Key words
Antibodies - Autoimmunity - Thyroid Autoimmunity
Introduction
Autoimmune thyroid disease (AITD) is an organ-specific disease and the most common
human autoimmune endocrine disorder. AITD may present with hyperthyroidism, as in
Graves’ disease, or as hypothyroidism in Hashimoto’s thyroiditis,
the latter being much more common [1]
[2].
AITD is characterized by the presence of autoantibodies towards thyroid peroxidase
(TPOab) and thyroglobulin (TGab). Thyroid peroxidase is a membrane-anchored enzyme
in the thyroid cell with crucial functions in the process of thyroid hormone
synthesis [1]. TPOab has been reported to be
the most abundant thyroid autoantibody and the strongest predictor of ongoing
autoimmunity [3]
[4]. TPO-1 is a 933 amino acid (aa) long,
transmembrane-spanning protein; therefore, reproducing correct folding in an in
vitro system might be challenging. No single epitope has thus far been
described for autoantibody interaction, but immunodominant regions have been defined
in three regions across the TPO protein, from N to C-terminal: aa 210–225,
aa 549–563, and aa 599–617 [5]
[6]
[7]
[8]
[9]. Autoantibodies are
preferably measured in the liquid phase for the detection of conformation-dependent
epitopes [10]. Reduced TPOab binding following
protein denaturation has previously been described [11] and can also occur in other autoimmune diseases, including type 1
diabetes [12]
[13]. Previous studies have suggested that TPOab in healthy and clinically
diagnosed patients is directed towards similar epitopes, but only patient-derived
TPOab was found to interfere with enzyme activity and complement fixation processes
[1]
[14]
[15].
Thyroglobulin is an intracellular protein expressed in the thyroid follicular cells
and associated with thyroid hormone storage. The pathological role of TGab is less
defined compared to TPOab, although the prevalence is comparable [4]
[16].
AITD predominates in females with ratios being reported between 4-10/1
(female/male). [2]. The reported
prevalence of TPOab in the general population differs substantially between studies
and countries [17], including Australia (TPOab
female 15.0%, male 6.6%) [18],
Norway (TPOab female 13.9%, male 2.8%) [19], US (TPOab female 17.0%, male
8.7%) [4], China (TPOab female
11.7%, male 4.2%) [20]. This
discrepancy could be explained by natural differences between geographical regions
caused by genetic associations of human leukocyte antigen (HLA) and environmental
factors, similar to other common pediatric autoimmune diseases, type 1 diabetes, and
celiac disease [21]
[22]. An additional explanation for the
discrepancy in the reported prevalence of TPOab could be the variation in
sensitivity and specificity of antibody detection assays, together with the absence
of adequate international reference standards for cut-off determination [23].
This study aimed to develop and validate a novel TPOab radiobinding assay (RBA) and
determine the prevalence of TPOab and TGab in the Swedish population.
Methods
Study populations
All experiments including human participants were ethically approved prior to
sample collection. The control study groups included (demographics, [Table 1]):
Table 1 Demographics of the study cohort.
|
n
|
Born
|
Gender (female, %)
|
|
Blood donors
|
476
|
1947–2000
|
n=193 (40.6%)
|
|
Schoolchildren
|
297
|
1996
|
n=157 (52.9%)
|
-
Blood donors (n=476; n=475 included to assess
TPOab diagnostic validity and prevalence). Serum samples were collected
at the blood-donation center in Malmö, Sweden, during the spring
of 2019. Blood samples were donated anonymously; only age and gender
were reported.
-
School children (n=297; n=295 included to assess
TGab diagnostic validity and prevalence) [24]. Serum samples from 13-year-old
schoolchildren were collected in Jämtland county, Sweden,
between April 2009 to January 2011, on separate occasions; April,
October, and November 2009, May, November and December 2010, and January
2011.
Two patient cohorts were included for the purpose of assay validation (a and b
were also used during validation):
-
Graves’ disease patients. Samples were donated at the
clinic at 6 weeks (n=27) and 6 months (n=20) following
diagnosis.
