The Mediation Role of Thyrotropin Receptor Antibody in the Relationship Between Age and Severity of Hyperthyroidism in Graves' Disease

Abstract

Background:

The severity of hyperthyroidism in Graves' disease (GD) has been reported to be worse in younger patients and to gradually improve with advancing age, accompanied by declining thyrotropin (TSH) receptor antibody (TRAb) values. This study was conducted to explore the extent to which the declining TRAb production may contribute to a decrease in severe hyperthyroidism with advancing age in patients with GD.

Methods:

This study was a cross-sectional analysis of retrospectively reviewed data. The medical records of patients newly diagnosed with GD at Ito Hospital, between January 2005 and June 2019, were examined. Patients were divided into age-stratified groups for evaluation. Multivariable logistic regression was performed to estimate the odds ratio (OR) of severe hyperthyroidism by increasing age. Mediation analyses were also conducted to quantify the association between age and declining severity of hyperthyroidism mediated through decreased TRAb productivity.

Results:

A total of 21,018 patients with newly diagnosed GD (3848 male and 17,170 female) were included. A correlation was observed between TRAb value and thyroid hormone values in each age-stratified group, which became weaker with an increase in age. Patients aged <40 years had a higher risk of severe hyperthyroidism (free thyroxine [fT4] level >7.0 ng/dL [n = 5616], OR [confidence interval, CI] = 1.80 [1.68–1.92]; free triiodothyronine [fT3] level >25 pg/mL [n = 4501], OR [CI] = 2.06 [1.92–2.23]) than those aged ≧40 years. In examining the relationship between age and severe hyperthyroidism, the proportion mediated through TRAb productivity was 8.5% and 8.4% using fT4 and fT3 as an outcome index, respectively.

Conclusions:

Declining TRAb value mediated only 8.5% of the negative association between age and severity of hyperthyroidism. The presence of other underlying mechanisms, such as the decline in the reactivity of thyrocytes to TSH stimulation, requires further investigation.

Introduction

Graves' disease (GD) is a common cause of hyperthyroidism and develops frequently in people of younger age with female predominance. An earlier report showed that the severity of hyperthyroidism in GD was higher in male individuals in their younger ages and declined with advancing age in both sexes.¹ Similar results were reported from several countries,^2

–6 which demonstrated that people of younger ages and male sex had more severe hyperthyroidism, or that elderly patients had a milder disease. However, only a few reports have discussed the underlying mechanisms of these age-related changes. According to Aizawa et al and Aizawa and Yamada,^3,7 these changes could be caused by reduced responsiveness of thyrocytes to the thyrotropin (TSH) receptor antibody (TRAb), or less activity of TRAb with advancing age, but not by a reduced amount or incidence of TRAb titers. However, they did not demonstrate the correlation between TRAb value or the incidence and severity of GD in their studies.

A more recent study¹ reported that TRAb value also declined with advancing age, suggesting that the declining of TRAb value could be a cause of the age-related declining of hyperthyroidism in GD. Therefore, in this study, we aimed to quantify the extent to which the declining TRAb productivity contributes to declining hyperthyroidism due to age among patients with GD.

Materials and Methods

Participants

This study is a cross-sectional analysis study of retrospectively reviewed data. Between January 2005 and July 2019, 21,595 patients aged 10–92 years (3947 male and 17,648 female) were newly diagnosed with GD at Ito Hospital (as identified by the Information Technology Department, using the search term “newly diagnosed untreated GD” in the electronic health records). As patients with GD were diagnosed and initially treated with medication at the outpatient department of our hospital, all patients included in this study were outpatients. Figure 1 shows the patient flowchart. We excluded 577 patients aged >70 years due to the relatively smaller sample size of this age group, resulting in a final analytical sample of 21,018 patients (3848 male and 17,170 female).

FIG. 1.

Participant flowchart. A total of 21,595 subjects were diagnosed with Graves' disease during the observation period, and 577 patients who were >70 years old were excluded from the analysis due to the small number, thereby 21,018 patients as eligible participants for the primary analysis. We included 3443 participants with information on Tg and negative for TgAb in the sensitivity analysis. Tg, thyroglobulin; TgAb, antithyroglobulin antibody.

