doi: 10.1111/j.1365-2796.2007.01865.x

Clinical and biochemical implications of low thyroidhormone levels (total and free forms) in euthyroid patientswith chronic kidney disease

J. J. Carrero1, A. R. Qureshi1, J. Axelsson1, M. I. Yilmaz1, S. Rehnmark2, M. R. Witt2, P. Barany1,O. Heimburger1, M. E. Suliman1, A. Alvestrand1, B. Lindholm1 & P. Stenvinkel1

From the 1Divisions of Renal Medicine and Baxter Novum; Department of Clinical Science, Intervention and Technology,Karolinska Institutet, Stockholm, Sweden, 2Karo Bio AB, Novum, Huddinge, Sweden

Abstract. Carrero JJ, Qureshi AR, Axelsson J, Yilmaz

MI, Rehnmark S, Witt MR, Barany P, Heimburger O,

Suliman ME, Alvestrand A, Lindholm B, Stenvinkel

P (Karolinska Institutet, Stockholm; and Karo

Bio AB, Novum, Huddinge; Sweden). Clinical and

biochemical implications of low thyroid hormone

levels (total and free forms) in euthyroid

patients with chronic kidney disease. J Intern Med

2007; 262: 690–701.

Objectives. In this study, we explore the associations

of decreased thyroid hormone levels with inflamma-

tion, wasting and survival in biochemically euthyroid

patients with end-stage renal disease (ESRD).

Design. After exclusion of 23 patients with thyroid-

stimulating hormone (TSH) values outside the normal

range (0.1–4.5 mIU L)1), 187 clinically and biochem-

ically euthyroid incident ESRD stage 5 patients start-

ing dialysis were followed for a median of 20 (range

1–60) months. Measurements of total and free forms

of thyroid hormones, s-albumin, hs-CRP, interleukin

(IL)-6, vascular adhesion molecule (VCAM)-1 and

insulin-like growth factor 1 (IGF-1) were performed

at baseline.

Results. In this population, 17 out of 210 patients (8%)

were defined as subclinically hypothyroid. Multivariate

analysis, according to receiver operating characteristic

(ROC) curves, showed that mortality was best pre-

dicted by total triiodothyronine (T3). When using the

cut-off levels derived from ROC, low T3 levels were

associated with increased inflammation (higher

hs-CRP, IL-6 and VCAM-1) and lower concentration

of both s-albumin and IGF-1. Finally, low T3 but not

low free triiodothyronine was associated with worse

all-cause (Likelihood ratio = 45.4; P < 0.0001) and

cardiovascular mortality (Likelihood ratio = 47.8;

P < 0.0001) after adjustment for confounding factors.

Conclusion. This study showed that low T3 levels are

independent predictors of all-cause and also cardio-

vascular disease mortality in biochemically euthyroid

patients, perhaps due to an intimate association with

inflammation. Based on these results, the use of T3

levels in studies assessing the relationship between

thyroid dysfunction and mortality risk is recom-

mended.

Keywords: chronic kidney disease, inflammation,

mortality, Triiodothyronine, wasting.

Introduction

Chronic kidney disease (CKD) causes alterations in

thyroid hormones in the absence of an underlying

intrinsic thyroid disorder, known as the syndrome of

nonthyroidal illness [1, 2]. This syndrome is mainly

characterized by a decrease in total (T3) and free tri-

iodothyronine (fT3) plasma concentration, whilst thy-

roid-stimulating hormone (TSH) levels are usually

normal. Indeed, Lo et al. [3] recently reported that

approximately one fourth of end-stage renal disease

(ESRD) patients present with low levels of fT3.

690 ª 2007 Blackwell Publishing Ltd

Original Article |

Traditionally, decreases in plasma T3 concentration

have been interpreted as an attempt to conserve body

energy stores by reducing metabolic rate, e.g. in peri-

ods of starvation [4]. There are, however, data indicat-

ing that low T3 and fT3 concentrations are patterns

present in approximately 30% of patients with con-

gestive heart failure (CHF) [5, 6]. As low T3 and fT3

are independent predictors of death in the general

population [7–9], it has been suggested that the thy-

roid system can also play a role in the development

of cardiac disease. Indeed, decreased active thyroid

hormone levels are considered to be part of the

deranged neuroendocrine ⁄proinflammatory system that

is associated with CHF [10]. This proinflammatory

stimuli and in particular interleukin (IL) signalling,

has been reported to downregulate the peripheral con-

version of total thyroxine (T4) into T3 in both experi-

mental [11, 12] and clinical studies [13, 14].

Therefore, low T3 could be associated with pathologi-

cal conditions and maladaptation leading to decreased

survival rather than a sign of physiological adaptation

to energy shortage.

A constellation of wasting, persistent subclinical

inflammation and cardiovascular disease (CVD), pres-

ent in a large proportion of ESRD patients [15] could

be explained in part by these alterations in the thyroid

system, acting as intermediate link between the

inflammatory stress and the impaired cardiovascular

response. This hypothesis was recently presented by

Zoccali et al. [16] in prevalent dialysis patients, show-

ing that subjects with low fT3 levels presented with

increased inflammatory markers and a worse all-cause

mortality when compared to those with higher fT3

levels [16–18]. However, these studies assessed only

fT3 levels, not providing information on the whole

thyroid system. On the other hand, the ‘low T3 syn-

drome’ is known to specifically affect cardiovascular

mortality in the general population [7, 9], yet the

impact of low T3 syndrome in cardiovascular-CKD

related mortality is unknown. Furthermore, subclinical

hypothyroidism has been associated to increased

cardiovascular risk [19]. Previous analyses included

patients with TSH values above the normal range

[16–18], which could have masked some of

the reported findings. Taking into consideration the

possible clinical significance of alterations in thyroid

hormone levels in CKD, we studied total and free lev-

els of thyroid hormones in relation to inflammation,

wasting and survival in incident CKD 5 patients start-

ing dialysis therapy who were followed prospectively

for upto 60 months.

