Current Thyrotropin Immunoassays Recognize Macro-Thyrotropin Leading to Hyperthyrotropinemia in Females of Reproductive Age

Abstract

Background:

Macro-thyrotropin (macro-TSH) is a high molecular weight form of TSH that leads to hyperthyrotropinemia. This study was undertaken to examine the prevalence and nature of macro-TSH in females of reproductive age.

Methods:

Blood samples were taken from 1794 female patients who visited the Hamada Obstetrics and Gynecology Hospital in Tokyo, Japan, complaining of infertility. The serum of 305 patients with TSH concentrations >2.5 mIU/L was screened for macro-TSH by the polyethylene glycol method. Samples with TSH precipitation ratios by polyethylene glycol >70% were further analyzed using gel filtration chromatography (GFC), protein G columns, and ¹²⁵I-TSH binding experiments.

Results:

Screening of the 305 patients revealed that 63 had serum TSH precipitation ratios >70%. GFC revealed that immunoreactive TSH, with a molecular weight of approximately 150 kDa, eluted at higher ratios (79.6 ± 24.4%) in 27 of the 63 patients compared to 0.4 ± 2.0% in the control group. Serum TSH concentrations in 24 of the 27 patients were spuriously elevated due to human anti-mouse antibodies. Macro-TSH was found in the other three patients, and one of them had detectable anti-TSH autoantibodies. Eight of the remaining 36 patients who did not have high-molecular-weight TSH assessed by GFC had immunoglobulin G–associated TSH. Three commercially available TSH immunoassays (Elecsys^®, Centaur^®, and Architect^®) all recognized macro-TSH leading to the elevated serum TSH concentrations.

Conclusions:

Macro-TSH was present in 0.17% of infertile women. Commercial TSH immunoassays recognized macro-TSH, resulting in the diagnosis of hyperthyrotropinemia.

Introduction

Subclinical hypothyroidism is a condition in which serum thyroid hormone concentrations are normal, despite elevated thyrotropin (TSH) concentrations. Thyroid hormone replacement therapy is usually considered when serum TSH concentrations are >10 mIU/L (1). However, this is not the case with pregnant women because serum TSH concentrations are suppressed due to elevated human chorionic gonadotropin during pregnancy (1,2). There are several complications such as miscarriage and placental abruption during pregnancy in patients with subclinical hypothyroidism (3). In determining the therapeutic strategy, the TSH concentrations in females of reproductive age should be carefully evaluated.

Macro-TSH is a high molecular weight form of TSH, with a molecular mass >150 kDa. It can be distinguished from TSH, which has a molecular weight of 28 kDa, by its behavior in gel filtration chromatography (GFC) (4 –11). Due to its large molecular size, the clearance of macro-TSH from the blood may be delayed relative to TSH, permitting it to accumulate in the circulation and lead to elevated serum TSH concentrations. The bioactivity of macro-TSH is low (4,10), so that thyroid hormone concentrations can be normal in the presence of macro-TSH, a situation that may mimic subclinical hypothyroidism. The pathogenesis of macro-TSH is not fully understood, but is due at least in part to a complex of TSH with anti-TSH autoantibodies (4 –11).

The prevalence of macro-TSH has been reported to be 0.61% in patients with serum TSH >10 mIU/L (9) and 0.79% in patients with subclinical hypothyroidism (TSH >4 mIU/L) (11). However, the prevalence of macro-TSH in females of reproductive age is not known. Careful evaluation of serum TSH concentrations is important in this population to determine if hormone replacement therapy is needed in order to avoid complications during pregnancy. The present study assessed the prevalence of macro-TSH in females of reproductive age and determined the detectability of macro-TSH by commercially available TSH immunoassays.

Materials and Methods

Patients

Of 1794 female patients between 16 and 39 years of age who visited Hamada Obstetrics and Gynecology Hospital in Tokyo, Japan, mostly seeking assistance with infertility, 305 patients were selected with serum TSH concentration >2.5 mIU/L and normal free thyroxine (fT4) concentrations. Blood sera from these patients were screened for macro-TSH with a polyethylene glycol (PEG) precipitation method. Samples with TSH precipitation ratios by PEG >70% were subjected to GFC to identify the presence of macro-TSH. Twenty patients with subclinical hypothyroidism whose serum TSH precipitation ratios by PEG were <70% comprised the control group. This study was approved by the Clinical Research Review Board of Hamada Obstetrics and Gynecology Hospital.

TSH and fT4 assays

The concentration of TSH in all samples was measured with an enzyme immunoassay (EIA) that was established in the authors' laboratory (10). Serum TSH concentrations were also measured with three commercial TSH assays to determine if they recognized macro-TSH. These commercial assays were: the Elecsys^® immunoassay, which employs the Cobas e411 analyzer^® (Roche Diagnostics GmbH, Mannheim, Germany); the ADVIA Centaur XP^® immunoassay system (Siemens Healthcare Diagnostics, Munich, Germany); and the Architect^® TSH immunoassay, which employs the Architect i2000 analyzer^® (Abbot Diagnostics, Abbott Park, IL). fT4 concentrations were measured with the ADVIA Centaur XP^® immunoassay system (Siemens Healthcare Diagnostics), with a reference range of 0.88–1.81 ng/dL. The conversion factor of fT4 concentration from ng/dL to pmol/L was 12.87.