Subjects were first divided into six groups by age as follows: 10–19 years (10s) (n = 1194, 167 male and 1027 female), 20–29 years (20s) (n = 4116, 672 male and 3444 female), 30–39 years (30s) (n = 6206, 1179 male and 5027 female), 40–49 years (40s) (n = 4820, 955 male and 3865 female), 50–59 years (50s) (n = 3109, 559 male and 2550 female), and 60–69 years (60s) (n = 1573, 316 male and 1257 female).

GD was diagnosed based on the status of both hyperthyroidism and positivity for TRAb or thyroid-stimulating antibody (TSAb) or diffuse high uptake of radioactive iodine (¹²³I) by the thyroid gland. Among the 21,018 patients, 20,218 (96.2%) were diagnosed with GD based on positive TRAb results, whereas 44 (0.2%) were diagnosed on the bases of positive TSAb results. The remaining 756 patients (3.65%) were diagnosed with GD based on the radioactive iodine test. Because the number of subjects who were diagnosed with GD on the basis of positive TSAb results was smaller than those diagnosed based on positive TRAb results, TRAb was used as a parameter in the analyses.

Thyroglobulin (Tg) was used as an index for measuring the reactivity of thyrocytes to stimulation through the TSH receptor. Because the currently used measurement kit of Tg was affected by the presence of antithyroglobulin antibody (TgAb), we analyzed only the Tg values of the patients who were negative for TgAb (n = 3443, 689 male and 2754 female).

This study was approved by the Ethics Committee of Ito Hospital (approval no.: 307), and written consent was not required.

Baseline characteristics and laboratory measurements

Age and sex were self-reported by the patients at the time of blood examination on the initial visits. The methods used to assess serum TSH, free triiodothyronine (fT3), free thyroxine (fT4), TRAb, and TSAb levels were the same as those previously reported.¹ Two different TRAb measurement kits were used: the TRAb CT kit and the ECLusys TRAb kit. The ECLusys TRAb kit has an upper detectable limit of 40 IU/L, whereas the TRAb CT kit does not have any upper detectable limit. Hence, the TRAb values converted from the TRAb CT kit that showed ≧40 IU/L were unified as 40 IU/L. We compared the two values using a formula derived from the regression curve representing the relationship between the TRAb CT and ECLusys TRAb values.¹ The Tg level was measured using the ECLusys Tg electrochemiluminescence immunoassay kit (normal range <35 ng/dL; Roche Diagnostics), and the TgAb level was evaluated using the TgAb radioimmunoassay kit (normal range ≦40 IU/mL; Roche Diagnostics). Thyroid volume was calculated using the methods reported in the previous studies.^1,8 Measurement details are described in the Supplementary Data.

Statistical analyses

Data are expressed as medians and interquartile ranges for continuous variables. Spearman's rank-sum test was used to evaluate the correlation between TRAb values and thyroid hormone level. The multivariable logistic regression analysis was conducted to evaluate the association between age (10–39 years [younger age group] and 40–69 years [older age group; reference]) and severity of hyperthyroidism: fT4 > 7.0 ng/dL (n = 5616) and fT3 > 25.0 pg/mL (n = 4501). These outcome cutoff values of fT4 (>7.0 ng/dL) and fT3 (>25.0 pg/mL) were determined on the basis of the upper 25% of fT4 and fT3 values. The Steel–Dwass test was conducted to compare Tg levels between all the age groups.

Causal mediation analyses were conducted to quantify the degree to which decreased TRAb productivity contributed to the association between age and declining severity of hyperthyroidism among patients with GD (Supplementary Fig. S1). A marginal structural approach within the counterfactual framework was used to decompose the total odds ratio (OR) into the direct OR and indirect OR.⁹ In this mediation analysis, we defined the exposure as age <40 years versus ≧40 years, the mediator as TRAb ≧40 IU/L versus TRAb <40 IU/L, the outcome as the severity of hyperthyroidism (fT4 > 7.0 ng/dL or fT3 > 25.0 pg/mL), and the confounders as sex. The direct OR represents the association between age and declining severity of hyperthyroidism through pathways that do not involve decreased TRAb production. The indirect OR represents the association between age and declining severity of hyperthyroidism through decreased TRAb production.

