0021-972X/90/7105-1239$02.00/0 Journal of Clinical Endocrinology and Metabolism Copyright© 1990 by The Endocrine Society
Vol. 71, No. 5 Printed in U.S.A.
Decreased Serum Triiodothyronine Is Associated with Increased Concentrations of Tumor Necrosis Factor ARSHAG D. MOORADIAN, RICHARD L. REED, DAN OSTERWEIL, RONALD SCHIFFMAN, AND PHILIP SCUDERI Division of Restorative Medicine (A.D.M., R.L.R.) and Veterans Administration Medical Center (A.D.M., R.S.), the Department of Microbiology and Immunology, University of Arizona College of Medicine (P.S.), Tucson, Arizonia 85719; and the Department of Medicine, University of California School of Medicine (D.O.), Los Angeles, California 90024
ABSTRACT. Previous studies in laboratory animals have shown that tumor necrosis factor-a (TNF) may alter thyroid function tests. To determine whether elevated serum TNF levels are associated with altered serum concentrations of T4, T3, free T4, rT3, and TSH, we measured these parameters in 29 nursing home residents with detectable serum TNF levels and compared the levels to those found in 36 patients with undetectable serum TNF levels. The 2 groups were matched for age, sex, clinical problems, use of medications, and nutritional status. Patients with detectable serum TNF levels had significantly lower serum T3 concentrations compared to those with undetectable levels
P
ROFOUND changes in thyroid function tests occur in patients with nonthyroidal illness (1, 2). The mechanism underlying these changes is probably multifactorial. Recent studies in laboratory animals have indicated that tumor necrosis factor-a (TNF) may have a pathogenetic role in altered thyroid hormone economy associated with illness (3, 4). Administration of TNF to rats and mice results in reduced thyroid radioiodine uptake, reduced response to TSH, and low serum concentrations of T4, T3, and rT 3 (3, 4). In addition, it has been shown that there are specific TNF receptors in human and rat thyroid cells that are regulated by TSH (5). It has been suggested that TNF may be a mediator of the changes in thyroid function tests seen in nonthyroidal illness (3-5). It has also been shown that administration of interleukin-1 (IL-1), -a or -/3, to mice will alter thyroid hormone metabolism in a manner similar to the effect of TNF (6). The clinical relevance of these observations, however, has not been addressed. If these findings in laboratory animals can be extrapolated to human subjects, one would expect that serum thyroid hormone levels in patients with elevated serum TNF will Received March 1, 1990. Address all correspondence and requests for reprints to: Arshag D. Mooradian, University of Arizona, 1821 East Elm Street, Tucson, Arizonia 85719.
[1.072 ± 0.588 vs. 1.621 ± 0.594 nmol/L (mean ± SD); P < 0.01]. Differences in other tests did not achieve statistical significance. Thyroid function tests were not significantly different when patients with detectable interleukin-la levels, another cytokine secreted during endotoxemia, were compared to those with undetectable levels. These observations taken together with the previous findings in laboratory animals suggest that some of the alterations in thyroid hormone levels seen in nonthyroidal illness are associated with elevated serum concentrations of TNF. (J Clin Endocrinol Metab 71:1239-1242, 1990)
differ from those in patients with comparable illness but without detectable serum TNF. To test this hypothesis, we compared the thyroid function tests in elderly nursing home patients with and without elevated serum TNF or IL-1 levels. The results suggest that elevated serum TNF levels are associated with a significant reduction in total serum T 3 levels. Materials and Methods Human serum samples Sera were obtained from the residents of the Los Angeles Jewish Home for the Aging. This site was chosen because a previous study had indicated that these patients have a relatively high prevalence of increased serum cytokine levels independent of their clinical or nutritional status (7). On admission, all of these residents had been screened for thyroid disease with appropriate thyroid function tests. A total of 29 patients with elevated serum TNF levels were compared to 36 patients with undetectable serum TNF levels. The 2 groups were comparable in age and sex distribution, medical problems, nutritional status, and the number of medications used (Table 1). None of the subjects had clinically evident infections or active malignancy. The types of medications used in the 2 groups were not different. In a previous study of a larger number of patients from the same institution, we could not find any correlation between serum TNF levels or any particular medical diagnosis or medication use (7). A similar comparison was made between 1239
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 19 November 2015. at 16:59 For personal use only. No other uses without permission. . All rights reserved.