Patients and methods

The Ethics Committee of the Karolinska Institute,

Sweden, approved the study at the Karolinska Univer-

sity Hospital at Huddinge. Informed consent was

obtained from each patient. Incident CKD 5 patients

were investigated as part of an ongoing prospective

cohort study part of which have been published else-

where [15]. In the present study, post hoc analyses

were performed to examine the association between

thyroid hormones, inflammation and mortality. The

study population consisted of 210 consecutive patients

with CKD stage 5 [glomerular filtration rate (GFR)

<15 mL min)1; 63% males, median and range; age

55 (19–70) years] who were enrolled at initiation of

renal replacement therapy in the renal program of the

Karolinska University Hospital Huddinge (Sweden)

between 1994 and 2006. The study exclusion criteria

were age below 18 years or above 70 years, clinical

signs of acute infection, active vasculitis or liver

disease at the time of evaluation, or unwillingness

to participate. The causes of CKD were chronic

glomerulonephritis in 52 patients, diabetic nephro-

pathy in 69 patients, polycystic kidney disease in 23

patients, nephrosclerosis in eight patients and other, or

unknown, etiologies in 58 patients. Thirty-two (15%)

patients were diagnosed as type-1 diabetics whilst 38

patients (18%) were diagnosed as type-2 diabetics.

Most of the patients initially taking oral antiglycemic

agents or on restricted diets had been switched to

insulin therapy at the time of inclusion in the study.

The majority of patients were on antihypertensive

medications (angiotensin-converting enzyme inhibitors

and ⁄or angiotensin II receptor antagonists; n = 107,

betablockers; n = 123, calcium-channel blockers;

n = 73) and other commonly used drugs in CKD,

such as phosphate and potassium binders, diuretics,

erythropoesis-stimulating agents, iron substitution and

vitamin B, C and D supplementation. Only 49

J. J. Carrero et al. | Low total triiodothyronine levels in chronic kidney disease

ª 2007 Blackwell Publishing Ltd Journal of Internal Medicine 262; 690–701 691

patients were on lipid-lowering medication (HMG-

CoA-reductase inhibitors).

None of the patients showed clinical signs of hyper or

hypothyroidisms. However, 187 patients were defined

as biochemically euthyroid, presenting both TSH

(0.1–4.5 mIU L)1) and T4 (57.9–169.9 nmol L)1;

4.5–13.2 lg dL)1) within the normal range [3, 20].

Because subclinical hypothyroidism has been associ-

ated with increased cardiovascular risk [19], 17 (8%)

subclinically hypothyroid patients [TSH > 4.5

mIU L)1 and T4 < 57.9 nmol L)1 (4.5 lg dL)1)] [3,

20], and six subclinically hyperthyroid patients [3, 20]

were excluded from this analysis. For comparative

purposes, analysis including all 210 investigated

subjects will, in some cases, be also presented.

Methods

After an overnight fast, venous blood samples were

drawn and stored at )70 �C for biochemical analyses.

In the patients, GFR was estimated as the mean of

creatinine and urea clearance from a 24 h collection

of urine. Plasma analysis of thyroid hormones were

performed on an Immulite� system (Immulite; DPC

Corp, Los Angeles, CA, USA) using their commer-

cially available immunometric assays for T3 [analyti-

cal sensitivity (AS), 0,54 nmol L)1; total coefficient

of variation (CVs), 13.2% and 5.4% at the levels of

0.95 and 6.02 nmol L)1], fT3 (AS, 0.97 pg mL)1;

total CVs, 7.1% and 4.5% at the levels of 4.43 and

12.63 pg mL)1), T4 (AS, 5 nmol L)1; total CVs,

8.4% and 6.3% at the levels of 49 and

167 nmol L)1), free thyroxine (fT4; AS,

0.31 ng dL)1; total CVs, 9% and 6% at the levels of

0.69 and 2.09 ng dL)1), TSH (AS, 0.004 mIU L)1;

total CVs, 12.5%, and 4.6% at the levels of 0.016

and 1.3 mIU L)1) and thyroxine-binding globulin

(TGB; AS, 1.1 lg mL)1; total CVs, 9.2% and 7.3%

at the levels of 13 and 52 lg mL)1). Results are

expressed as the average of two measurements. The

plasma concentrations of insulin-like growth-factor

(IGF)-1 and serum levels of IL-6 were quantified on

the Immulite� system (Immulite; DPC Corp). Plasma

concentration of vascular adhesion molecule

(VCAM)-1 and levels of high sensitivity (hs)-CRP

were measured using commercially available ELISA

kits (R&D System Inc., Minneapolis, MN, USA). The

levels of serum cholesterol and triacylglycerols were

analysed by standard enzymatic procedures (Roche

Diagnostics GmbH, Mannheim, Germany). The

remaining biochemical analyses were performed using

routine methods at the Department of Clinical Chem-

istry at Huddinge Hospital.

Subjective global assessment (SGA) was used to eval-

uate overall protein-energy nutritional status [21, 22],

and patients presenting with an SGA score of 2–4 were

defined as malnourished. Body mass index (BMI) was

calculated as weight in kg per (height in m)2.

Survival was determined from the day of examination,

with a median follow-up period of 20 (range 1–60)

months. Patients were censored at transplantation or

when completing the follow-up period, with no loss

of follow-up of any patient. Within the follow-up

period, 66 patients died and 73 patients underwent

transplantation. Cardiovascular mortality was defined

as death due to coronary heart disease, sudden death,

stroke or complicated peripheral vascular disease.

Statistical analysis

Results are expressed as mean and standard deviation

or median and range unless otherwise indicated.

A P-value <0.05 was considered to be statistically

significant. Comparisons between two groups were

assessed for continuous variables with the Student’s

unpaired t-test, Mann–Whitney test or Fisher exact test,

as appropriate. As many values were not normally

distributed, Spearman’s rank correlation was used to

determine correlations between thyroid hormones lev-

els and other variables. To evaluate the sensitivity and

specificity of total and free thyroid forms as predictors

of mortality, a receiver operating characteristic (ROC)

analysis was performed [23], using the statistical soft-

ware NCSS 2007 and PASS 2005 (Number Cruncher

Statistical Systems, Keysville, UT, USA). The optimum

cut-off value, with the combination of the highest

sensitivity and specificity, was calculated. Survival

analyses were made with the Kaplan–Meier survival

curve or the Cox proportional hazard model. The

J. J. Carrero et al. | Low total triiodothyronine levels in chronic kidney disease

692 ª 2007 Blackwell Publishing Ltd Journal of Internal Medicine 262; 690–701

relative risks for mortality were determined by multi-

variate Cox regression analysis and presented as hazard

ratio [Hazards ratio (HR); 95% confidence intervals

(CI)]. Finally, stepwise method of variable selection

was employed to choose the ‘best’ Cox proportional

hazard model in the survival analyses for all-cause and

cardiovascular mortality. The statistical analysis was

performed using statistical software SAS version 9.1

(SAS Campus Drive, Cary, NC, USA).