-
Thyroid autoantibody-positive children (n=124,
n=109 TGab diagnostic validity). This group of children from the
Diabetes Prediction in Skåne Study (DiPiS) has been described
previously [25]. In short, TPOab
and/or TGab-positive children at 10 years of age were asked to
leave a confirmatory sample between 11–16 years of age. Samples
were previously analyzed using radioimmunoassay (RIA) (RSR Limited, art.
no. RS-TP/100 and RS-TG/100 respectively) and
electrochemiluminescence (ECL) (Roche, Anti-TPO REF 190,06368590). The
analyses were performed according to the manufacturer’s
instructions [25].
Thyroid peroxidase autoantibody determination using radio-binding
assay
The principle of the RBA [26] is based on
the interaction between blood-derived TPO autoantibodies and
35S-radiolabeled TPO-antigen. Overnight incubation enables the
formation of autoantibody-antigen complexes. Precipitation of these complexes is
mediated using Protein A Sepharose. A higher concentration of measured TPO
autoantibodies is correlated to the amount of precipitated radiolabeled
antigen.
The cDNA was constructed to include the amino acid methionine (added for the
intitiation of protein synthesis), followed by TPO aa 204–785. A
standard RBA was developed, including coupled in vitro
transcription-translation (ITT). TPO cDNA constructs were subcloned into a pTNT
vector (Thermo Fisher Scientific). Radiolabeled TPO-antigens were expressed
using rabbit reticulocyte lysate (Promega Corporation) and
35S-methionine (Promega Corporation) at 30°C, 90 min
ITT-reaction. Unlabeled methionine was removed from incorporated antigens
through size-restricted filtration using Nap-5 Columns (GE Healthcare).
Serum samples (2.5 µl) in duplicates were incubated in 96-well
analytical plates (Nunc V96 MicroWell, Nunc A/S) with
60 µL radiolabeled TPO-antigens (diluted to
400 cpm/µL in assay-buffer; 150 mmol/L
NaCl, 20 mmol/L Tris-HCl (pH 7.4), 0,15%
v/v polysorbate 20, 0,1% w/v bovine
serum albumin) at 4 °C overnight. Filtration plates (MultiScreen HTS-DV
Plates, Millipore) were blocked with 2.5% milk solution overnight. The
next day, Protein A Sepharose was diluted to 20% in assay buffer and
then added to filtration plates with 50 µL of the overnight
incubated antigen-antibody mixture. Incubation was for 1 h at
4°C before unbound antigens were removed by vacuum-washing (405LS
Microplate Strip Washer, Biotek Instruments, Inc.). The addition of
scintillation liquid, 50 µL per well, was followed by analysis
in luminescence BetaCounter (1450 MicroBeta TriLux β-counter,
PerkinElmer). TPO autoantibody levels were expressed as in-house arbitrary units
per milliliter (U/mL). A positive serum sample was diluted in nine steps
(1000 U/mL, 500 U/mL, 250 U/mL,
125 U/mL, 63.5 U/mL, 31.25 U/mL,
15.63 U/mL, 7.81 U/mL, and
3.91 U/mL) and used for conversion of counts per minute to
U/mL. The TPO RBA intra-assay CVs, as a measure of precision, and
inter-assay, as a measure of reproducibility were 5.6 and 5.5%,
respectively.
Thyroglobulin electrochemiluminescence autoantibody determination
TG autoantibodies were analyzed using YHLO iFlash 1800 Chemiluminescence
Immunoassay Analyzer (C89003G). The principle of the assay is a sandwich complex
formation of 1) TGab in a patient sample, 2) microbeads coated with
thyroglobulin antigen, and 3) anti-TG acridinium-ester labeled conjugates.
Following magnetic-based separation of unbound material, pre-trigger (hydrogen
peroxide solution) and trigger solutions (sodium hydroxide solution) were added
to the mixture. The resulting chemiluminecent signal was detected as relative
light units (RLUs). Conversion to International Units Per Milliliter
(IU/mL) was automatically calculated by a 3-point standard curve; the
amount of antigen in the sample was proportional to the detected signal of
relative light units.
