Thyrotropin Receptor Autoantibody Assessment in Thyroid Eye Disease: Does the Assay Type Matter?
Article information
Abstract
Purpose
Thyroid receptor antibodies can quantify thyroid eye disease activity, predict outcomes and aid timing of interventions. The type and generation of assay is frequently unspecified, complicating meta-analyses. To determine the clinical and biochemical relationships between a second-generation thyrotropin receptor-binding inhibition antibody (TRAb) immunoassay, detecting stimulatory and blocking antibodies, with the thyroid stimulating immunoglobulin (TSI) bridging immunoassay detecting the stimulatory component only.
Methods
Retrospective review of 100 consecutive patients attending a regional specialist service. For each patient and visit, both a TRAb and TSI were performed, and a clinical activity score (CAS) recorded.
Results
A significant positive correlation between TRAb and TSI (rho = 0.828, p < 0.01) but a weaker correlation between the assays and CAS (TRAb: rho = 0.439, p < 0.01; TSI: r = 0.357, p < 0.01) were found. In 10% of the episodic data, patients had a TRAb level that was disproportionately high (39.41 ± 52.84 IU/L), compared to their TSI levels (9.53 ± 12.10 IU/L) with a higher-than-average CAS (2.47 ± 1.78; range, 0–5). Within 12 months of diagnosis, a significant positive correlation between CAS and TRAb (rho = 0.503, p < 0.01) as well as between CAS and TSI (rho = 0.329, p < 0.01) were found. In patients with a diagnosis over 12 months, the correlation with CAS for both TSI and TRAb were Spearman rank correlation coefficient of 0.347 (p < 0.01) and 0.327 (p < 0.01), respectively.
Conclusions
TRAb and TSI correlate strongly and to a lesser extent with the CAS. For most patients, TRAb can be replaced with the more economical TSI. TRAb also correlates better with newly diagnosed, more active patients than TSI. In a subset of patients, blocking antibodies may play a significant pathogenic role, requiring different treatment and monitoring. Further studies are required to investigate this relationship.
Thyroid eye disease (TED) can manifest with several signs including proptosis, exposure keratopathy and compressive optic neuropathy [1]. Although the pathogenesis of TED remains incompletely understood, orbital fibroblasts play a key role [2]. In TED patients, these cells express a higher level of thyrotropin receptor (TSHR) and insulin-like growth factor-1 receptor than ordinary fibroblasts [3]. Circulating thyroid receptor autoantibodies, detectable in TED patients [4–6], act through these receptors and correlate with clinical activity [7,8] which can be predictive [9–11] and indicative [12] of the disease course over time. This antibody-mediated signaling via such receptors leads to fibroblast proliferation and differentiation into myofibroblasts and adipocytes [13–15]. This cascade stimulates excess glycosaminoglycan production, cytokine, and reactive oxygen species release [16] through interaction with T cells resulting in tissue edema orbital expansion seen in TED [17–19].
Disease activity and thyroid antibody levels fluctuate depending on the natural history of the disease; a pattern known as Rundle’s curve. Activity increases, reaching a maximal point, then abates and plateaus, improving, but not returning to baseline [20,21]. On average, the active phase lasts for 12 months in nonsmokers, with TSHR-binding inhibition antibody (TRAb) levels normalizing with a mean of 18.5 ± 6.5 months, and 2 to 3 years in smokers. In untreated patients, activity peaks between 13 to 24 months. Patient characteristics influence Rundle’s curve. In patients who smoke, normalization of TRAb is delayed by almost a year whilst those patients on immunosuppressive treatment, normalize quicker [12,22].
Although thyroid stimulating immunoglobulins (TSIs) are thought to be the main immunopathogenic cause of Graves disease [23], it is unclear whether the immunopathogenic mechanism in TED is due to inhibitory and stimulating or stimulating antibodies alone [16,24]. The literature is inconsistent with the reporting of antibody nomenclature and the type and generation of assay used, frustrating comparisons and formal meta-analyses [25].
There are currently three assay types of biochemical tests to quantify antibody levels: competition assays, bioassays, and assays applying bridge technology. Competition assays, also named TSH-binding inhibition immunoglobulin (TBII) assays, are the most widely used with three generations of assays developed [26].
First-generation TBII assays used autoantibody ability to prevent binding of radiolabeled TSH to porcine thyroid extract. These assays were improved using recombinant TSHR but eventually replaced with second-generation assays which utilize immobilized TSHR on solid phase and have f luorescent rather than radioactive detection. The third-generation assay replaces the labeled TSH with a labeled human monoclonal thyroid stimulating antibody M22 which competes with TRAb for binding to immobilized TSHR.