PEG precipitation

PEG was used to precipitate γ-globulin that was associated with TSH, as previously reported (10). In brief, 50 μL serum samples were treated either with 50 μL of 25% PEG (final concentration of PEG was 12.5%) to precipitate γ-globulin fractions, leaving the free TSH in the supernatant, or with 50 μL of water as a control (serum TSH). The percentage of PEG-precipitable TSH was calculated as follows: (serum TSH − free TSH)/serum TSH × 100.

To examine the effects of γ-globulin on TSH precipitation by PEG, human TSH reference preparation (hTSH; hTSH-SIAFPI-8; AFP-3959A; National Institute of Diabetes and Digestive and Kidney Diseases [NIDDK], Bethesda, MD) was incubated at room temperature overnight in 50 μL of 0.01 mol/L phosphate buffer (pH 7.0) containing 0.1 mol/L NaCl (phosphate-buffered saline [PBS]), PBS with 40 mg/mL bovine serum albumin (BSA; Wako Pure Chemical Industries Ltd., Osaka, Japan), PBS with 10 mg/mL human γ-globulin (Wako), and PBS with 40 mg/mL BSA plus 10 mg/mL human γ-globulin. Next, these samples were treated with 50 μL of 25% PEG, and the hTSH concentrations in the supernatants were measured. The percentage of the precipitation ratios of hTSH by PEG treatment was calculated as follows: (hTSH in PBS − hTSH in the supernatant of PBS containing BSA, γ-globulin, and BSA plus γ-globulin)/hTSH in PBS × 100.

The effects of serum dilution on TSH precipitation by PEG were examined in serum that had previously been shown to have macro-TSH (10). The TSH concentrations of the serum, the ratios of immunoglobulin G (IgG)-associated TSH and the ratios of high molecular weight TSH on GFC in two patients having macro-TSH are shown in the footnote of Table 1. As shown in Table 1, serum dilution up to fourfold did not affect TSH precipitation ratios by PEG.

Table 1.

Effects of Serum Dilution on TSH Precipitation Ratios by PEG

	Added PEG				PEG precipitation ratios of serum TSH (%)
Serum (μL)	(%)	(μL)	PEG final (%)	Serum dilution x-fold	Patient 1	Patient 2
50	75.0	10	12.5	1.2	97.2	97.7
50	37.5	25	12.5	1.5	98.1	97.7
50	25.0	50	12.5	2.0	99.0	97.9
50	18.8	100	12.5	3.0	98.7	98.0
50	16.7	150	12.5	4.0	98.5	98.3

The sera were obtained from two patients previously shown to have macro-TSH (10). In patient 1, serum TSH concentration was 716 mIU/L, the ratio of IgG-associated TSH was 96.3%, the ratio of high-molecular-weight TSH was 100%, and there was no interference due to HAMA. In patient 2, serum TSH concentration was 26.9 mIU/L, the ratio of IgG-associated TSH was 60.5%, the ratio of high-molecular-weight TSH 100%, and there was no interference due to HAMA.

TSH, thyrotropin; PEG, polyethylene glycol; IgG, immunoglobulin G; HAMA, human anti-mouse antibodies.

GFC

GFC was performed with a 1 × 60 cm column (Ultrogel AcA 44^®; IBF, La Garenne, France) equilibrated with 0.01 mol/L sodium phosphate buffer (pH 7.0), containing 0.1 mol/L NaCl, 0.1% (w/v) BSA, and 0.05% (w/v) NaN₃. Serum samples (50–200 μL) were applied to the column, and 1 mL fractions were collected for TSH measurements. The column was calibrated with various molecular weight markers (Sigma–Aldrich, St. Louis, MO).

IgG-bound TSH

To determine the percentage of IgG-bound TSH, protein G Sepharose^® (GE Healthcare, Uppsala, Sweden) was used, which characteristically binds IgG. Serum samples (100–200 μL) were applied to a protein G column equilibrated with 0.02 mol/L of sodium phosphate buffer (pH 7.0), and the column was washed with 4 mL of the same buffer. Then, the bound IgG-associated TSH was eluted with 4 mL of 0.1 mol/L glycine-HCl buffer (pH 2.7) and immediately neutralized with 1 mol/L of Tris-HCl buffer (pH 9.0). The amount of IgG-bound TSH (%) was calculated with the equation: TSH in the bound fraction/(TSH in the unbound fraction + TSH in the bound fraction) × 100.

HAMA

HAMA can produce spuriously high TSH values in sandwich-type immunoassay systems that employ two kinds of mouse monoclonal antibodies because HAMA bridge the capture antibody and detection antibody in the absence of TSH (12,13). The EIA system used in this study did not employ techniques to prevent the influence of HAMA. Therefore, the TSH in sera containing HAMA behaves like macro-TSH on protein G and GFC, causing spuriously high serum TSH concentrations. To determine if the serum contained HAMA, a mixture of three HAMA blockers was added to the EIA system to counteract the effects of HAMA: mouse IgG (THBR1) and mouse IgM (THBR2), both from Cosmobio (Tokyo, Japan), and mouse whole serum from MP Biomedicals (Santa Ana, CA). The percentage of (TSH concentration in the presence of HAMA blockers)/(TSH concentration in the absence of HAMA blockers) was 106.8 ± 12.2% in the 20 controls. The serum was considered as containing HAMA if the percentage was <82.4% (mean − 2 SD in the controls by the addition of HAMA blockers).