MOORADIAN ET AL
1240
TABLE 1. The clinical characteristics of patients with detectable serum TNF levels (positive TNF) and those with undetectable serum TNF levels (negative TNF)
Age (yr) Male/female ratio Wt (kg) Ht(m) No. of diagnoses No. of medications
Positive TNF (n = 29)
Negative TNF (n = 36)
p
88.83 ± 3.56 5/25 52.8 ± 10.4 1.514 ±0.117 4.26 ± 0.63 10.5 ± 4.63
89.26 ± 5.09 6/36 54.1 ± 6.9 1.511 ± 0.124 4.05 ± 1.11 9.54 ± 4.06
NS NS NS NS
NS NS
TABLE 2. The clinical characteristics of patients with detectable serum IL-la levels (positive IL-1) and those with undetectable levels (negative IL-1)
Age (yr) Male/female ratio Wt (kg) Ht(m) No. of diagnoses No. of medications
Positive IL-1 (n = 25) 89.03 ± 4.05 3/23 52.3 ± 8.7 1.473 ± 0.103 4.20 ± 0.40 9.92 ± 4.57
Negative IL-1 (n = 38) 89.10 ± 4.77 8/26 53.9 ± 8.1 1.506 ± 0.114 4.11 ± 1.13 9.95 ± 4.21
p
JCE & M • 1990 Vol 71 • No 5
made on an ELISA reader set at 405 nm within 10 min of substrate addition. The concentration of TNF in serum was determined by comparison with a set of standards made with recombinant human TNF. (Genentech). The lower limit of detection with this assay system is 2 pM. The interassay coefficient of variation was 4.8%. The ELISA for human IL-la has been described previously (11) and was carried out using essentially the same protocol as that used for the TNF assay. In this assay system, 10 ng ascitic protein containing the IL-la-specific mAb C12 (Olympus, Inc., Lake Success, NY) were used to coat each well of the assay plates. A rabbit anti-IL-la-specific antiserum was used as the second reagent. The serum concentrations of IL-la was determined by comparison with a set of standards made by adding known amounts of recombinant human IL-la to serum. The lower limit of detection with this assay system is 2.8 pM. The interassay coefficient of variation was 5.1%. All samples were measured on the same day and in duplicate.
NS
Thyroid function tests
NS NS NS NS
Total T 3 (3,5,3'-L-triiodothyronine) and free T4 were measured by commercial competitive binding assay kits (Ciba-Corning Diagnostics, Medfield, MA) (12). The precision (i.e. interassay coefficient of variation) of the T3 method is 16.0% at 1.229 nmol/L and 9.3% at 3.840 nmol/L. The sensitivity is 0.307 nmol/L. The normal range is 1.536-3.072 nmol/L. The precision of the free T4 method is 12.0% at 7.7 pmol/L and 7.7% at 57.9 pmol/L. The sensitivity is 2.6 pmol/L. The normal range is 10-36 pmol/L. Total T4 was measured by the TDX therapeutic drug monitoring fluorescence polarization immunoassay method (Abbott Diagnostics, Irving, TX) (13). The precision of the T4 method is 7.6% at 57 nmol/L and 4.1% at 212.3 nmol/L. The normal range is 57-154 nmol/L. TSH was measured by a microparticle enzyme IMx immunoassay (Abbott Diagnostics) (14). The precision of the method is 11.3% at 0.4 mU/L and 3.4% at 29.9 mU/L. The sensitivity is 0.05 mU/L, and the normal range is 0.48-4.0 mU/L. rT 3 was measured by a competitive RIA kit (Techland, Belgium) employing [125I]rT3 tracer, rabbit anti-rT3, and second antibody precipitation separation. The precision of the method is 8.4% at 1.029 nmol/L and 12.5% at 0.661 nmol/L. The normal range is 0.307-0.768 nmol/L. All results are expressed as the mean ± SD. Statistical analysis was performed using one-way analysis of variance, followed by Student's t test for unpaired variables with Bonferoni correction. P < 0.05 was considered the limit for the statistical significance.
