chronic kidney disease patients with isolated TSH elevations were associated with a slower decline in renal function over time (5, 8). Patients with dyslipidemia, high anti-TPO antibody titers, and symptoms to suggest hypothyroidism may have received L-T4 therapy, whereas those without these findings may have remained untreated, potentially biasing the findings (5). Because data to indicate benefit in chronic nonthyroidal illnesses with isolated TSH elevations is lacking, L-T4 therapy should probably be reserved for patients with elevated TSH and reduced free T4 values determined by a method not affected by reduced serum T4 binding in vitro.
J Clin Endocrinol Metab, November 2014, 99(11):4015– 4026
Is there evidence that L-T4 therapy for isolated TSH elevations has an acceptable risk-to-benefit ratio in the general population or the elderly? The benefit of L-T4 therapy in the general population with isolated TSH elevations remains controversial (13– 15, 30, 54 –57). Pre-emptive L-T4 therapy has not been shown to prevent progression to overt hypothyroidism, consistently improve symptoms and signs of hypothyroidism or quality of life, or reduce cardiovascular events or mortality (30). Data for the increased risk of harm from subclinical hyperthyroidism are stronger than data for potential benefit from treatment of isolated TSH elevations in the general population (58 – 61). Several lines of evidence indicate that in older people, isolated elevation in TSH need not be routinely treated. In the elderly, treatment for TSH levels ⬍10 mU/L has not been shown to be beneficial (62) and does not improve cognitive function (63). A large observational study of the United Kingdom General Practitioner Research Database found that treatment with L-T4 was associated with fewer ischemic heart disease events and reduced all-cause mortality during an 8-year period in 40 to 70 year olds with isolated TSH elevations of 5–10 mU/L, but not in those ⬎70 years of age (64). In the Cardiovascular Health Study, which included subjects ⱖ65 years old, isolated TSH elevations between 4.5 and 20 mU/L were not associated with increased coronary heart disease (CHD), cerebrovascular disease, cardiovascular risk, or all-cause mortality over 13 years of follow-up, and only patients with TSH levels ⱖ10 mU/L were more likely to progress to overt hypothyroidism (22, 23, 39). In an elderly subgroup from the Cardiovascular Health Study, TSH levels increased 13% during 13 years of follow-up but were not associated with increased mortality (26). In a meta-analysis from 11 prospective studies of 55 287 adult patients, many of whom were elderly, increased CHD events and CHD mortality were significantly associated only with isolated TSH elevations between 10 and 20 mU/L (12). A reduction of all-cause mortality was observed with isolated TSH elevations between 5 and 10 mU/L in a large retrospective cohort study in Denmark over 5.5 years of follow-up (59), whereas incident ischemic heart disease events and related mortality were increased with isolated baseline TSH 6 –15 mU/L in the Whickham survey cohort after 20 years of follow-up (65). Differences in outcomes may relate to the length of follow-up or characteristics of the study population. In the Denmark study, the risk of major adverse cardiovascular events was increased with overt and subclinical hyperthyroidism over a median of 5.5 years
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 December 2014. at 11:16 For personal use only. No other uses without permission. . All rights reserved.
doi: 10.1210/jc.2014-1850
(59). In a population-based study of mortality in individuals ⱖ 60 years old and not on L-T4 therapy, a single low TSH concentration was associated with increased mortality, particularly due to circulatory and cardiovascular disease (60). In elderly patients from the Cardiovascular Health Study, higher free T4 levels were associated with death (26). The commonest cause of a sustained reduction in TSH concentration in the general population is overtreatment with L-T4 (30). A retrospective cohort study of 52 298 patients from the United Kingdom was followed for up to 5 years after initiating L-T4 therapy (58). Thirty-one percent had L-T4 initiated with an isolated TSH elevation between 4 –10 mU/L, no prior cardiovascular risk factors, or classic hypothyroid symptoms. L-T4 was continued long term in 90%, with up to 6% having suppressed TSH levels and 10% with values between 0.1 and 0.5 mU/L during follow-up. These patients may have been overtreated according to recent guidelines (16). Prior studies indicate that 15 to 25% of patients taking L-T4 are overtreated and develop a low TSH level, and overtreatment may be associated with an increased risk of fractures and atrial fibrillation (58). These findings raise concern about the adverse effect of L-T4 overtreatment of mild TSH elevations in advanced age or with cardiovascular risk factors.
jcem.endojournals.org
4023
Is there evidence that L-T4 therapy for isolated TSH elevations has an acceptable risk-to-benefit ratio in chronically ill patients? No data were found to address this question. A major concern with arbitrary treatment of isolated TSH elevations in patients with chronic nonthyroidal illnesses is overtreatment without demonstrated benefit. The risk of L-T4 therapy for isolated TSH elevations in patients with heart disease or chronic kidney disease who are at risk for concurrent cardiac disease is unintentional induction of iatrogenic subclinical hyperthyroidism with the inherent complications (15). It would appear prudent to recommend that patients with isolated TSH elevations be followed without L-T4 therapy.
What are the criteria to diagnose overt hypothyroidism in chronically ill patients? Are free T4 immunoassays reliable in patients with chronic nonthyroidal illnesses? Free T4 values estimated by immunoassays may be reduced with chronic illness due to decreases in serum binding protein concentrations unrelated to hypothyroidism. In a study employing nine different free T4 immunoassays in euthyroid hemodialyzed ESRD patients with normal TSH values, low free T4 values were observed in 4.5 to 59% (66). The presence and magnitude of biases dependent on serum T4 binding capacity for these free T4 assays were demonstrated by a decrease in measured free T4 values for all imSuspect hypothyroidism in munoassays with serum dilution chronic illness (66). Because free T4 immunoassays Persistent TSH elevation may provide spuriously low free T4 estimates in patients with reduced se10 to 20 mU/L >20 mU/L 4.5 to