¹²⁵I-hTSH binding study

Human TSH (hTSH-SIAFPI-8) was radio-iodinated with Na¹²⁵I (PerkinElmer, Waltham, MA) in an IODO-GEN^® pre-coated iodination tube (Pierce, Rockford, IL) according to the manufacturer's instructions. An ¹²⁵I-hTSH binding study was performed by incubating 20 μL of serum with ¹²⁵I-hTSH (20,000 cpm/4 μL) in the presence or absence of excess unlabeled hTSH (1 μg) at 37°C for 1 h, after which 24 μL of 25% cold PEG solution was added and mixed vigorously. The mixture was centrifuged, and the pellet was washed once with 500 μL of 12.5% PEG. The mixture was centrifuged again, and the radioactivity in the pellet was measured with a gamma counter (Wizard 3; PerkinElmer).

Statistical analysis

All measurements are expressed as the mean ± standard deviation (SD). Values differing from the mean ± 2 SD were considered to be significantly different. The Kolmogorov–Smirnov test (StatMate III; 3B Scientific Co., Niigata, Japan) was used to examine whether the data from TSH precipitation ratios by PEG were normally distributed (Gaussian distribution).

Results

Figure 1A shows the distribution of TSH precipitation ratios by PEG in 305 female patients of reproductive age with serum TSH concentrations >2.5 mIU/L and normal fT4 concentrations. The precipitation ratios varied widely from 17.6% to 96.2% (M ± SD = 63.4 ± 12.3%), and exhibited a normal (Gaussian) distribution.

FIG. 1.

(A) Distributions of the polyethylene glycol (PEG) precipitation ratios of thyrotropin (TSH) among sera of 305 female patients who had serum TSH >2.5 mIU/L and normal free thyroxine (fT4) concentrations. The mean ± standard deviation (SD) precipitation ratio was 63.4 ± 12.3%. Shading indicates that serum samples that contained high molecular weight TSH, as shown by gel filtration chromatography (GFC), were included. Upper numbers indicate the number of samples that contained high molecular weight TSH; lower numbers indicate the total number of serum samples within that range. (B) Effects of human γ-globulin (γ-G) on the precipitation of human TSH reference preparation (hTSH) with PEG. hTSH was incubated at room temperature overnight in 50 μL of 0.01 mol/L phosphate buffer (pH 7.0) containing 0.1 mol/L NaCl (phosphate-buffered saline [PBS]), PBS with 40 mg/mL of bovine serum albumin (BSA), PBS with 10 mg/mL γ-G, and PBS with 40 mg/mL of BSA plus 10 mg/mL of γ-G. These samples were treated with 50 μL of 25% PEG, and hTSH concentrations in the supernatant were measured. The percentage of the precipitation ratios of hTSH by PEG treatment was calculated as follows: (hTSH in PBS − hTSH in the supernatant in PBS containing BSA and/or γ-globulin)/hTSH in PBS × 100 (n = 11). *p < 0.001 compared to results in PBS containing BSA. **p < 0.001 compared to results in PBS containing BSA and those in PBS containing γ-G.

The precipitation ratios presented in Figure 1A were high. So, the effects of γ-globulin on the precipitation of human TSH reference preparation (hTSH) by PEG were examined in the serum-free condition. As shown in Figure 1B, 10 mg/mL of human γ-globulin significantly increased the precipitation ratios of hTSH by PEG (26.5 ± 5.7%; p < 0.001) compared to those measured in the presence of 40 mg/mL of BSA (15.8 ± 4.3%). Human γ-globulin (10 mg/mL) plus BSA (40 mg/mL) further increased the ratios to 58.7 ± 2.3% (p < 0.001), almost comparable to the average TSH precipitation ratios by PEG in sera.

The presence of high molecular weight TSH was assessed using GFC in the sera from 63 patients in the sample population who had TSH precipitation ratios >70% (mean +0.5 SD in 305 female patients). High molecular weight TSH (molecular mass >150 kDa) including spurious results due to HAMA was found in 27 of the 63 patients. The numbers above the gray columns in Figure 1A indicate the prevalence of high molecular weight TSH in the range of the TSH precipitation ratios by PEG. The numbers above the lines represent those of the serum samples that had high molecular weight TSH, and the numbers below the lines indicate the total number of serum samples within the range. The greater the TSH precipitation ratios were, the more frequently high molecular weight TSH was observed.

The clinical characteristics of the 27 patients with high molecular weight TSH are shown in Table 2. Of these patients, 21 had infertility issues, four complained of irregular menstruation, one had ovarian failure, and one had hyperprolactinemia. The percentage of high molecular weight TSH was 79.6 ± 24.4% in these patients compared to 1.1 ± 2.7% in the controls. IgG-associated TSH was elevated in all 27 patients, with ratios that ranged from 17.1% to 99.8% (72.8 ± 21.5%) compared to 0.4 ± 2.0% in the controls. Serum TSH concentrations significantly decreased (<82.4%; mean − 2 SD in the controls) by the addition of HAMA blockers in 24 of these 27 samples (samples 4–27) after the addition of HAMA blockers to the sera, suggesting that these sera contained HAMA that could have behaved as macro-TSH: spuriously elevated serum TSH concentrations, high molecular weight TSH on GFC, and IgG-associated TSH by protein G. The remaining three patients (patients 1–3) were diagnosed as having macro-TSH.