The mediated proportion was computed as the log of the indirect OR divided by the log of the total OR. The interaction term of the exposure (age) and mediator (TRAb) was included in the model, but there was no statistical indication of the interaction. In mediation analyses, robust confidence intervals [CIs] were estimated by repeating the analysis on 10,000 bootstrapped samples.

Three analyses were performed. First, the subgroup analysis by sex was performed based on a previous report showing sex heterogeneity in the severity and remission rate of hyperthyroidism in GD.¹ Second, as a sensitivity analysis, we additionally adjusted for thyroid volume and smoking status using the subset of patients (n = 6921) with ultrasound examination and smoking status information.

Finally, we assessed the difference in Tg (as a marker of reactivity of thyrocytes to stimulation through the TSH receptor) and Tg divided by thyroid volume according to the age-stratified groups. Considering that the Tg value is generally influenced by TgAb, we analyzed the data of TgAb-negative patients (n = 3390, 688 male and 2702 female) in this additional analysis. Among the 3390 TgAb-negative patients, thyroid volume data were available in 2068 patients (439 male and 1269 female).

Among the 21,018 participants, fT3, fT4, and TRAb data were missing among 3 patients (0.01%), 2 patients (0.01%), and 33 patients (0.02%), respectively. We applied complete-case analyses for each outcome.

Statistical analyses were conducted using the JMP version 14.0 software (SAS Institute, Cary, NC) and R 4.0.1 (R Foundation for Statistical Computing, Vienna, Austria). A p-value of <0.05 was considered to be statistically significant.

Results

Characteristics of subjects and correlation between aging and severity of hyperthyroidism in GD

Among the 21,018 patients included in this study, 82% (n = 17,170) were female (Table 1). The number of patients with GD was largest in the 30s in both sexes. The median values of fT3, fT4, and TRAb showed a gradual decline with advancing age. Severe hyperthyroidism that was defined as fT4 > 7.0 ng/dL and fT3 > 25 pg/mL was observed among 5616 and 4501 patients, respectively.

Table 1.

Baseline Characteristics of the Participants and Median Values of Free-Triiodothyronine, Free-Thyroxine, and Thyrotropin Receptor Antibody in Each Age Group

		Sex
	Total No. (No. of smokers)	Male	Female	Median fT3 (pg/mL) (n = 21,011)	Median fT4 (ng/dL) (n = 21,016)	Median TRAb (IU/L) (n = 20,986)	Median Tg (ng/dL) (n = 3443)	Median volume (mL) (n = 6921)
All ages	21,018 (5386)	3848	17,170	15.7 (9.3–23.5)	4.6 (3.0–7.2)	11.6 (5.5–25.9)	113.3 (51.1–229.3)	29.0 (121.1–42.0)
Age 10–19 years	1194 (25)	167	1027	20.5 (12.0–29.1)	5.9 (3.58–7.8)	17.0 (7.5–35.2)	127.5 (50.0–264.3)	32.7 (23.6–48.0)
20–29 years	4116 (1020)	672	3444	18.2 (10.8–26.5)	5.3 (3.39–7.8)	13.3 (6.3–29.8)	118.3 (57.1–272.4)	31.5 (22.7–45.6)
30–39 years	6206 (1796)	1179	5027	16.6 (9.5–24.7)	4.8 (3.07–7.7)	11.9 (5.8–26.7)	113.5 (50.9–227.0)	30.4 (22.0–43.3)
Younger age group	11,516 (2841)	2018	9498	17.6 (10.1–25.8)	5.1 (3.2–7.8)	12.8 (6.1–28.7)	115.8 (52.4–242.9)	31.1 (22.4–44.7)
Age 40–49 years	4820 (1474)	955	3865	15.5 (9.4–22.6)	4.5 (3.02–6.9)	11.7 (5.5–24.8)	113.8 (54.0–227.9)	29.0 (21.4–42.0)
50–59 years	3109 (800)	559	2550	13.8 (8.7–20.4)	4.0 (2.83–6.0)	9.7 (4.7–22.0)	114.0 (46.2–224.3)	25.6 (19.3–36.2)
60–69 years	1573 (271)	316	1257	11.9 (7.7–17.0)	3.6 (2.60–5.0)	8.5 (4.1–19.2)	94.1 (44.7–169.5)	23.0 (16.6–31.5)
Older age group	9502 (2545)	1830	7672	14.1 (8.8–20.9)	4.2 (2.9–6.3)	10.4 (5.0–22.7)	109.6 (50.3–220.5)	27.0 (19.9–38.3)

Data are expressed as median and interquartile range.

fT3, free triiodothyronine; fT4, free thyroxine; TRAb, thyrotropin receptor antibody.