Results

Associations with inflammation and wasting

General characteristics of the study population are

summarized in Table 1. Table 2 shows the Spearman’s

Rank coefficients between serum thyroid hormones

and other relevant biochemical parameters. T3 and

fT3 correlated inversely with serum levels of IL-6 as

well as with hs-CRP and VCAM-1, and were posi-

tively correlated with s-albumin and IGF-1 (Fig. 1).

These associations were stronger for T3 than for fT3

values. Furthermore, when patients were divided

according to the presence of inflammation (defined as

hs-CRP > 10 mg L)1) and ⁄or wasting (defined as

SGA > 1), concentrations of T3 gradually decreased

as none (n = 109; 83.6 ± 16.8 ng dL)1), one (n = 51;

77.7 ± 20.1 ng dL)1) or two (n = 27; 68.5 ± 20.2

ng dL)1) of these conditions were present (P = 0.002

by v2 test).

Thyroid hormones as predictors of mortality usingreceiver operating characteristic curves.

Area under the curve (AUC) and ROC cut-off levels

for different thyroid hormones versus all-cause mor-

tality are shown in Table 3. When only euthyroid

patients are considered (n = 187) only T3, exhibiting

the highest AUC (Fig. 2), was a significant predictor

of all-cause mortality, whilst fT3, T4, fT4 and TSH

Table 1 General characteristics of 187 chronic kidneydisease (CKD) stage 5 euthyroid patients

Parameter Value

Age, year 55 (19–70)

Male, n (%) 119 (63%)

GFR, mL min)1 6.4 ± 2.2

Diabetes mellitus, n (%) 61 (32%)

Cardiovascular disease, n (%) 67 (36%)

Wasting (SGA > 1), n (%) 48 (26%)

BMI, kg m)2 24.4 ± 4.2

S-albumin, g dL)1 3.3 ± 0.6

CRP, mg L)1 5.0 (0.2–218.0)

Interleukin-6, pg mL)1 6.5 (0.7–48.4)

Thyroid hormones

T3, ng dL)1 79.5 ± 19.1

fT3, pg mL)1 1.9 ± 0.6

T4, lg dL)1 5.9 ± 1.7

fT4, ng dL)1 1.1 ± 0.2

TSH, mIU L)1 1.5 ± 0.9

TBG, ng mL)1 16.7 ± 3.9

Note: To convert T3 in ng dL)1 to nmol L)1, multiply by 0.01536;fT3 in pg mL)1 to pmol L)1, multiply by 0.01536; T4 in lg dL)1

to nmol L)1, multiply by 12.87; fT4 in ng dL)1 to pmol L)1, multi-ply by 12.87.Data are presented as media ± SD or median (range) unless other-wise stated. TBG, Thyroxine binding globulin; TSH, thyroid-stimu-lating hormone.

Table 2 Spearman’s Rank coeffi-cients between serum thyroidhormones and other relevantparameters in 187 euthyroidCKD 5 patients

IL-6 hs-CRP VCAM-1 Cholesterol S-Albumin Haemoglobin IGF-1a

T3 )0.36e )0.32e )0.19c ns 0.32e 0.29e 0.21c

fT3 )0.29e )0.21c )0.20c ns 0.35e 0.15b 0.18b

T4 )0.15b ns ns )0.14b 0.23e 0.20d ns

fT4 )0.20c ns ns )0.10 0.31e 0.17b ns

TSH ns ns ns ns ns ns ns

TBG )0.17c ns ns ns ns ns )0.16b

ns, not significant; TBG, Thyroxine binding globulin; TSH, thyroid-stimulating hormone; VCAM, vas-cular adhesion molecule; IGF-1, insulin-like growth factor 1. an = 129. Data is presented as rho valuesand superscript letters indicating statistical significance: bP < 0.05; cP < 0.01; dP < 0.001;eP < 0.0001.

J. J. Carrero et al. | Low total triiodothyronine levels in chronic kidney disease

ª 2007 Blackwell Publishing Ltd Journal of Internal Medicine 262; 690–701 693

were not significantly associated with death. For this

reason, further analyses were restricted to T3 only.

Interestingly, when all the investigated subjects

(n = 210) were included both T3 and fT3, were sig-

nificant predictors of all-cause mortality and T4 levels

were borderline significant (P � 0.05).

Comparison of patients with low or high T3 levels

The patient’s material was divided into two cohorts,

based on cut-off levels of T3 derived from the ROC

analysis: T3 £ 78.5 ng dL)1 (1.2 nmol L)1; n = 88),

classified as ‘Low T3’ and T3 > 78.5 ng dL)1

(1.2 nmol L)1; n = 99), classified as ‘High T3’. In

Table 4 risk factors in ESRD euthyroid patients are

presented in relation to this division. Patients with

low T3 levels exhibited a higher inflammatory status

and endothelial dysfunction, measured as higher levels

of hs-CRP, IL-6 and VCAM-1. Patients with low T3

also showed a worse nutritional status, presented as

lower levels of s-albumin, Hb and IGF-1.

All-cause and cardiovascular mortality

During the follow-up period 66 patients died, 34

of them (i.e. 54% of total deaths) of cardiovascular

0.4 0.8 4.0 8.0 40.0 80.00.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0hs-CRP mg L–1

T3

nm

ol L

–1

Rho = –0.32 ; p < 0.0001

100200

300400

500600

7000.5

0.6

0.7

0.8

0.9

1.0

1.1

1.21.31.41.51.61.71.81.92.0

IGF-1, ng L–1

T3

nm

ol L

–1

Rho = 0.21 ; p < 0.0001

1 1 2 4 6 8 20 40 60

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0IL-6, mg L–1

Rho = –0.36 ; p < 0.0001

5 10 15 20 25 30 35 40 45 500.5

0.6

0.7

0.8

0.9

1.0

1.1

1.21.31.41.51.61.71.81.92.0

Albumin, g L–1

Rho = 0.32 ; p < 0.0001

Fig. 1 Spearman rank correlation between total triiodothyronine (T3) levels and markers of inflammatory (IL-6 and hs-CRP)and nutritional (s-albumin and IGF-1) status in 187 chronic kidney disease (CKD) stage 5 pre-dialysis euthyroid patients.n = 129 patients for IGF-1; to convert T3 in nmol L)1 to ng dL)1, divide by 0.01536; albumin in g L)1 to g dL)1, divideby 10.