Table 2.

Clinical Characteristics of Patients Whose Sera Contained Large Molecular Weight TSH on Gel Filtration Chromatography

Patients	Age (years)	Diagnosis	fT4 (ng/dL)	TSH^a (mIU/L)	Free TSH^b (mIU/L)	PEG^c (%)	GFC^d (%)	PG^e (%)	HAMA^f (%)
1	38	Irregular menses	1.36	14.93	1.58	89.4	93.3	27.3	109.0
2	34	Infertility	1.25	4.07	0.59	85.6	79.4	61.4	103.0
3	32	Irregular menses	1.11	4.43	1.22	72.4	27.6	54.5	104.5
4	37	Infertility	1.33	12.21	1.18	90.4	66.7	17.1	50.0
5	35	Infertility	1.00	3.80	0.86	77.4	100.0	94.0	36.7
6	27	Infertility	1.29	12.94	0.54	95.8	95.7	95.4	8.8
7	30	Infertility	1.33	6.33	1.58	75.0	100.0	88.1	46.8
8	29	Infertility	1.50	5.20	1.27	75.7	60.9	60.0	73.7
9	27	Ovarian failure	1.01	9.91	2.35	76.3	96.8	89.1	33.6
10	30	Infertility	1.29	12.58	4.48	74.4	23.6	40.5	77.8
11	32	Infertility	1.23	7.33	1.58	78.4	81.0	65.7	56.7
12	23	Irregular menses	1.05	10.18	2.04	80.0	95.6	82.4	20.7
13	39	Infertility	1.62	8.64	1.40	83.8	94.3	94.4	39.8
14	31	Infertility	1.43	4.12	0.63	84.6	79.7	74.1	41.5
15	33	Infertility	1.11	8.19	0.63	92.3	99.3	95.5	9.6
16	35	Infertility	1.40	5.47	1.18	78.5	86.6	73.3	26.2
17	37	Infertility	1.69	9.95	2.44	75.5	100.0	78.8	48.0
18	33	Infertility	1.57	4.43	0.77	82.7	78.6	66.7	42.4
19	35	Infertility	1.37	7.15	0.81	88.6	95.6	92.9	15.7
20	37	Infertility	1.48	5.93	0.23	96.2	100.0	99.8	9.1
21	31	Infertility	1.26	4.30	0.59	86.3	83.7	77.8	35.9
22	30	Infertility	0.94	4.12	0.86	79.1	100.0	87.7	48.1
23	28	Hyperprolactinemia	1.07	3.44	0.81	76.3	78.5	46.2	52.6
24	36	Infertility	1.02	2.94	0.63	78.5	92.1	60.0	40.0
25	38	Infertility	1.04	4.07	0.86	78.9	64.4	75.0	54.5
26	22	Irregular menses	0.99	3.03	0.63	79.1	62.8	88.0	29.6
27	33	Infertility	1.29	6.15	2.22	74.0	13.4	79.4	75.8
Mean	32.3		1.26	6.88	1.26	81.7	79.6	72.8
SD	4.4		0.21	3.42	0.88	6.8	24.4	21.5
Control (n = 20)
Mean	32.1		1.10	5.85	2.68	54.0	1.1	0.4	106.8
SD	5.9		0.19	1.28	0.66	6.3	2.7	2.0	12.2

TSH concentrations were measured by an enzyme immunoassay (EIA), which was also used to evaluate free TSH, polyethylene glycol (PEG), gel filtration chromatography (GFC), protein G (PG), and human anti-mouse antibodies (HAMA) experiments. HAMA blockers were not included in the routine usage of EIA, except for HAMA blocker experiments.

Free TSH: TSH concentrations in the supernatant after PEG treatment of sera.

PEG: TSH precipitation ratios by polyethylene glycol (PEG) were defined as (TSH concentrations − free TSH concentrations)/TSH concentrations × 100 (%).

GFC: the proportion of high molecular weight TSH determined by GFC.

PG: the proportion of TSH that bound to the protein G (PG) columns (IgG-associated TSH).

HAMA: the ratios of (TSH concentration in the presence of HAMA blockers)/(TSH concentration in the absence of HAMA blockers) are shown. The percentage <82.4% (mean − 2 SD in the controls) indicates that the serum contained HAMA, which interfered with the EIA assay and spuriously increased the serum TSH concentration results.

Figure 2 shows immunoreactive TSH profiles separated by GFC in the sera from the three patients with macro-TSH, representative ones from those with HAMA, and from controls. In patients with HAMA, the peak associated with high molecular weight TSH disappeared when TSH concentrations were assayed in the fractions containing HAMA blockers (Fig. 2D, open circles).

FIG. 2.

(A–C) Immunoreactive TSH profiles on GFC in sera containing macro-TSH, (D) representative profiles in the sera containing human anti-mouse antibodies (HAMA), and (E) control sera. The profile of TSH shown with open circles in (D) was demonstrated when the TSH concentration in each fraction was measured in the presence of HAMA blockers.

The involvement of anti-TSH autoantibodies was examined in three patients with macro-TSH (Fig. 3). The amount of ¹²⁵I-labeled TSH bound to the serum from patient 1 was high and was displaced by the addition of excess unlabeled TSH, indicating the presence of anti-TSH autoantibodies in this serum sample. The binding was within the reference range in the serum from patient 2. Binding of the labeled TSH in the serum of patient 3 was elevated, but it was not displaced by excess unlabeled TSH.