The relationships between TRAb and fT4 values are shown in Figure 2. We found a correlation between TRAb value and both fT3 and fT4 values in each age group. The correlation coefficient gradually decreased by age in both fT3 and fT4, suggesting that the correlation became weaker among older patients than younger patients.

FIG. 2.

Correlation between TRAb and fT4 according to age categories. Correlations are present between TRAb and fT4 values in each age group. As shown in each graph, the correlation coefficient (R) gradually decreases with advancing age, suggesting a stronger correlation in patients of younger age. fT4, free thyroxine; TRAb, thyrotropin receptor antibody.

The sex-adjusted multivariable logistic regression analysis showed that younger patients aged 10–39 years showed 2.18-fold higher odds of fT3 > 25 pg/mL (OR = 2.18, CI 2.04–2.34) and 1.92-fold higher odds of fT4 > 7.0 ng/dL (OR = 1.92, CI 1.80–2.04) than older patients aged 40–69 years (Table 2). When we additionally adjusted for TRAb value (i.e., the potential mediator), we observed that younger patients aged 10–39 years showed 2.06-fold higher odds of exceeding fT3 > 25 pg/mL (OR = 2.06, CI 1.92–2.23) and 1.80-fold higher odds of exceeding fT4 > 7.0 ng/dL (OR = 1.80, CI 1.68–1.92) than older patients aged 40–69 years. The ORs in this model were smaller than those in the model without adjustment for TRAb value (Table 2).

Table 2.

Association Between Age and Severity of Hyperthyroidism Among Patients with Graves' Disease

		Outcome
		fT4 > 7 ng/dL		fT3 > 25 pg/mL
		Model 1	Model 2	Model 1	Model 2
Age <40 vs. age ≥40	OR	1.92	1.80	2.18	2.06
	[CI]	[1.80–2.04]	[1.68–1.92]	[2.04–2.34]	[1.92–2.23]

Multivariable logistic regression was used to estimate the adjusted OR of each outcome (fT4 > 7 ng/dL or fT3 > 25 pg/mL) by age category (age <40 vs. age ≥40 [Ref.]). Model 1 included sex. Model 2 included TRAb in addition to sex.

CI, confidence interval; OR, odds ratio.

Association between declining TRAb value and declining severity in hyperthyroidism in patients with GD

In examining the relationship between age and severe hyperthyroidism, the proportion mediated through TRAb productivity was 8.5% using fT4 as an index (total OR [CI] = 1.92 [1.80–2.04], direct OR [CI] = 1.82 [1.71–1.94], and indirect OR [CI] = 1.06 [1.04–1.07]) and 8.4% using fT3 as an index (total OR [CI] = 2.18 [2.04–2.34], direct OR [CI] = 2.06 [1.93–2.21], and indirect OR [CI] = 1.07 [1.05–1.09]) (Fig. 3). We found a similar proportion mediated through TRAb productivity (8–9%) for both males and females (Supplementary Table S1), whereas the OR was greater in males than in females. When we additionally adjusted for thyroid volume and smoking status using the subset of 6921 patients, we observed a smaller magnitude of the proportion mediated compared with our main analysis (Supplementary Fig. S2).

FIG. 3.

Mediation of TRAb on the association between age and fT4. In the relationship between age and severe hyperthyroidism, the proportion mediated through TRAb productivity was 8.5% using fT4 as an index and 8.4% using fT3 as an index. fT3, free triiodothyronine; CI, confidence interval; OR, odds ratio.

Relationship between Tg values and aging

The median values of Tg gradually decreased with advancing age (Fig. 4). We found statistically significant differences between the age groups of 10s and 60s (p = 0.024), 20s and 60s (p = 0.006), and 40s and 60s (p = 0.020). However, when the Tg level was corrected by thyroid volume, this pattern was not recognized (Supplementary Fig. S3).