J. J. Carrero et al. | Low total triiodothyronine levels in chronic kidney disease

694 ª 2007 Blackwell Publishing Ltd Journal of Internal Medicine 262; 690–701

complications. Survival curves for all-cause mortality

in euthyroid patients (n = 187) are shown in Fig. 3.

In the non-adjusted analysis, low plasma T3 (Log-

Rank (v2) = 9.41; P = 0.002), but not low plasma fT3

(Log-Rank (v2) = 1.43; ns), was associated with a

higher all-cause mortality (Fig. 4, Panel A), even after

adjustment for confounding factors (age, gender, pres-

ence of diabetes mellitus, CVD and inflammation

(hs-CRP > 10 mg mL)1); Fig. 4, Panel B; Likelihood

ratio for low T3 = 45.4; P < 0.0001). Low plasma T3

concentration was associated to an increased all-cause

mortality (HR 2.0; 95% CI: 1.1–3.5, P = 0.02).

Interestingly, when all the investigated patients

(n = 210, including those biochemically non-euthy-

roid) were considered, both low plasma T3 (Log-Rank

(v2) = 14.9; P < 0.0001) and low plasma fT3 (Log-

Rank (v2) = 3.6; P < 0.05) had an impact on survival,

that persisted after adjustment for confounding factors

(Likelihood ratio for low T3 = 35.9; P < 0.0001 and

Likelihood ratio for low fT3 = 35.7; P < 0.0001). In

this scenario, presenting with low T3 concentrations

conferred a risk of death 3.0 higher (HR; 95%

CI: 1.7–5.3, P = 0.0002).

Figure 4 shows similar results in the survival curves

for cardiovascular mortality in biochemically euthy-

roid patients. In the nonadjusted analysis, low plasma

T3 (Figure 4, Panel A. Log-Rank (v2) = 6.13;

P = 0.01), but not low plasma fT3 (Log-Rank

(v2) = 2.30; ns) was associated with increased cardio-

vascular mortality (Figure 4, Panel A) that persisted

Table 3 Area under the curve(AUC) and cut-off values ofreceiver operating characteristic(ROC) curves for prediction ofall-cause mortality by thyroidhormone levels in CKD stage 5patients starting renal replace-ment therapy

Curve AUC SE P-value Cut-off level Sensitivity Specificity

Euthyroid patients only (n = 187)

T3, ng dL)1 0.65 0.04 0.001 78.5 64.2% 60%

fT3, pg mL)1 0.58 0.05 0.06 1.9 56.9% 54.7%

T4, lg dL)1 0.56 0.04 NS 6.1 54.7% 51.5%

fT4, ng dL)1 0.52 0.04 NS 1.1 59.2% 46.0%

TSH, mIU mL)1 0.54 0.04 NS 1.3 58.5% 50.7%

All patients (n = 210)

T3, ng dL)1 0.69 0.05 <0.0001 78.3 64.0% 63.1%

fT3, pg mL)1 0.61 0.05 0.01 1.9 62.0% 54.4%

T4, lg dL)1 0.58 0.05 0.05 76.0 56.3% 54.4%

fT4, ng dL)1 0.52 0.05 NS 1.1 58.3% 46.3%

TSH, mIU mL)1 0.54 0.05 NS 1.4 52.9% 52.3%

Note. To convert T3 in ng dL)1 to nmol L)1, multiply by 0.01536; fT3 in pg mL)1 to pmol L)1, mul-tiply by 0.01536; T4 in lg dL)1 to nmol L)1, multiply by 12.87; fT4 in ng dL)1 to pmol L)1, multi-ply by 12.87.SE, standard error; NS, non significant; TSH, thyroid-stimulating hormone.

Sen

sitiv

ity

T3

T4

fT3

1.00

0.75

0.50

0.25

0.001.000.750.500.250.00

1-Specificity

ROC Curve

Fig. 2 Receiver operating characteristic (ROC) curves ver-sus all-cause mortality generated with serum total (T3) andfree triiodothyronine (fT3) in 187 chronic kidney disease(CKD) stage 5 euthyroid pre-dialysis patients. Only T3(P = 0.001) levels were significant predictors of all-causemortality, showing the highest area under the curve.

J. J. Carrero et al. | Low total triiodothyronine levels in chronic kidney disease

ª 2007 Blackwell Publishing Ltd Journal of Internal Medicine 262; 690–701 695

after adjusting for age, gender, presence of diabetes

mellitus and inflammation (Figure 4, Panel B; Likeli-

hood ratio = 47.8; P < 0.0001). In this adjusted

model, low plasma T3 concentrations was related to

almost threefold increased cardiovascular mortality

risk (HR 2.7; 95% CI: 1.2–6.3, P = 0.02).

When all the investigated subjects (n = 210) were

included in the survival analysis, both low T3 and fT3

were associated with increased cardiovascular mortality

in an unadjusted (Log-Rank (v2) for T3 = 16.25;

P < 0.0001; Log-Rank (v2) for fT3 = 4.30; P < 0.03)

and adjusted (Likelihood ratio for T3 = 37.6;

P < 0.0001. Likelihood ratio for fT3 = 36.5;

P < 0.0001) analysis. In this model, presenting with

low T3 concentrations conferred a risk of death 4.0

higher (HR; 95% CI: 1.8–8.8, P = 0.0006).

Stepwise Cox regression analysis was used to iden-

tify predictors of all-cause and cardiovascular mortal-

ity in the euthyroid population (Table 5). For

predictors of all-cause mortality (Table 5, Panel A),

the presence of diabetes, CVD, low T3 levels and

s-albumin levels were stepwise included. In this

analysis, euthyroid patients with lower plasma T3

levels, had a fully adjusted relative risk for all-cause

death almost two times higher (HR: 1.9, 95% CI:

1.1–3.4; P = 0.02) than that observed in those with

higher T3 values.