FIG. 3.

Binding of ¹²⁵I-hTSH (20,000 cpm per assay) in sera from the three patients with macro-TSH in the absence (gray bars) and presence (white bars) of excess unlabeled hTSH. The dotted line and shaded area show the mean ± 2 SD of the bound ¹²⁵I-hTSH in the sera from 20 controls.

The sera of 36 of the 63 patients with TSH precipitation ratios by PEG >70% did not have high molecular weight TSH when assessed by GFC. Studies with HAMA blockers and protein G were also performed in these samples (Table 3). The TSH concentrations in the sera were not reduced by HAMA blockers: the ratio of TSH concentrations with HAMA blockers to the TSH concentrations without HAMA blockers ranged from 88.9% to 120% (105.0 ± 7.6%). In eight of the 36 patients, the ratios of IgG-associated TSH were elevated and ranged from 7.2% to 50.0% (24.5 ± 14.0%) compared to 0.4 ± 2.0% in the controls. This suggests that these eight patients have high TSH precipitation ratios by PEG and IgG-associated TSH, but no evidence of high molecular weight TSH by GFC.

Table 3.

Characterization of 63 Patients with TSH Precipitation Ratios by PEG >70%

GFC	High molecular weight TSH^a (+)	High molecular weight TSH (−)
n	27	36
TSH precipitation ratios by PEG^b	81.7 ± 6.8% (72.4–96.2%)	76.0 ± 3.3% (71.2–86.4%)
HAMA (+)^c	24	0
Protein G (+)^d	27	8
	72.8 ± 21.5%	24.5 ± 14.0%
	(17.1–99.8%)	(7.2–50.0%)

The presence of high molecular weight TSH was examined using GFC. Spurious results due to HAMA are included.

TSH precipitation ratios by PEG were calculated as follows: (serum TSH − free TSH)/serum TSH × 100. Mean ± SD (range).

HAMA (+): The serum was judged to contain HAMA if the percentage of (TSH concentration in the presence of HAMA blockers)/(TSH concentration in the absence of HAMA blockers) was <82.4% (mean − 2 SD in 20 controls [106.8 ± 12.2%]).

Protein G (+): The serum was judged to contain protein G-bound TSH (IgG-bound TSH) if the percentage of TSH in the bound fraction/(TSH in the unbound fraction + TSH in the bound fraction) × 100 was >4.4% (mean +2 SD) in the controls (0.4 ± 2.0%). Spurious results due to HAMA are included.

The influence of macro-TSH and HAMA on TSH measurements in sera (with TSH >2.5 mIU/L as measured by HAMA blockers containing EIA) was assessed with three commercially available TSH immunoassay systems (Elecsys^®, Centaur^®, and Architect^®) and EIA including HAMA blockers. The results are shown in Table 4. Taking the high nonspecific TSH precipitation ratios by PEG into account, true TSH concentrations were estimated from the free TSH values. The ratio of serum TSH to free TSH concentrations in the 20 controls was 2.21 ± 0.27. The estimated TSH concentrations were determined by the equation: free TSH concentrations × 2.21 [mean of (TSH/free TSH) in 20 controls]. The upper limit of the estimated TSH concentrations was determined by the equation: free TSH concentrations × 2.75 [mean +2 SD of (TSH/free TSH) in 20 controls]. Serum TSH values greater than the upper limit of the estimated TSH concentrations were judged as being influenced by macro-TSH or HAMA or heterophile antibodies. In patients 1–3, whose sera contained macro-TSH, and in patient 4, whose serum contained HAMA, the TSH concentrations measured by the four immunoassays were all above the upper limit.

Table 4.

Influence of Macro-TSH or HAMA or Heterophile Antibodies on TSH Measurements by Immunoassay Systems

	Free TSH^b	Estimated TSH^c	Upper limit^d		EIA HAMA blocker (+)	Elecsys^®	Centaur^®	Architect^®
Patients^a	(mIU/L)	(mIU/L)	(mIU/L)	HAMA^e	(mIU/L)	(mIU/L)	(mIU/L)	(mIU/L)
1	1.58	3.49	4.35	−	16.27^*	11.52^*	9.60^*	9.71^*
2	0.59	1.30	1.62	−	4.19^*	4.26^*	3.84^*	4.12^*
3	1.22	2.70	3.36	−	4.63^*	4.11^*	3.78^*	3.80^*
4	1.18	2.61	3.25	+	6.11^*	4.14^*	4.08^*	4.03^*
7	1.58	3.49	4.35	+	2.96	4.36^*	4.38^*	3.90
8	1.27	2.81	3.49	+	3.83^*	3.82^*	3.30	3.24
9	2.35	5.19	6.46	+	3.33	4.00	3.78	4.08
10	4.48	9.90	12.32	+	9.78	7.02	6.24	6.46
11	1.58	3.49	4.35	+	4.16	3.85	3.24	3.65
13	1.40	3.09	3.85	+	3.44	0.84	0.78	0.76
17	2.44	5.39	6.71	+	4.78	4.93	4.44	4.46
27	2.22	4.91	6.11	+	4.66	5.53	4.62	5.01

Out of 27 patients with high molecular weight TSH as shown by GFC, those with serum TSH concentrations >2.5 mIU/L measured by EIA in the presence of HAMA blockers are shown.