Two (low control, high control) control samples were included in each run; the
inter-assay as a measure for reproducibility was 5.1%. The TgAb assay
had a calibration range of up to 1000 IU/mL, and levels higher than 1000
IU/mL were reported as 1000 IU/mL.
Statistical analysis
Autoantibody levels were found to be non-normally distributed. Nonparametric
test, two-tailed Mann-Whitney U, were used for comparison of group differences
of continuous variables. Comparison between group frequencies was assessed using
the χ2 test or Fisher’s exact test. Correlations were
assessed using Spearman Rho. The receiver operator characteristic (ROC) analysis
was used to determine the area under the ROC curve (AUC) for the TPO and the TG
autoantibody assays. A p-value≤0.05 was considered statistically
significant. All statistical analyses were conducted using GraphPad Prism (Prism
9.1.0, La Jolla, CA) and MedCalc (MedCalc Software Ltd.: https://www.medcalc.org/features/roccurves.php).
Results
Diagnostic validity
TPOAb in the 27 Graves’ patients and 475 blood donors analyzed by a ROC
curve resulted in an AUC value of 0.82 (0.78–0.85, p<0.0001),
with a sensitivity of 77.8% and a specificity of 91.8% ([Fig. 1a]). The ROC analysis, including 124
thyroid autoantibody-positive children compared to 297 13-year-old
schoolchildren, resulted in an AUC value of 0.70 (0.65–0.74,
p<0.0001). The sensitivity of the assay was 53.2%, and the
specificity was 95.3% ([Fig.
1b]).
Fig. 1 Receiver Operator Characteristic analysis was used to
determine the Area Under the Curve (AUC) for a) Thyroid
peroxidase autoantibodies in the group of patients with Graves’
disease (n=27) compared to adult blood donors (n=475),
to a value of 0.82 (0.78–0.85, p<0.0001), with a
sensitivity of 77.8% and a specificity of 91.8%.
b) Thyroid peroxidase autoantibodies in the group of
autoimmune thyroid disease (AITD) autoantibody-positive children in
prospective follow-up for increased genetic risk of type 1 diabetes
(n=124) compared to 13-year-old schoolchildren (n=297),
to a value of 0.70 (0.65–0.74, p<0.0001), with a
sensitivity of 53.2% and a specificity of 95.3%.
c) Thyroglobulin autoantibodies in the group of patients with
Graves’ disease (n=27) compared to adult blood donors
(n=476), to a value of 0.87 (0.83–0.90,
p<0.0001), with a sensitivity of 85.2% and a specificity
of 88.5%. d) Thyroglobulin autoantibodies in AITD
autoantibody-positive children in prospective follow-up for increased
genetic risk of type 1 diabetes (n=109) compared to 13-year-old
schoolchildren (n=295), to a value of 0.76 (0.72–0.80,
p<0.0001), with a sensitivity of 66.1% and a specificity
of 92.2%.
TGab was determined using ECL. ROC analysis, including 27 Graves’
patients compared to 476 blood donors, resulted in an AUC value of 0.87
(0.83–0.90, p<0.0001), with a sensitivity of 85.2% and a
specificity of 88.5% ([Fig. 1c]).
ROC analysis, including 109 thyroid autoantibody-positive children compared to
295 13-year-old schoolchildren, resulted in an AUC value of 0.76
(0.72–0.80, p<0.0001), with a sensitivity of 66.1% and a
specificity of 92.2% ([Fig.
1d]).
Radiobinding TPOab assay in comparison to standardized
electrochemiluminescence and radioimmunoassay
A strong correlation could be demonstrated for TPOab, in Graves’ patients
(n=27) and thyroid antibody-positive children (n=124), in RBA
compared to ECL (Spearman’s correlation coefficient
ρ=0.90 (0.85–0.93), p<0.0001, [Fig. 2a]) and compared to RIA
(Spearman’s correlation coefficient ρ=0.93
(0.90–0.95), p<0.0001, [Fig.
2b]). A total of 16 individuals deviated from RBA to ECL, with
3/124 (2.5%) being RBA-only positive and 13/124
(10.5%) being ECL-only positive. In comparison to RBA, 31/124
(25%) individuals were RIA-only positive.