For predictors of cardiovascular mortality (Table 5,

Panel B), the presence of diabetes, low T3 levels and

previous history of CVD were stepwise included, and

the euthyroid patients with lower plasma T3 levels,

had a fully adjusted relative risk for cardiovascular

Table 4 Characteristics of theCKD stage 5 euthyroid patientssubdivided into low and high T3plasma concentrations (accordingto ROC analysis)

Parameter Low T3 (n = 88) High T3 (n = 99) P-value

General

Age, year 55 (23–70) 55 (19–70) ns

Male, n (%) 61 (69%) 58 (59%) ns

GFR, mL min)1 6.5 ± 2.2 6.3 ± 2.2 ns

Diabetes mellitus, n (%) 33 (37%) 28 (29%) ns

Cardiovascular disease, n (%) 35 (40%) 32 (32%) ns

Nutritional markers

Wasting (SGA > 1), n (%) 25 (30%) 23 (24%) ns

BMI, kg m)2 24.8 ± 4.1 23.9 ± 4.3 ns

Haemoglobin, g dL)1 10.1 ± 1.3 10.8 ± 1.5 0.001

s-albumin, g dL)1 3.1 ± 0.6 3.5 ± 0.6 <0.0001

IGF-1, ng mL)1a 146 (27–468) 191 (41–562) 0.03

Inflammation, adhesion molecules and oxidative stress biomarkers

CRP, mg L)1 8.1 (0.3–163.0) 3.1 (0.2–218.0) <0.0001

IL-6, pg mL)1 8.9 (1.1–42.6) 4.7 (0.7–48.4) <0.0001

sVCAM-1, ng mL)1 1412 (672–4238) 1301 (546–2521) <0.05

Thyroid hormones

T4, lg dL)1 5.1 ± 1.6 6.7 ± 1.5 <0.0001

fT4, ng dL)1 0.9 ± 0.2 1.2 ± 0.2 0.0001

TSH, mIU L)1 1.6 ± 1.0 1.4 ± 0.9 <0.05

TBG, ng mL)1 15.9 ± 4.3 17.3 ± 3.4 <0.01

Note: To convert T3 in ng dL)1 to nmol L)1, multiply by 0.01536; fT3 in pg mL)1 to pmol L)1,multiply by 0.01536; T4 in lg dL)1 to nmol L)1, multiply by 12.87; fT4 in ng dL)1 to pmol L)1,multiply by 12.87.Data are presented as media±SD or median (range) unless otherwise stated; ns, non significant.a68 ⁄ 61 patients.

J. J. Carrero et al. | Low total triiodothyronine levels in chronic kidney disease

696 ª 2007 Blackwell Publishing Ltd Journal of Internal Medicine 262; 690–701

death three times higher (HR: 3.1, 95% CI: 1.4–7.1;

P = 0.006) than that observed in those with higher

values.

When all patients (n = 210) were considered in the

stepwise regression, the hazard ratio for patients pre-

senting with low T3 levels was in general higher: In

100

80

60

40

Pat

ient

s su

rviv

al, %

20

00 365 730 1095

Days

1460

Low T3

High T3

1825

100

80

60

40

Pat

ient

s su

rviv

al, %

20

00 365 730 1095

Days

1460

Low T3

High T3

1825

Log-Rank (χ2) = 9.41

p = 0.002

Likelihood ratio = 45.4

p < 0.0001

(b)

(a)

Fig. 3 Unadjusted (Panel A) and adjusted (Panel B) all-cause mortality according to plasma levels of T3 in 187CKD stage 5 euthyroid pre-dialysis patients. Adjustment forpotential confounding factors [age, gender, presence of dia-betes mellitus, cardiovascular disease and inflammation(hs-CRP > 10 mg mL)1)] was done with the Cox propor-tional hazards model. Cut-off value for T3 was assessedfrom the ROC analysis.

00

20

40

60

80High T3

Pat

ient

s su

rviv

al, %

Pat

ient

s su

rviv

al, %

Low T3

100

(b)

(a)

0

20

40

60

80

100

365 730 1095

Days

1460 1825

0 365 730 1095Days

1460 1825

High T3

Low T3

Log-Rank (χ2) = 6.13

p = 0.01

Likelihood ratio = 47.8

p < 0.0001

Fig. 4 Unadjusted (Panel A) and adjusted (Panel B) cardio-vascular mortality according to plasma levels of T3 in 187chronic kidney disease (CKD) stage 5 euthyroid pre-dialysispatients. Adjustment for potential confounding factors (age,gender, presence of diabetes mellitus, cardiovascular diseaseand inflammation [hs-CRP > 10 mg mL)1]) was done withthe Cox proportional hazards model. Cut-off value for T3was assessed from the ROC analysis.

J. J. Carrero et al. | Low total triiodothyronine levels in chronic kidney disease

ª 2007 Blackwell Publishing Ltd Journal of Internal Medicine 262; 690–701 697

this analysis, diabetes (HR 3.2, 95% CI: 1.6–6.3, P <

0.0001), age (HR 3.1, 95% CI: 1.9–5.8, P = 0.002) and

low T3 (HR 3.3, 95% CI: 1.9–5.8, P = 0.0002) were

independent predictors of all-cause mortality, whereas

diabetes (HR 4.2, 95% CI: 1.9–9.0, P = 0.0001), CVD

(HR 3.7, 95% CI: 1.6–8.3, P = 0.0005) and low T3

(HR 4.2, 95% CI: 1.9–9.1, P = 0.0005) independently

predicted cardiovascular mortality. In both cases, and

because the univariate association of T3 was stronger

for IL-6 than for hs-CRP, we repeated the stepwise Cox

regression and the survival analysis adjusting instead

for IL-6 levels as a marker of inflammation (above ver-

sus below 6.8 pg mL)1 as defined from ROC). Both

models remained significant (not shown).

Discussion

In the present study we compared, for the first time, the

prognostic value of different thyroid hormones (both

total and free forms) on outcome in biochemically

euthyroid nondialysed CKD patients. Results showed

that T3 levels, but not fT3, appear to be an independent

predictor of not only all-cause but also CVD mortality

perhaps due to an intimate association with inflamma-

tion. The novelty of the present analysis is that low

T3 levels, even in CKD patients with normal thyroid

status, contribute to mortality in this patient population.