Serum samples were treated with 25% PEG (final concentration of PEG was 12.5%) to precipitate γ-globulin fractions, and free TSH concentrations were defined as those in the supernatant.

The estimated TSH concentrations were determined by the equation: free TSH concentrations × 2.21 (mean of [TSH/free TSH] in 20 controls).

The upper limits of the estimated TSH concentrations were determined by the equation: free TSH concentrations × 2.75 (mean +2 SD of [TSH/free TSH] in 20 controls).

The serum was determined as containing HAMA when the percentage of (TSH concentration in the presence of HAMA blockers)/(TSH concentration in the absence of HAMA blockers) was <82.4% (mean − 2 SD in the controls by the addition of HAMA blockers).

Serum TSH values greater than the upper limit of the estimated TSH concentrations are judged as being influenced by macro-TSH or HAMA or heterophile antibodies.

Discussion

This study surveyed the presence of macro-TSH in 1794 females of reproductive age who were mainly experiencing infertility. Among this cohort, 305 patients had serum TSH >2.5 mIU/L and normal fT4 concentrations. These patients were chosen to screen for macro-TSH because for the following reasons. Negro et al. (14) reported a significantly higher pregnancy loss rate in thyroid peroxidase (TPO) autoantibody-negative women with TSH concentrations between 2.5 and 5.0 mIU/L compared to those with TSH concentrations <2.5 mIU/L. Recently published guidelines of the American Thyroid Association for the diagnosis and management of thyroid disease during pregnancy and postpartum recommend evaluation of the TPO autoantibody in pregnant women with TSH concentrations >2.5 mIU/L when determining the therapeutic strategy for subclinical hypothyroidism (15). It is most likely that macro-TSH is present in high concentrations in serum because its large molecular size causes it to be cleared slowly. In addition, since the biological activity of macro-TSH is low, fT4 concentrations in serum may be normal, mimicking subclinical hypothyroidism (10).

The PEG precipitation method was used to screen for macro-TSH. Macro-TSH is mostly associated with IgG (4 –11), and PEG precipitates γ-globulin fractions. Therefore, TSH precipitation ratios are expected to be high if the sera contain macro-TSH. However, the average serum TSH precipitation ratios by PEG (63.4 ± 12.3%) were significantly higher than those of prolactin (36.6 ± 10.2%) (16). To determine why these precipitation ratios were unexpectedly high, the precipitation of hTSH by PEG was assessed in the absence of serum. It was observed that phosphate buffer containing γ-globulin plus albumin exhibited comparable precipitation ratios to those of serum TSH. The high precipitation ratios of serum TSH may be attributable to the nature of the TSH molecule, which is prone to be precipitated with PEG in the presence of γ-globulin plus albumin. The high nonspecific serum TSH precipitation ratios and their broad distribution led us to set the cutoff value for further analysis using GFC and protein G at 70% (mean +0.5 SD in 305 patients), so as not to miss serum samples that contained macro-TSH.

Human anti-mouse antibodies bind and link capture and detection mouse monoclonal antibodies, producing false signals in sandwich-type immunoassays without antigens (12,13). The EIA system used in this study was unprotected from HAMA. If the serum contained HAMA, the serum TSH concentration measured by EIA would be high, TSH precipitation ratios by PEG would be elevated, GFC would show a high molecular weight TSH, and IgG-associated TSH separated by the protein G column would increase. However, all these results would be due to the interference of HAMA in the immunoassays, and would not reflect true TSH concentrations. Sixty-three samples had serum TSH precipitation ratios by PEG >70%, and 27 of these had high molecular weight TSH when assessed by GFC. In 24 of the 27 samples, TSH concentrations were significantly suppressed by HAMA blockers, consistent with the presence of HAMA. Because it is unlikely that HAMA was involved in the sera that showed neither high serum TSH concentrations nor high TSH precipitation ratios by PEG, the prevalence of HAMA in females of reproductive age would be 1.3% (24/1794).

TSH concentrations in the remaining three of the 27 serum samples were not suppressed by HAMA blockers, suggestive of the presence of macro-TSH. Macro-TSH in one of them was a complex of TSH with anti-TSH autoantibodies because a significant amount of the ¹²⁵I-hTSH was bound to the serum and the binding was displaced by excess unlabeled hTSH. In one of the remaining two patients, ¹²⁵I-hTSH bound to the serum at a significantly higher ratio compared to controls, but the binding was not displaced by excess unlabeled TSH, suggesting either that the binding capacity of the autoantibodies was very high or that TSH bound to other serum components. In another patient, ¹²⁵I-hTSH did not bind to the serum, which made the presence of autoantibodies unlikely.