patients who had elevated serum IL-la levels (n = 25) and those with undetectable levels of serum IL-la (n = 38). Since there is a correlation between serum TNF and IL-la levels (7), some of these subjects (15 of 25) had also been included in the group with elevated serum TNF levels. In this comparison, the control group did not have detectable IL-la levels, but 12 of 38 controls had detectable TNF levels. The clinical characteristics of these patients are shown in Table 2. The 2 groups were also comparable in age and sex distribution, medical problems, medications, and nutritional status (Table 2). Patients with known thyroid disease were excluded from the study. Since glucocorticoids suppress TNF production (8) and have independent effects on thyroid function tests (9), patients receiving systemic glucocorticoid therapy were also excluded. TNF and IL-la enyzme-linked immunosorbant assay (ELISA) systems An ELISA system specific for human TNF was used to detect serum levels. The details of this assay system have been previously described (10). In brief, 96-well Immulon II (Dynatech Inc., McLean, VA) plastic plates were coated with mAb 6E, a murine monoclonal antibody specific for human TNF. This antibody was a gift from Genentech, Inc. (South San Francisco, CA). Sera were added to the wells and incubated for 1 h at room temperature. The plates were washed three times with phosphate-buffered saline containing 0.05% Tween-20, and a rabbit anti-TNF reagent was subsequently added. After a similar incubation and wash procedure, a peroxidase-conjugated goat antirabbit immunoglobulin G was added to the wells. The peroxidase substrate used was 2,2'-azino-bis-3-ethylbenzthiazoline-6-sulfonic acid (Sigma, St. Louis, MO). Optical density readings were
Results The results of thyroid function tests in the two groups of patients studied are summarized in Table 3. The only significant difference was found in the level of T3. The serum T 3 concentration in those subjects with elevated TNF was significantly reduced (1.072 ± 0.588 us. 1.621 ± 0.594 nmol/L; P < 0.01). The serum TSH concentration in patients with elevated TNF levels was not signif-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 19 November 2015. at 16:59 For personal use only. No other uses without permission. . All rights reserved.
TNF AND THYROID FUNCTION TABLE 3. The thyroid function test results (mean ± SD) in patients with detectable serum TNF levels (positive TNF) and those with undetectable serum TNF levels (negative TNF) Test T4 (nmol/L) Free T4 (pmol/L) T3 (nmol/L) rT3 (nmol/L) TSH (mU/L)
Positive TNF (n = 29)
Negative TNF (n = 36)
93.43 ± 17.89 ± 1.072 ± 0.387 ± 2.42 ±
91.51 ± 16.99 ± 1.621 ± 0.403 ± 1.61 ±
20.20 3.22 0.588 0.128 1.38
TABLE 4. The thyroid function test results (mean ± SD) in patients with detectable serum IL-la levels (positive IL-1) and those with undetectable serum levels of IL-la (negative IL-1) regardless of serum TNF status Test
NS NS
Vol. 71, No. 5 Printed in U.S.A.
Decreased Serum Triiodothyronine Is Associated with Increased Concentrations of Tumor Necrosis Factor ARSHAG D. MOORADIAN, RICHARD L. REED, DAN OSTERWEIL, RONALD SCHIFFMAN, AND PHILIP SCUDERI Division of Restorative Medicine (A.D.M., R.L.R.) and Veterans Administration Medical Center (A.D.M., R.S.), the Department of Microbiology and Immunology, University of Arizona College of Medicine (P.S.), Tucson, Arizonia 85719; and the Department of Medicine, University of California School of Medicine (D.O.), Los Angeles, California 90024
ABSTRACT. Previous studies in laboratory animals have shown that tumor necrosis factor-a (TNF) may alter thyroid function tests. To determine whether elevated serum TNF levels are associated with altered serum concentrations of T4, T3, free T4, rT3, and TSH, we measured these parameters in 29 nursing home residents with detectable serum TNF levels and compared the levels to those found in 36 patients with undetectable serum TNF levels. The 2 groups were matched for age, sex, clinical problems, use of medications, and nutritional status. Patients with detectable serum TNF levels had significantly lower serum T3 concentrations compared to those with undetectable levels
P
ROFOUND changes in thyroid function tests occur in patients with nonthyroidal illness (1, 2). The mechanism underlying these changes is probably multifactorial. Recent studies in laboratory animals have indicated that tumor necrosis factor-a (TNF) may have a pathogenetic role in altered thyroid hormone economy associated with illness (3, 4). Administration of TNF to rats and mice results in reduced thyroid radioiodine uptake, reduced response to TSH, and low serum concentrations of T4, T3, and rT 3 (3, 4). In addition, it has been shown that there are specific TNF receptors in human and rat thyroid cells that are regulated by TSH (5). It has been suggested that TNF may be a mediator of the changes in thyroid function tests seen in nonthyroidal illness (3-5). It has also been shown that administration of interleukin-1 (IL-1), -a or -/3, to mice will alter thyroid hormone metabolism in a manner similar to the effect of TNF (6). The clinical relevance of these observations, however, has not been addressed. If these findings in laboratory animals can be extrapolated to human subjects, one would expect that serum thyroid hormone levels in patients with elevated serum TNF will Received March 1, 1990. Address all correspondence and requests for reprints to: Arshag D. Mooradian, University of Arizona, 1821 East Elm Street, Tucson, Arizonia 85719.