Low T3 – associations with inflammation and wasting

Low T3 levels showed a stronger inverse correlation

with IL-6 and hs-CRP than fT3 did. Low T3 levels

are common in ESRD patients [3] and may not only

be a physiological adaptation to chronic illness that

minimize catabolism but perhaps also directly related

to the progression of CHF [5–9]. Several reasons have

been proposed for these low levels in CKD, including

iodine accumulation in the thyroid gland, increased

protein binding and peripheral tissue metabolism as

well as altered tissue concentrations and availability

[2]. Additionally, several studies have shown that pro-

inflammatory cytokines may inhibit T3 production or

increase tissue turnover [11–14, 24]. A proposed

mechanistic rationale for this effect is that cytokine-

induced competition for limited amounts of coactivators

decreases hepatic type I iodothyronine 5¢-deiodinaseexpression, resulting in decrease T3 production [25].

This is relevant because persistent inflammation and

wasting are common amongst CKD patients and prob-

ably contribute to CVD [15, 26].

Additionally, inflammation in CKD is associated with

a state of protein-energy wasting [15, 26], both hav-

ing additive effects on mortality [26, 27]. In our

study, T3 levels were, indeed, incrementally lower

when inflammation, wasting or both conditions were

present. However, the association between thyroid

hormone levels and wasting was not so evident since

no difference was found in the percentage of wasted

patients or BMI levels when comparing low versus

high-T3 levels. The lower serum albumin concentra-

tion in the low-T3 group could be more related to

inflammation than to wasting per se [28]. Further-

more, in agreement with a previous study [29], we

Table 5 Stepwise Cox regressionfor predictors of all-cause (A)and cardiovascular mortality (B)in 187 biochemically euthyroidpatientsa

Variable Hazard ratio v2 P-value

(A) All-cause mortality

Diabetes mellitus (yes versus no) 2.1 (1.4–4.5) 5.42 0.01

Cardiovascular disease (yes versus no) 2.5 (1.4–4.5) 9.73 0.001

T3 (below versus above 78.3 ng dL)1) 1.9 (1.1–3.4) 4.74 0.02

S-albumin (below versus above 3.3 g dL)1) 2.1 (1.1–4.0) 5.51 0.01

(B) CVD mortality

Diabetes mellitus (yes versus no) 3.9 (1.6–9.1) 10.1 0.001

T3 (below versus above 78.3 ng dL)1) 3.1 (1.4–7.1) 7.5 0.006

Cardiovascular disease (yes versus no) 5.1 (1.9–12.6) 11.7 0.0006

aBoth models also included the following variables that were stepwise excluded: gender (female versusmale), age (above versus below 55 years) and inflammation (hs-CRP above versus below 10 mg L)1).

J. J. Carrero et al. | Low total triiodothyronine levels in chronic kidney disease

698 ª 2007 Blackwell Publishing Ltd Journal of Internal Medicine 262; 690–701

found an inverse association between T3 and IGF-1,

whose concentrations might be a valid and novel

nutritional marker relating to lean body mass for dial-

ysis patients [30]. However, an appropriate cut-off

point of IGF-1 to discriminate wasting from non-

wasting is not yet defined. As IGF-1 is part of the

major anabolic pathway mediating growth hormone in

skeletal muscle and influenced by thyroid hormone

actions [31], it is not surprising that a state of wasting

also relates to low plasma concentrations of IGF-1,

also as a reflection of a hypothyroid state.

Low T3 – a sensitive predictor of mortality in CKD

The ROC curves showed that T3 appeared as a

more sensitive predictor than fT3, with an AUC

considered to represent a ‘fair predictor’ [32]. This

observation might have important clinical implica-

tions, as T3 measurement is a simple, inexpensive,

and reliable blood test that can be readily measured

and is routinely implanted in many units. In fact,

Pingitore et al. [9] showed that T3 might be a better

prognostic factor for CHF than fT3, specially in crit-

ically ill patients where frequent heparinization inter-

feres with assays for fT3 and fT4 [33], as occurs in

HD patients [34]. This issue, together with the lower

specificity of the radio immunoassay (RIA) methods

[35] and the false increases in fT3 in stored samples

[36] could explain why the fT3 levels reported by

Zoccali et al. [16–18] were slightly elevated when

compared with previous reports [3, 37] and our

study. Thyroxine binding globulin (TBG) is a nega-

tive acute phase reactant [37] and probably contrib-

utes to explain the stronger associations between T3

and markers of inflammation and wasting. However,

TBG was not related to s-albumin levels in our

study. Furthermore, both s-albumin and low T3 lev-

els were stepwise included in the all-cause mortality

Cox regression analysis, suggesting that other

pathways might be contributing to the observed

relations.

We also show for the first time that low T3 levels are

independent predictors of cardiovascular mortality in

CKD patients. This is consistent with the adverse prog-

nosis associated with increased peripheral production

of biologically inactive reverse T3 after acute myo-

cardial infarction in the general population [38], but it

had never been described in the CKD population

before. The recent observation by Trivieri et al. [39]that thyroid hormones specifically enhance contractility

and prevent pressure overload-induced cardiac dys-

function, might provide a mechanistic explanation on

how low T3 levels can contribute to the development

of CHF. It also supports the observed association

between low T3 levels and left ventricular hypertrophy

in HD patients [40].

Previous studies in CKD by Zoccali et al. [16–18]

included patients with TSH values above the normal

range. As many, but not all studies, have associated a

state of subclinical hypothyroidism with increased car-

diovascular risk [19], we restricted our analysis to

only biochemically euthyroid patients [3, 20]. Thus,

the inclusion of subclinically hypothyroid patients

could have masked some of the previously reported

findings. However, even following the exclusion of

these patients, we confirm the association between

low thyroid hormones, inflammation and increased

mortality in the CKD population [16–18]. In addition,

we believe that our analysis, in a more restricted pop-

ulation of incident CKD patients, adds important

information to the literature as it implies a higher risk

in patients with normal thyroid values. When bio-

chemically noneuthyroid patients are included in our

analysis (that is, all studied patients with a similar dis-

tribution of subclinical hypothyroidism (8%) as

reported by Zoccali et al. [16–18]), low fT3 levels

showed indeed a clear impact on all-cause and cardio-

vascular mortality, with a prognostic value similar

than that attributed to T3. However, when only bio-

chemically euthyroid patients were studied, the prog-

nostic value for fT3 was lost and the impact of low

T3 was in general lower. The loss of prediction power

of fT3 in analyses restricted to biochemically euthy-

roid patients is difficult to explain and needs to be

confirmed. It could be speculated that in this patient

group the superior predictive power of T3 in bio-

chemically euthyroid patients is due to the fact that

circulating hormone and the carrier protein might cap-

ture greater prognostic information, possibly beyond

thyroid function per se.