Within the group of 63 patients with precipitation ratios >70%, there were 36 patients whose sera did not have high molecular weight TSH as assessed by GFC. None of these serum samples contained HAMA. This result is logically expected because HAMA elute at the position of γ-globulin (molecular weight approximately 150 kDa) and produce a spurious high molecular weight peak in the TSH profile of GFC. It is interesting that eight out of the 36 patients had significantly higher binding ratios of serum TSH to the protein G column, indicating that TSH might be associated with IgG in these patients. GFC has been used as the gold standard for the diagnosis of macroprolactinemia (17). However, dissociation of TSH from the binding components might occur during GFC. Blunted peaks and poor separation of serum TSH in patients with macro-TSH (patients 2 and 3) might be evidence of this occurrence. Thus, it is possible that these eight patients also have macro-TSH consisting of TSH and weakly bound anti-TSH autoantibodies. If so, 11 patients in this study would have macro-TSH (3 + 8), and the prevalence of macro-TSH in infertile women would be between 0.17% (3/1794) and 0.61% (11/1794). It should be noted that this prevalence of macro-TSH may not be replicated in all females of reproductive age because TSH is routinely assessed in infertile women but not in the general population.

None of the methods are perfect, and all of them have shortcomings. PEG at a final concentration of 12.5% mostly precipitates γ-globulin fraction (18), but not all IgA class immunoglobulins (19). A false-negative PEG precipitation test was reported in a patient with an IgA macroprolactin (20). Thus, screening by PEG may miss macro-TSH consisting of TSH and IgA-type anti-TSH autoantibodies if they exist. Protein G binds IgG but not IgA (instructions of protein G; GE Healthcare). Therefore, this method cannot detect IgA-type macro-TSH either. Blockers of HAMA (high affinity anti-mouse antibodies directed against the assay reagents) may not be able to eliminate the interference by heterophile antibodies (weak polyspecific antibodies directed against the assay reagents) (21) if they do not carry these epitopes. GFC is not a sensitive method for macro-TSH in which TSH and the autoantibodies are weakly bound.

Commercial TSH immunoassays employ several methods to avoid the interference of HAMA and heterophile antibodies, such as using mouse serum components, antibodies of different species, and chimeric antibodies (22,23). Treatment with PEG could mostly eliminate the influence of macro-TSH, as well as HAMA or heterophile antibodies. Thus, free TSH concentrations after treatment with PEG could be used to estimate serum TSH values unaffected by these factors, although the background TSH precipitation ratios were high and variable. In three patients with macro-TSH, TSH concentrations measured by the three commercial immunoassays and EIA including HAMA blockers all exceeded the estimated upper limits of TSH values calculated from the free TSH concentrations. In other words, all the immunoassays recognized macro-TSH and serum TSH concentrations were overestimated due to macro-TSH in these patients. Although the serum from patient 4 contained HAMA, TSH concentrations measured by all four immunoassays exceeded the estimated upper limit of TSH. This serum characteristically showed a high TSH precipitation ratio by PEG (90.4%), but the involvement of IgG was the least (17.1% of protein G-bound TSH) among the 24 HAMA-containing sera, raising the possibility that some additional factors other than HAMA such as an immunoglobulin class/subclass that does not interact with protein G might be involved in this patient.

Hormone replacement therapy is unnecessary when elevated serum TSH is due to macro-TSH because of the low bioactivity of macro-TSH (4,10). Therefore, diagnosing macro-TSH is important, especially in females of reproductive age, because misdiagnosis may lead to unnecessary long-term replacement therapy. The optimal time and method for the diagnosis of macro-TSH in females of reproductive age has not been established. In this study, the prevalence of serum TSH concentrations >2.5 mIU/L was 17% (305/1794 patients in total) and the prevalence of TSH precipitation ratios >70% among them was 21% (63/305). However, these values would have been lower if HAMA blockers had been added to the present EIA. If the more commonly used cutoff value of 4 mIU/L measured by immunoassays protected from HAMA or heterophile antibodies were to be applied, fewer serum samples would have been screened for macro-TSH. At present there is no single method that can reliably detect macro-TSH. The cost performance of the PEG method is good (cheap and easy), whereas GFC and protein G methods are time and labor intensive. At present, diagnosis of macro-TSH needs purification of high molecular weight TSH by PEG, GFC, and protein A/G in which TSH should be measured by immunoassays protected from HAMA and heterophile antibodies. A ¹²⁵I-TSH binding study can help establishing the diagnosis. To avoid these problems, developing TSH immunoassays that do not recognize macro-TSH would be desirable, similar to the prolactin immunoassays, some of which are designed to minimize the influence of macroprolactin (24,25).

Footnotes

Acknowledgments

We thank Geoff Gillespie for his assistance in preparing this manuscript. This work was supported by grants from the Ministry of Culture, Sports, Science, and Technology of Japan, from the Ministry of Health, Labor, and Welfare of Japan, and from Ritsumeikan University.

Author Disclosure Statement

No competing financial interests exist.

References

Biondi

, Cooper

. 2008. The clinical significance of subclinical thyroid dysfunction. Endocr Rev, 29:76–131.

Moleti

, Trimarchi

, Vermiglio

. 2014. Thyroid physiology in pregnancy. Endocr Pract, 20:589–596.

Maraka

, Ospina

, O'Keeffe

, Espinosa De Ycaza

, Gionfriddo

, Erwin

, Coddington CC 3rd, Stan

, Murad

, Montori

. 2016. Subclinical hypothyroidism in pregnancy: a systematic review and meta-analysis. Thyroid, 26:580–590.

Spitz

, Le Roith

, Hirsch

, Carayon

, Pekonen

, Liel

, Sobel

, Chorer

, Weintraub

. 1981. Increased high-molecular-weight thyrotropin with impaired biologic activity in a euthyroid man. N Engl J Med, 304:278–282.