FIG. 4.

Comparison of the median values of Tg in age-stratified groups. Statistical differences were observed between the 10s and 60s (p = 0.001), 20s and 60s (p = 0.032), 30s and 60s (p = 0.001), and 40s and 60s (p = 0.009).

Discussion

Studies have reported that the severity of hyperthyroidism in GD in terms of clinical symptoms, such as palpitations,⁴ the exact values of thyroid hormones,^2,3,6 and therapeutic response to antithyroid drugs,^5,10 decreases with advancing age. It might be because the younger generation was diagnosed with a more aggressive presentation and milder cases of hyperthyroidism were diagnosed later, although our results did not support the theory. Nevertheless, the causes of this alteration have not yet been completely established. In this study, we found that TRAb might be a key factor contributing to the severity of hyperthyroidism in GD among both males and females. Our findings of the mediation analysis also indicated the presence of other pathways, not through TRAb, responsible for the declining severity of hyperthyroidism by age in patients with GD. Although TRAb is considered a primary factor inducing GD development, the proportion of its mediating role in the relationship between age and the severity of hyperthyroidism was ∼8% to 9%.

Several pathways responsible for hyperthyroidism severity in GD, not through TRAb, have been discussed in previous studies, such as declining reactivity of thyrocytes simulated from TSH or TRAb,³ decreased TRAb activity,⁷ inactivation of TSH receptor,¹¹ and structural degeneration of TSH receptor.^12,13 For instance, Yamada et al¹⁴ reported that after administrating thyrotropin-releasing hormone, the rate of the variability of TSH increase was significantly larger in older subjects with female diffuse goiter, although T3 levels decreased with advancing age. Their results suggested that the reactivity of thyrocytes in subjects with diffuse goiter declined with advancing age. Therefore, the decline in thyrocyte reactivity with age may also occur in the absence of autoimmune thyroiditis. Furthermore, as GD is an autoimmune thyroid disease, several changes in the immune system of patients with GD have previously been reported.^15,16

According to Al-Humaidi,¹⁶ both Th1 and Th2 cells are activated in patients with GD, causing long-term inflammation and thyrocyte destruction. These pathological changes may impact the reactivity of thyroid tissue. In the present study, declining Tg was associated with advancing age, which is consistent with a previous study.⁷ This supported our hypothesis that the reactivity of thyrocytes to stimulation through the TSH receptor decreases with age. However, this pattern was not evident after correction for thyroid volume. When we additionally adjusted for thyroid volume and smoking status using ∼30% of the study sample available for these data, we found smaller mediation of TRAb on the relationship between age and decreasing hyperthyroidism severity. These results indicate that thyroid volume slightly contributes to the reactivity of thyrocytes and severity in patients with GD; this requires further investigation with a larger sample size.

This study has several limitations. We lacked some data on demographic characteristics and other medical conditions. We cannot eliminate the possibility of violating one of the assumptions in the mediation analysis, that is, the presence of exposure-induced mediator-outcome confounder, because there would be some (measured or unmeasured) confounders between TRAb and thyroid hormone levels affected by age. Our data on TRAb and thyroid hormone levels were collected at the same visit; thus, the time ordering between these variables was unclear. Our data did not allow us to quantify the extent to which TRAb at the diagnosis contributes to GD progression after the treatment. In addition, it may be possible that people with a less severe hyperthyroid phenotype might be more likely to receive their diagnosis later than those with a more severe hyperthyroid phenotype. We did not include inpatients and exclude individuals >70 years of age, which might contribute to potential selection bias.

Finally, as this study was a single-center investigation performed at a hospital specializing in thyroid disease in an Asian country, our findings are not necessarily generalizable to other populations.

Conclusion

This study revealed the association between age and severity of hyperthyroidism, which was partially mediated through the TRAb titer. As the proportion mediated by TRAb was 8.5%, there may be several other pathways responsible for the decreased severity of hyperthyroidism among patients with GD by advancing age and these deserve further study.