J. J. Carrero et al. | Low total triiodothyronine levels in chronic kidney disease

ª 2007 Blackwell Publishing Ltd Journal of Internal Medicine 262; 690–701 699

Limitations of the study and future perspectives

To properly address the implications of the present

study, several limitations should be taken into account.

First, the classification of CVD and cardiovascular

mortality included only patients with clinically signifi-

cant disease, which may limit and underestimate the

true prevalence of CVD [41]. Moreover, measurements

of the peripheral concentration of thyroid hormones do

not describe the intracellular thyroid hormone effect

on the different intracellular receptor sites. Clearly, we

need to further elucidate the role of thyroid dysregula-

tion in CKD. Although the use of thyroid hormone

therapy in nonthyroidal illnesses is still controversial

[42], interventional studies designed to reduce CVD in

renal disease is a tempting idea. In this regard, it

should be emphasized that a recent 8-week dietary

intervention with low protein diet in CKD patients

decreased levels of proinflammatory cytokines with

concomitant increases of both T3 and T4 [43].

Conclusion

About 8% of the clinically euthyroid incident CKD

five patients showed biochemical signs of subclinical

hypothyroidism [3, 20]. Furthermore, the present

study showed that T3 levels appear to be an indepen-

dent predictor of not only all-cause but also CVD

mortality in biochemically euthyroid patients, perhaps

due to its intimate association with inflammation.

Based on these results, the use of T3 in studies

assessing the relationship between thyroid dysfunction

and mortality risk is recommended.

Conflict of interest statement

BL is employed by Baxter Healthcare. SR and MRW

are employed by Karo Bio AB. The other authors

declare no conflict of interest.

Acknowledgements

The authors wish to express their appreciation to Ann

Dreiman-Lif, Annika Nilsson and Anki Emmoth (clini-

cal studies), Dr Bjorn Anderstam, Monica Eriksson and

Ann-Christin Bragfors-Helin (biochemical analysis).

This study was partially supported by unrestricted

grants from the Swedish Research Council, Heart-

Lung Foundation, MEC (EX2006-1670), Baxter

Healthcare, Karolinska Institutet Center for Gender-

based Research, the General Export Association of

Sweden and the ERA-EDTA. Analyses of thyroid

hormones were supported by Karo Bio AB, Sweden.

References

1 Chopra IJ. Nonthyroidal illness syndrome or euthyroid sick

syndrome? Endocr Pract 1996; 2: 45–52.

2 Kaptein EM. Thyroid hormone metabolism and thyroid diseases

in chronic renal failure. Endocr Rev 1996; 17: 45–63.

3 Lo JC, Chertow GM, Go AS, Hsu CY. Increased prevalence of

subclinical and clinical hypothyroidism in persons with chronic

kidney disease. Kidney Int 2005; 67: 1047–52.

4 Douyon L, Schteingart DE. Effect of obesity and starvation on

thyroid hormone, growth hormone, and cortisol secretion.

Endocrinol Metab Clin North Am 2002; 31: 173–89.

5 Manowitz NR, Mayor GH, Klepper MJ, DeGroot LJ. Subclini-

cal Hypothyroidism and Euthyroid Sick Syndrome in Patients

with Moderate-to-Severe Congestive Heart Failure. Am J Ther

1996; 3: 797–801.

6 Kozdag G, Ural D, Vural A et al. Relation between free triiodo-

thyronine ⁄ free thyroxine ratio, echocardiographic parameters

and mortality in dilated cardiomyopathy. Eur J Heart Fail

2005; 7: 113–8.

7 Iervasi G, Pingitore A, Landi P et al. Low-T3 syndrome: a

strong prognostic predictor of death in patients with heart

disease. Circulation 2003; 107: 708–13.

8 Hamilton MA, Stevenson LW, Luu M, Walden JA. Altered

thyroid hormone metabolism in advanced heart failure. J Am

Coll Cardiol 1990; 16: 91–5.

9 Pingitore A, Landi P, Taddei MC, Ripoli A, L’Abbate A, Iervasi

G. Triiodothyronine levels for risk stratification of patients with

chronic heart failure. Am J Med 2005; 118: 132–6.

10 Opie LH. Cellular basis for therapeutic choices in heart failure.

Circulation 2004; 110: 2559–61.

11 Ozawa M, Sato K, Han DC, Kawakami M, Tsushima T,

Shizume K. Effects of tumor necrosis factor-alpha ⁄ cachectin on

thyroid hormone metabolism in mice. Endocrinology 1988; 123:

1461–7.

12 Torpy DJ, Tsigos C, Lotsikas AJ, Defensor R, Chrousos GP,

Papanicolaou DA. Acute and delayed effects of a single-dose

injection of interleukin-6 on thyroid function in healthy humans.

Metabolism 1998; 47: 1289–93.

13 Bartalena L, Brogioni S, Grasso L, Velluzzi F, Martino E. Rela-

tionship of the increased serum interleukin-6 concentration to

changes of thyroid function in nonthyroidal illness. J Endocri-

nol Invest 1994; 17: 269–74.

14 Van der Poll T, Romijn JA, Wiersinga WM, Sauerwein HP.

Tumor necrosis factor: a putative mediator of the sick euthyroid

syndrome in man. J Clin Endocrinol Metab 1990; 71: 1567–72.

J. J. Carrero et al. | Low total triiodothyronine levels in chronic kidney disease

700 ª 2007 Blackwell Publishing Ltd Journal of Internal Medicine 262; 690–701

15 Stenvinkel P, Heimburger O, Paultre F et al. Strong association

between malnutrition, inflammation, and atherosclerosis in

chronic renal failure. Kidney Int 1999; 55: 1899–911.

16 Zoccali C, Tripepi G, Cutrupi S, Pizzini P, Mallamaci F. Low

triiodothyronine: a new facet of inflammation in end-stage renal

disease. J Am Soc Nephrol 2005; 16: 2789–95.