The TSHR antibody bioassay is a cell-based assay (e.g., Thyretain bioassay) that assesses the function of the thyroid stimulating antibodies directly by measuring the induction of cyclic adenosine monophosphate mediated by binding of TSI to the TSHR. Although this may discriminate between stimulating and blocking antibodies, it is costly and time-consuming, due to long turnaround times and higher labor intensity as a consequence of maintaining cell lines in continuous tissue culture [27].
In 2016, Siemens introduced a new automated bridge assay for the detection of stimulating TSHR autoantibodies [9]. The assay makes use of a chimeric receptor that contains epitopes of the TSHR that binds stimulating immunoglobulins. The bridge assay is fully automated, with fast turnaround times, with minimal labor required, making it more economical and thus introduced in our laboratory to replace the more costly, laborious Brahms second-generation luminescent assay [27].
The TBII and bioassays are significantly positively correlated with clinical activity score (CAS), but TSI may offer advantages in assessing muscle restriction and inflammation [7,28]. Eckstein et al. [9], showed that TBII values <2.3 offer a more than 15-fold better chance for a good course of TED whilst results above >8.7 result in a 31 times higher risk of a severe course.
After the introduction of the new TSI bridging immunoassay, we sought to compare the results from this new assay with a second-generation TRAb assay. As far as we are aware, our study is the first which assesses the clinical activity and biochemical relationship between a next generation TSI assay employing bridging technology with a second-generation TBII competitive assay (Brahms Kryptor TRAK assay, Thermofisher Scientific). We evaluated the clinical activity by measuring CAS, in our population of TED patients helping us to evaluate effectiveness in a clinical setting.
Materials and Methods
Ethical statements
As this study is an anonymized retrospective chart review, neither Institutional Review Board authorization nor the informed consents from patients was necessary. The study adheres to the principles outlined in the Declaration of Helsinki.
Patient identification
We retrospectively reviewed all patients diagnosed with TED, who attended our regional specialist Thyroid Eye Disease Service at Addenbrookes Hospital, Cambridge University Hospitals NHS Foundation Trust between October 2014 and May 2018. Inclusion criteria consist of all patients with a diagnosis of both Graves disease and TED. All age groups, ethnicities, and both sexes were included. Both active and inactive patients were included. Exclusion criteria consist of patients where the diagnosis was unclear, or the presence of any other diagnosis, such as Hashimoto thyroiditis or immunoglobulin G4 orbitopathy, were excluded. Patients where concurrent TRAb and TSI readings were unavailable were excluded. Medical case notes and blood results were accessed through an electronic patient record system database (Epic, Epic Systems Corp). All patients were seen by the same TED specialist and for this 4-year time period had combined assessment of TRAb (Brahms Kryptor TRAK assay) to measure both stimulatory and inhibitory immunoglobulins as well as measurement of TSI (Immulite 2000, Siemens) at their clinic visits. All patients were under multidisciplinary care with an endocrinologist. New patients were rendered biochemically euthyroid, within 6 weeks of the first clinic visit to the ophthalmologist, on a block and replace regimen. This consisted of carbimazole and thyroxine replacement, as is the standard method for our institution. All new patients had an magnetic resonance imaging (MRI) scan on first visit and patients with sight threatening disease were inducted on prednisone and ciclosporin after their second visit following control of triggers and drives including smoking cessation and infectious drives. Follow-up patients were biochemically euthyroid, with a minority post-radioiodine or post-thyroidectomy and therefore on thyroxine replacement alone. The patient cohort captured all patients with TED seen in the specialist service and included patients with inactive and active disease during the review period.
Clinical thyroid eye disease evaluation
All patients had full subjective and objective records for each visit to calculate CAS and VISA (vision, inflammation, strabismus, and appearance) score, Hess charts, fields of binocular single vision, Humphrey visual field, orthoptic assessment, exophthalmometry, and full ocular examination. MRI scans of the head and orbit were performed for all new patients and in those on immunosuppression, prior to cessation or referral for rehabilitative surgery.