[1.072 ± 0.588 vs. 1.621 ± 0.594 nmol/L (mean ± SD); P < 0.01]. Differences in other tests did not achieve statistical significance. Thyroid function tests were not significantly different when patients with detectable interleukin-la levels, another cytokine secreted during endotoxemia, were compared to those with undetectable levels. These observations taken together with the previous findings in laboratory animals suggest that some of the alterations in thyroid hormone levels seen in nonthyroidal illness are associated with elevated serum concentrations of TNF. (J Clin Endocrinol Metab 71:1239-1242, 1990)
differ from those in patients with comparable illness but without detectable serum TNF. To test this hypothesis, we compared the thyroid function tests in elderly nursing home patients with and without elevated serum TNF or IL-1 levels. The results suggest that elevated serum TNF levels are associated with a significant reduction in total serum T 3 levels. Materials and Methods Human serum samples Sera were obtained from the residents of the Los Angeles Jewish Home for the Aging. This site was chosen because a previous study had indicated that these patients have a relatively high prevalence of increased serum cytokine levels independent of their clinical or nutritional status (7). On admission, all of these residents had been screened for thyroid disease with appropriate thyroid function tests. A total of 29 patients with elevated serum TNF levels were compared to 36 patients with undetectable serum TNF levels. The 2 groups were comparable in age and sex distribution, medical problems, nutritional status, and the number of medications used (Table 1). None of the subjects had clinically evident infections or active malignancy. The types of medications used in the 2 groups were not different. In a previous study of a larger number of patients from the same institution, we could not find any correlation between serum TNF levels or any particular medical diagnosis or medication use (7). A similar comparison was made between 1239
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 19 November 2015. at 16:59 For personal use only. No other uses without permission. . All rights reserved.
MOORADIAN ET AL
1240
TABLE 1. The clinical characteristics of patients with detectable serum TNF levels (positive TNF) and those with undetectable serum TNF levels (negative TNF)
Age (yr) Male/female ratio Wt (kg) Ht(m) No. of diagnoses No. of medications
Positive TNF (n = 29)
Negative TNF (n = 36)
p
88.83 ± 3.56 5/25 52.8 ± 10.4 1.514 ±0.117 4.26 ± 0.63 10.5 ± 4.63
89.26 ± 5.09 6/36 54.1 ± 6.9 1.511 ± 0.124 4.05 ± 1.11 9.54 ± 4.06
NS NS NS NS
NS NS
TABLE 2. The clinical characteristics of patients with detectable serum IL-la levels (positive IL-1) and those with undetectable levels (negative IL-1)
Age (yr) Male/female ratio Wt (kg) Ht(m) No. of diagnoses No. of medications
Positive IL-1 (n = 25) 89.03 ± 4.05 3/23 52.3 ± 8.7 1.473 ± 0.103 4.20 ± 0.40 9.92 ± 4.57
Negative IL-1 (n = 38) 89.10 ± 4.77 8/26 53.9 ± 8.1 1.506 ± 0.114 4.11 ± 1.13 9.95 ± 4.21
p
JCE & M • 1990 Vol 71 • No 5
made on an ELISA reader set at 405 nm within 10 min of substrate addition. The concentration of TNF in serum was determined by comparison with a set of standards made with recombinant human TNF. (Genentech). The lower limit of detection with this assay system is 2 pM. The interassay coefficient of variation was 4.8%. The ELISA for human IL-la has been described previously (11) and was carried out using essentially the same protocol as that used for the TNF assay. In this assay system, 10 ng ascitic protein containing the IL-la-specific mAb C12 (Olympus, Inc., Lake Success, NY) were used to coat each well of the assay plates. A rabbit anti-IL-la-specific antiserum was used as the second reagent. The serum concentrations of IL-la was determined by comparison with a set of standards made by adding known amounts of recombinant human IL-la to serum. The lower limit of detection with this assay system is 2.8 pM. The interassay coefficient of variation was 5.1%. All samples were measured on the same day and in duplicate.