Fig. 2 Correlations were assessed in a combined group of
Graves’ diseases patients (n=27) and thyroid
autoantibody-positive children (n=124) and could be demonstrated
for thyroid peroxidase radiobinding assay (RBA) in comparison to
a) Electrochemiluminescence assay (Spearman’s
correlation coefficient ρ=0.90 (0.85–0.93),
p<0.0001)). A total of 16 individuals diverged between RBA
compared to ECL, with 3/124 (2.5%) being RBA-only positive and
13/124 (10.5%) being ECL-only positive. b)
Radioimmunoassay (Spearman’s correlation coefficient
ρ=0.93 (0.90–0.95), p<0.0001)). Compared
to RBA, 31/124 (25%) individuals were RIA-only positive, and
none were RBA-only positive.
Thyroid autoantibody prevalence
The prevalence of TPOab was 6.3% (30/475) in the adult group of blood
donors and 3.0% (9/297) in the group of 13-year-old schoolchildren,
respectively, using a cut-off level of ≥16 U/mL ([Fig. 3a]). TPOab was more frequent in the
group of females (6.8%, 24/351) compared to males (3.6%,
15/421, p=0.0386).
Fig. 3
a) Median AITD autoantibody levels, TPOab and TGab were
comparable between single-positive and double-positive individuals;
TPOab single-positive 96.2 U/mL- double-positive 115.4 U/mL,
TGab single-positive 527.5 IU/mL double-positive 360.5 IU/mL,
p=0.4908 and p=0.5594 respectively. b) Thyroid
autoantibody levels were demonstrated to decrease in the group of
Graves’ disease patients (n=20) from 6 weeks compared to
6 months following diagnosis (TPOab p=0.004 and TGab
p=0.0043).
The prevalence of TGab was 7.6% (36/476) in the adult group of blood
donors and 3.7% (11/295) in the group of 13-year-old schoolchildren,
respectively, using a cut-off level of ≥114 IU/mL ([Fig. 3b]). TGab was more frequent in the
group of females (8.0%, 28/349) compared to males (4.5%,
19/421, p=0.0428).
In the group of adult blood donors, 10.1% (48/476) were TPOab
and/or TGab positive. In the group of TPOab-positive individuals,
60.0% (18/30) were also TGab-positive, and in the group of
TGab-positive individuals, 50.0% (18/36) were also
TPOab-positive. Single positivity was 2.5% (12/476) and
3.8% (18/476) for TPOab and TGab, respectively. TPOab and TGab
double-positivity was 3.8% (18/476).
In the group of adult blood donors, there was no difference in median
autoantibody levels between TPOab and TGab single-positive compared to
double-positive individuals (TPOab single-positive 96.2 U/mL,
double-positive 115.4 U/mL; TGab single-positive 527.5
IU/mL, double-positive 360.5 IU/mL, p=0.4908 and
p=0.5594 respectively, [Fig.
4a]).
Fig. 4 Quantile-Quantile plots showing the distribution of
a) Thyroid peroxidase autoantibodies, with a cut-off of
≥ 16 U/mL. The prevalence of TPOab was
6.3% (30/475) in the group of adult blood donors and
3.0% (9/297) in the group the 13-year-old schoolchildren,
b) Thyroglobulin autoantibodies, with a cut-off of
≥114 IU/mL. The prevalence of TGab was 7.6% (36/476) in
the group of adult blood donors and 3.7% (11/295) in the group
the 13-year-old schoolchildren.
Patients with Graves’ disease (n=20) were longitudinally analyzed
for TPOab and TGab at 6 weeks and 6 months following diagnosis. A general
reduction in thyroid autoantibody levels could be demonstrated between the
sampling times (TPOab p=0.004 and TGab p=0.0043, [Fig. 4b]).
Discussion
Here we report the development and validation of RBA for TPO autoantibodies. The
reported prevalence of TPO and TG autoantibodies in schoolchildren (3.0%,
3.7%) was in agreement with the reported prevalence among schoolchildren
from Finland (2.6%, 3.4%) [27]
and Sweden [28] (TPOab 2.8%). Thyroid
autoantibody prevalence varies substantially between studies. While geographical
differences might occur and potentially explain part of the fluctuation, the
methodological determination of autoantibodies is also speculated to contribute.