17 Zoccali C, Mallamaci F, Tripepi G, Cutrupi S, Pizzini P. Low

triiodothyronine and survival in end-stage renal disease. Kidney

Int 2006; 70: 523–8.

18 Enia G, Panuccio V, Cutrupi S et al. Subclinical hypothyroidism

is linked to micro-inflammation and predicts death in continuous

ambulatory peritoneal dialysis. Nephrol Dial Transplant 2007;

22: 538–44.

19 Surks MI, Ortiz E, Daniels GH et al. Subclinical thyroid dis-

ease: scientific review and guidelines for diagnosis and manage-

ment. JAMA 2004; 291: 228–38.

20 Hollowell JG, Staehling NW, Flanders WD et al. Serum TSH,

T(4), and thyroid antibodies in the United States population

(1988 to 1994): National Health and Nutrition Examination

Survey (NHANES III). J Clin Endocrinol Metab 2002; 87:

489–99.

21 Qureshi AR, Alvestrand A, Danielsson A et al. Factors predict-

ing malnutrition in hemodialysis patients: a cross-sectional

study. Kidney Int 1998; 53: 773–82.

22 Detsky AS, McLaughlin JR, Baker JP et al. What is subjective

global assessment of nutritional status? JPEN J Parenter

Enteral Nutr 1987; 11: 8–13.

23 Obuchowski NA. Receiver operating characteristic curves and

their use in radiology. Radiology 2003; 229: 3–8.

24 Boelen A, Platvoet-Ter Schiphorst MC, Wiersinga WM. Associ-

ation between serum interleukin-6 and serum 3,5,3¢-triiodothyro-nine in nonthyroidal illness. J Clin Endocrinol Metab 1993; 77:

1695–9.

25 Yu J, Koenig RJ. Regulation of hepatocyte thyroxine 5¢-deiodin-ase by T3 and nuclear receptor coactivators as a model of the

sick euthyroid syndrome. J Biol Chem 2000; 275: 38296–301.

26 Qureshi AR, Alvestrand A, Divino-Filho JC et al. Inflammation,

malnutrition, and cardiac disease as predictors of mortality in

hemodialysis patients. J Am Soc Nephrol 2002; 13(Suppl 1):

S28–36.

27 Avesani CM, Carrero JJ, Axelsson J, Qureshi AR, Lindholm B,

Stenvinkel P. Inflammation and wasting in chronic kidney

disease: Partners in crime. Kidney Int Suppl 2006; 70: S8–13.

28 Heimburger O, Qureshi AR, Blaner WS, Berglund L, Stenvin-

kel P. Hand-grip muscle strength, lean body mass, and plasma

proteins as markers of nutritional status in patients with chronic

renal failure close to start of dialysis therapy. Am J Kidney Dis

2000; 36: 1213–25.

29 Iglesias P, Bayon C, Mendez J, Gancedo PG, Grande C, Diez

JJ. Serum insulin-like growth factor type 1, insulin-like growth

factor-binding protein-1, and insulin-like growth factor-binding

protein-3 concentrations in patients with thyroid dysfunction.

Thyroid 2001; 11: 1043–8.

30 Axelsson J, Qureshi AR, Divino-Filho JC et al. Are insulin-like

growth factor and its binding proteins 1 and 3 clinically useful

as markers of malnutrition, sarcopenia and inflammation in

end-stage renal disease? Eur J Clin Nutr 2006; 60: 718–26.

31 Hussain MA, Schmitz O, Jorgensen JO et al. Insulin-like

growth factor I alters peripheral thyroid hormone metabolism in

humans: comparison with growth hormone. Eur J Endocrinol

1996; 134: 563–7.

32 Swets JA. Measuring the accuracy of diagnostic systems.

Science 1988; 240: 1285–93.

33 Peeters RP, Van der Geyten S, Wouters PJ et al. Tissue thyroid

hormone levels in critical illness. J Clin Endocrinol Metab

2005; 90: 6498–507.

34 Sonawane S, Kasbekar N, Berns JS. The safety of heparins in

end-stage renal disease. Semin Dial 2006; 19: 305–10.

35 Squire CR. Methods for the investigation of thyroid function.

Methods Mol Biol 2006; 324: 91–108.

36 Kilinc AS, Duzoylum A, Uncugil CF, Yucel D. Falsely

increased free triiodothyronine in sera stored in serum separator

tubes. Clin Chem 2002; 48: 2296–7.

37 Lim VS. Thyroid function in patients with chronic renal failure.

Am J Kidney Dis 2001; 38: S80–4.

38 Friberg L, Werner S, Eggertsen G, Ahnve S. Rapid down-regu-

lation of thyroid hormones in acute myocardial infarction: is it

cardioprotective in patients with angina? Arch Intern Med 2002;

162: 1388–94.

39 Trivieri MG, Oudit GY, Sah R et al. Cardiac-specific elevations

in thyroid hormone enhance contractility and prevent pressure

overload-induced cardiac dysfunction. Proc Natl Acad Sci U S A

2006; 103: 6043–8.

40 Zoccali C, Benedetto F, Mallamaci F et al. Low triiodothyro-

nine and cardiomyopathy in patients with end-stage renal

disease. J Hypertens 2006; 24: 2039–46.

41 Lauer MS, Blackstone EH, Young JB, Topol EJ. Cause of death

in clinical research: time for a reassessment? J Am Coll Cardiol

1999; 34: 618–20.

42 Bartalena L. The dilemma of non-thyroidal illness syndrome: to

treat or not to treat? J Endocrinol Invest 2003; 26: 1162.

43 Rosolowska-Huszcz D, Kozlowska L, Rydzewski A. Influence

of low protein diet on nonthyroidal illness syndrome in chronic

renal failure. Endocrine 2005; 27: 283–8.

Correspondence: Peter Stenvinkel, Division of Renal Medicine,

K56, Karolinska University Hospital at Huddinge, 141 86 Stock-

holm, Sweden.

(fax: +46 8 7114742; e-mail: peter.stenvinkel@ki.se).

J. J. Carrero et al. | Low total triiodothyronine levels in chronic kidney disease

ª 2007 Blackwell Publishing Ltd Journal of Internal Medicine 262; 690–701 701