NS
Thyroid function tests
NS NS NS NS
Total T 3 (3,5,3'-L-triiodothyronine) and free T4 were measured by commercial competitive binding assay kits (Ciba-Corning Diagnostics, Medfield, MA) (12). The precision (i.e. interassay coefficient of variation) of the T3 method is 16.0% at 1.229 nmol/L and 9.3% at 3.840 nmol/L. The sensitivity is 0.307 nmol/L. The normal range is 1.536-3.072 nmol/L. The precision of the free T4 method is 12.0% at 7.7 pmol/L and 7.7% at 57.9 pmol/L. The sensitivity is 2.6 pmol/L. The normal range is 10-36 pmol/L. Total T4 was measured by the TDX therapeutic drug monitoring fluorescence polarization immunoassay method (Abbott Diagnostics, Irving, TX) (13). The precision of the T4 method is 7.6% at 57 nmol/L and 4.1% at 212.3 nmol/L. The normal range is 57-154 nmol/L. TSH was measured by a microparticle enzyme IMx immunoassay (Abbott Diagnostics) (14). The precision of the method is 11.3% at 0.4 mU/L and 3.4% at 29.9 mU/L. The sensitivity is 0.05 mU/L, and the normal range is 0.48-4.0 mU/L. rT 3 was measured by a competitive RIA kit (Techland, Belgium) employing [125I]rT3 tracer, rabbit anti-rT3, and second antibody precipitation separation. The precision of the method is 8.4% at 1.029 nmol/L and 12.5% at 0.661 nmol/L. The normal range is 0.307-0.768 nmol/L. All results are expressed as the mean ± SD. Statistical analysis was performed using one-way analysis of variance, followed by Student's t test for unpaired variables with Bonferoni correction. P < 0.05 was considered the limit for the statistical significance.
patients who had elevated serum IL-la levels (n = 25) and those with undetectable levels of serum IL-la (n = 38). Since there is a correlation between serum TNF and IL-la levels (7), some of these subjects (15 of 25) had also been included in the group with elevated serum TNF levels. In this comparison, the control group did not have detectable IL-la levels, but 12 of 38 controls had detectable TNF levels. The clinical characteristics of these patients are shown in Table 2. The 2 groups were also comparable in age and sex distribution, medical problems, medications, and nutritional status (Table 2). Patients with known thyroid disease were excluded from the study. Since glucocorticoids suppress TNF production (8) and have independent effects on thyroid function tests (9), patients receiving systemic glucocorticoid therapy were also excluded. TNF and IL-la enyzme-linked immunosorbant assay (ELISA) systems An ELISA system specific for human TNF was used to detect serum levels. The details of this assay system have been previously described (10). In brief, 96-well Immulon II (Dynatech Inc., McLean, VA) plastic plates were coated with mAb 6E, a murine monoclonal antibody specific for human TNF. This antibody was a gift from Genentech, Inc. (South San Francisco, CA). Sera were added to the wells and incubated for 1 h at room temperature. The plates were washed three times with phosphate-buffered saline containing 0.05% Tween-20, and a rabbit anti-TNF reagent was subsequently added. After a similar incubation and wash procedure, a peroxidase-conjugated goat antirabbit immunoglobulin G was added to the wells. The peroxidase substrate used was 2,2'-azino-bis-3-ethylbenzthiazoline-6-sulfonic acid (Sigma, St. Louis, MO). Optical density readings were
Results The results of thyroid function tests in the two groups of patients studied are summarized in Table 3. The only significant difference was found in the level of T3. The serum T 3 concentration in those subjects with elevated TNF was significantly reduced (1.072 ± 0.588 us. 1.621 ± 0.594 nmol/L; P < 0.01). The serum TSH concentration in patients with elevated TNF levels was not signif-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 19 November 2015. at 16:59 For personal use only. No other uses without permission. . All rights reserved.
TNF AND THYROID FUNCTION TABLE 3. The thyroid function test results (mean ± SD) in patients with detectable serum TNF levels (positive TNF) and those with undetectable serum TNF levels (negative TNF) Test T4 (nmol/L) Free T4 (pmol/L) T3 (nmol/L) rT3 (nmol/L) TSH (mU/L)
Positive TNF (n = 29)
Negative TNF (n = 36)
93.43 ± 17.89 ± 1.072 ± 0.387 ± 2.42 ±
91.51 ± 16.99 ± 1.621 ± 0.403 ± 1.61 ±
20.20 3.22 0.588 0.128 1.38
TABLE 4. The thyroid function test results (mean ± SD) in patients with detectable serum IL-la levels (positive IL-1) and those with undetectable serum levels of IL-la (negative IL-1) regardless of serum TNF status Test
NS NS