There is a lack of standardization for the detection of TPOabs and TGabs. Different
laboratory methods result in a vast span of reference intervals for autoantibody
measurements. The reference interval in the same group of patients has been reported
to be between 0.2 to 10.0 IU/mL for TPOab and 0.53 to 14.23 IU/mL
for TGab, depending on the method used for autoantibody detection [29]
[30].
One explanation may be that the methods differ substantially in how they measure the
presence of autoantibodies; furthermore, methods that use labels for the detection
of autoantibodies may obscure antibody-antigen binding sites. Here, we present a
novel RBA with a higher sensitivity in adults compared to children. The RBA assay
showed a strong correlation to both ECL and RIA, with a few patients classified as
RBA-only negative and all high positive TPO samples identified using RBA. An
expanded analysis in larger cohorts is warranted to further establish the cut-off
level for positivity.
The current study reported a higher prevalence of thyroid autoantibodies in females
compared to males, which is in line with previous studies [4]
[17]
[18]
[19]
[20].
Two different explanatory mechanisms may be considered to understand this disparity
between males and females. First, the time point of seroconversion in the early
stages of autoimmune pathogenesis, and second, the timepoint in the progression from
autoimmunity to clinical onset of disease. Of interest, although the TPOab frequency
is lower among males, a 20-year longitudinal follow-up study reported that the
prognostic value for hypothyroidism, if found positive for anti-thyroid antibodies,
was increased in males compared to females (odds ratio: males 25, females 8) [31].
The progression from seroconversion to clinical onset of AITD is suggested to be
slow, and not all antibody-positive individuals develop disease. Previous studies
reported seroconversion as early as 2 years of age [25]. Our study adds to this understanding with an increased prevalence of
TPOab and TGab from adolescence to adulthood.
Previous studies reported higher median thyroid autoantibody levels among
TPOab/TGab double- compared to single-positive individuals, suggesting a
mechanism of enhanced thyroid antigen leakage [32]. In contrast, antibody levels in our study were comparable between
single and double-positive individuals. Further research is warranted to investigate
the pathogenic mechanisms of AITD, with a focus on clinical outcomes in relation to
individuals with single and double TPOab/TGab positivity.
One strength of the study is that longitudinal studies of thyroid autoantibody levels
in the group of clinically diagnosed patients with Graves’ disease are not
common. To our knowledge, this study is novel in determining autoantibody levels in
the interval between 6 weeks and 6 months following diagnosis. The reduction of
autoantibodies during this period is of particular interest, and further studies are
warranted to determine the clinical and long-term implications.
Weaknesses of the study include the missing determination of HLA-genotypes and
clinical characteristics of adult blood donors to solidify the suggested group
distinctions. Associations to HLA risk alleles in AITD is less defined in patients
and controls compared to many other autoimmune diseases [33]. However, it cannot be excluded that
specific HLA genotypes are related to seroconversion and autoimmune progression to
clinical disease, rather than development of autoantibodies.
The sensitivity for the novel TPOab assay was adequate but could be improved further.
The harmonization of TPOabs between assay platforms is a recognized issue. Two main
explanations were outlined firstly, the selection of reference populations and
secondly, the procedures for antigen preparation. A weakness in our study, and a
potential way to improve the assay, was that the control-groups were not selected
according to the standardized NACB criteria with the inclusion of individuals under
the age of 30 years, only males, and exclusion of subjects with a family history of
thyroid disease [34]. An international
standardization program, as present for T1D [35], would be needed, as it is currently a wide dispersion in TPOab
between assays and techniques. Other ways of improving the assays, including ours,
could entail analyses in expanded and multi-diverse populations, as well as in-depth
analysis of epitope recognition and assessment of autoantibody affinity.
Finally, we report an increased prevalence of thyroid autoantibodies from adolescence
to adulthood, suggesting the importance of future studies aiming to determine the
age of seroconversion.
Data Availability Statement
Data Availability Statement
The data that support the findings of this study are available from the corresponding
author upon reasonable request.