Footnotes

Authors' Contributions

Conceptualization: R.K. and N.S. Data Curation: N.S., R.Y., A.K., A.S., M.F., and M.M. Formal analysis: N.S. and K.I. Project administration: J.Y.N., A.Y., and N.W. Supervision: K.S. and K.I. Writing—original draft: N.S. and K.I.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

No funding was received for this article.

Supplementary Material

Supplementary Data

Supplementary Table S1

Supplementary Figure S1

Supplementary Figure S2

Supplementary Figure S3

References

Suzuki

, Noh

, Yoshimura

, et al. Does age or sex relate to severity or treatment prognosis in Graves' disease?. Thyroid, 2021; 31(9):1409–1415; doi: 10.1089/thy.2020.0881

Magri

, Zerbini

, Gaiti

, et al. Gender influences the clinical presentation and long-term outcome of Graves disease. Endocr Pract, 2016; 22(11):1336–1342; doi: 10.4158/EP161350.OR

Aizawa

, Ishihara

, Hashizume

, et al. Age-related changes of thyroid function and immunologic abnormalities in patients with hyperthyroidism due to Graves' disease. J Am Geriatr Soc, 1989; 37(10):944–948; doi: 10.1111/j.1532-5415.1989.tb07279.x

Nordyke

, Gilbert

Jr ., Harada

. Graves' disease. Influence of age on clinical findings. Arch Intern Med, 1988; 148(3):626–631; doi: 10.1001/archinte.148.3.626

Yamada

, Aizawa

, Koizumi

, et al. Age-related therapeutic response to antithyroid drug in patients with hyperthyroid Graves' disease. J Am Geriatr Soc, 1994; 42(5):513–516; doi: 10.1111/j.1532-5415.1994.tb04973.x

Allahabadia

, Daykin

, Holder

, et al. Age and gender predict the outcome of treatment for Graves' hyperthyroidism. J Clin Endocrinol Metab, 2000; 85(3):1038–1042; doi: 10.1210/jcem.85.3.6430

Aizawa

, Yamada

. Hyperthyroidism due to Graves' disease in aged patients. Trends Endocrinol Metab, 1991; 2(1):4–8; doi: 10.1016/1043-2760(91)90053-p

Sato

, Noh

, Sato

, et al. Comparison of efficacy and adverse effects between methimazole 15mg+inorganic iodine 38mg/day and methimazole 30mg/day as initial therapy for Graves' disease patients with moderate to severe hyperthyroidism. Thyroid, 2015; 25(1):43–50; doi: 10.1089/thy.2014.0084

VanderWeele

TJ.

Mediation analysis: A practitioner's guide. Annu Rev Public Health, 2016; 37:17–32; doi: 10.1146/annurev-publhealth-032315-021402

10.

Cappelli

, Gandossi

, Castellano

, et al. Prognostic value of thyrotropin receptor antibodies (TRAb) in Graves' disease: A 120 months prospective study. Endocr J, 2007; 54(5):713–720; doi: 10.1507/endocrj.k06-069

11.

Ando

, Latif

, Davies

. Concentration-dependent regulation of thyrotropin receptor function by thyroid-stimulating antibody. J Clin invest, 2004; 113(11):1589–1595; doi: 10.1172/JCI21334

12.

Chen

, Pichurin

, Nagayama

, et al. The thyrotropin receptor autoantigen in Graves disease is the culprit as well as the victim. J Clin invest, 2003; 111(12):1897–1904; doi: 10.1172/JCI17069

13.

Davies

, Marians

, Latif

. The TSH receptor reveals itself. J Clin invest, 2002; 110(2):161–164; doi: 10.1172/JCI16234

14.

Yamada

, Naka

, Komiya

, et al. Age-related alterations of pituitary-thyroid function in normal female subjects and in female patients with simple goitre. Acta Endocrinol (Copenh), 1984; 107(3):346–351; doi: 10.1530/acta.0.1070346

15.

Ueda

, Ichiyama

, Sugawa

. Thyroid specific T helper cell analysis by ELISPOT assay with thyrotropin receptor (TSH-R) peptides. Peptides, 2002; 23(1):103–107; doi: 10.1016/s0196-9781(01)00585-x

16.

Al-Humaidi

MA.

Serum cytokines levels in Graves' disease. Saudi Med J, 2000; 21(7):639–644.