Tamaki

, Takeoka

, Nishi

, Shindoh

, Tsukada

, Amino

. 1995. Novel thyrotropin (TSH)-TSH antibody complex in a healthy woman and her neonates. Thyroid, 5:299–303.

Verhoye

, Van den Bruel

, Delanghe

, Debruyne

, Langlois

. 2009. Spuriously high thyrotropin values due to anti-thyrotropin antibodies in adult patients. Clin Chem Lab Med, 47:604–606.

Rix

, Laurberg

, Porzig

, Kristensen

. 2011. Elevated thyroid-stimulating hormone level in a euthyroid neonate caused by macro thyrotropin-IgG complex. Acta Paediatr, 100:e135–e137.

Loh

, Kao

, Halsall

, Toh

, Chan

, Ho

, Tai

, Khoo

. 2012. Macro-thyrotropin: a case report and review of literature. J Clin Endocrinol Metab, 97:1823–1828.

Mills

, Jeffery

, Mackenzie

, Cranfield

, Ayling

. 2013. An immunoglobulin G complexed form of thyroid-stimulating hormone (macro thyroid-stimulating hormone) is a cause of elevated serum thyroid-stimulating hormone concentration. Ann Clin Biochem, 50:416–420.

10.

Hattori

, Ishihara

, Yamagami

, Shimatsu

. 2015. Macro TSH in patients with subclinical hypothyroidism. Clin Endocrinol, 83:923–930.

11.

Hattori

, Ishihara

, Shimatsu

. 2016. Variability in the detection of macro TSH in different immunoassay systems. Eur J Endocrinol, 174:9–15.

12.

Kricka

. 1999. Human anti-animal antibody interferences in immunological assays. Clin Chem, 45:942–956.

13.

Koshida

, Asanuma

, Kuribayashi

, Goto

, Tsuji

, Kobayashi

, Tanaka

, Watanabe

. 2010. Prevalence of human anti-mouse antibodies (HAMAs) in routine examinations. Clin Chim Acta. 411:391–394.

14.

Negro

, Schwartz

, Gismondi

, Tinelli

, Mangieri

, Stagnaro-Green

. 2010. Increased pregnancy loss rate in thyroid antibody negative women with TSH concentrations between 2.5 and 5.0 in the first trimester of pregnancy. J Clin Endocrinol Metab, 95:E44–E48.

15.

Alexander

, Pearce

, Brent

, Brown

, Chen

, Dosiou

, Grobman

, Laurberg

, Lazarus

, Mandel

, Peeters

, Sullivan

. 2017. Guidelines of the American Thyroid Association for the diagnosis and management of thyroid disease during pregnancy and the postpartum. Thyroid, 27:315–389.

16.

Hattori

, Ishihara

, Saiki

. 2009. Macroprolactinaemia: prevalence and aetiologies in a large group of hospital workers. Clin Endocrinol, 71:702–707.

17.

Gibney

, Smith, TP McKenna

. 2005. The impact on clinical practice of routine screening for macroprolactin. J Clin Endocrinol Metab, 90:3927–3932.

18.

Fried

, Chun

. 1971. Water-soluble nonionic polymers in protein purification. Methods Enzymol, 22:238–248.

19.

Ismail

. 2005. A radical approach is needed to eliminate interference from endogenous antibodies in immunoassays. Clin Chem, 51:25–26.

20.

Kilvington

, Rahim

, Kavanagh

, Smith

, Fahie-Wilson

. 2006. False negative PEG precipitation test in a case of hyperprolactinaemia with an IgA macroprolactin [Abstract T12]. Ann Clin Biochem, 43:35.

21.

Levinson

, Miller

. 2002. Towards a better understanding of heterophile (and the like) antibody interference with modern immunoassays. Clin Chim Acta, 325:1–15.

22.

Yamashita

, Kobayashi

, Tatematsu

, Sezutsu

, Noda

, Ishihara

. 2016. Sandwich ELISA using a mouse/human chimeric CSLEX-1 antibody. Clin Chem, 62:1516–1523.

23.

, Foster

, Datta

, Orswell

, Tasaico

, Mai

, Connolly

, Reamer

, Walsh

, Yang

, Barlow

, Bluestein

, Parsons

. 2000. Use of a dual monoclonal solid phase and a polyclonal detector to create an immunoassay for the detection of human cardiac troponin I. Clin Biochem, 33:255–262.

24.

Smith

, Suliman

, Fahie-Wilson

, McKenna

. 2002. Gross variability in the detection of prolactin in sera containing big big prolactin (macroprolactin) by commercial immunoassays. J Clin Endocrinol Metab, 87:5410–5415.

25.

Hattori

, Aisaka

, Shimatsu

. 2016. A possible cause of the variable detectability of macroprolactin by different immunoassay systems. Clin Chem Lab Med, 54:603–608.

Current Thyrotropin Immunoassays Recognize Macro-Thyrotropin Leading to Hyperthyrotropinemia in Females of Reproductive Age

Abstract

Background:

Methods:

Results:

Conclusions:

Introduction

Materials and Methods

Patients

TSH and fT4 assays

PEG precipitation

GFC

IgG-bound TSH

HAMA

125I-hTSH binding study

Statistical analysis

Results

Discussion

Footnotes

Acknowledgments

Author Disclosure Statement

References

¹²⁵I-hTSH binding study