TED activity was assessed using the CAS as outlined by Mourits et al. [29] In brief, the score allocates a single point to each of the following: pain (painful feeling on or behind the globe, during the last 4 weeks, pain on attempted upward or downward gaze during the last 4 weeks), redness (redness of the eyelids, diffuse redness of the conjunctiva, covering at least one quadrant), swelling (swelling of the eyelids, chemosis, swollen caruncle, increase of proptosis ≥2 mm during a period of 1–3 months), impaired function (decrease of eye movements in any direction ≥5° during a period of 1–3 months, decrease in visual acuity ≥1 line on Snellen chart [using a pinhole] during a period of 1–3 months) [29]. CAS was chosen to allow comparison with other publications and calculated using the 10-point score scale from the signs and symptoms recorded in the electronic patient record system database (Epic) for each patient visit. Lid retraction, proptosis, and strabismus all occur with chronic inactive disease as well as active disease, so isolated serial measurements, outside the CAS score, were not utilized to assess disease activity in this study; they were, however, recorded for each clinic visit. Correlation with MRI imaging was beyond the scope of this paper but has been published by our team previously [30].
Thyroid stimulating hormone receptor autoantibody assays
The Brahms Kryptor TRAK assay is a competitive two-step, second-generation, homogenous assay which is based on inhibition of binding of labeled bovine TSH to a testing vial coated with human recombinant TSHR (hTSHR). This TRAb assay does not distinguish between TSIs and thyroid blocking immunoglobulins and has a clinical decision limit of 0.9 IU/L. Results above 0.9 IU/L may be suggestive of a diagnosis of Graves disease when taken into consideration with other findings.
The Immulite 2000 TSI assay is a two-step, chemiluminescent immunoassay employing bridging technology. It utilizes a pair of recombinant hTSHR constructs in a bridging immunoassay format. The capture receptor is immobilized on a solid phase (polystyrene bead) and the signal receptor is an alkaline phosphatase labeled recombinant hTSHR in a buffered solution. In the first step, sample is incubated with the solid phase for 30 minutes allowing the TSI in the sample to bind through one arm to the capture receptor. This is followed by a wash step to remove residual sample. In the second step, the signal receptor is added to the reaction tube and incubated for 30 minutes. The complexed TSI binds the signal receptor through the second arm, forming a bridge. Unbound signal receptor is then removed by centrifugation and washes. Finally, chemiluminescent substrate is added to the reaction tube and a signal is generated in direct relation to the amount of TSI in the sample. The clinical cutoff value for positive results is 0.55 IU/L; results above this may be suggestive of a diagnosis of Graves disease when taken into consideration with other findings. Both antibody assays were performed at the same clinic visit as their clinical assessment or within two weeks of their assessment.
Statistical analysis
IBM SPSS ver. 24.0 (IBM Corp) was used for data analysis which included Spearman correlation coefficient to compare TRAb and TSI to CAS scores, and to each other. Linear regression analysis was also performed using the program and data were presented as mean ± standard deviation, unless otherwise indicated.
Results
Patients included 75 females and 25 males, ranging in age from 13 to 83 years with a mean of 52.5 years. Amongst this cohort, 163 concurrent TRAb and TSI assays were performed and evaluated for correlations. The average number of tests per patient was 1.629. The average first presentation TSI result was 8.12 IU/L, and the average follow-up TSI result was 6.54 IU/L. The average first presentation TRAb result was 20.23 IU/L, and the average follow-up TRAb result was 15.11 IU/L. The majority of patients (94 out of 100) were White, with three South Asian patients, two East Asian patients, and one Black patient. Eleven had undergone thyroidectomy. The mean TED duration was 20.47 months with 19% of patients considered to have active disease and 32% of all patients on immunosuppression.
In patients where concurrent TRAb and TSI levels were taken, the average (mean ± standard deviation) TSI level was 7.33 ± 11.59 IU/L with the average TRAb level 17.67± 43.24 IU/L. In 123 episodes, TSI and TRAb levels were compared to concurrent CAS. The average TRAb level was 13.54 ± 29.33 IU/L (range, 0.1–236.2 IU/L) and the average associated CAS was 1.23 ± 1.58 (range, 0–7). Where TSI levels were compared to the CAS, the average TSI level was 6.76 ± 11.1 IU/L (range, 0.1–40.0 IU/L) and the average associated CAS was 1.25 ± 1.59 (range, 0–7). In 10% of the episodic data, amounting to 19 episodes the patients had a TRAb level that was disproportionately high relative to their TSI levels. In this group, the average TRAb value was 39.41 ± 52.84 IU/L with the average TSI being 9.53 ± 12.10 IU/L. The average CAS in this group was also higher at 2.47 ± 1.78 (range, 0–5).
When evaluating patients who had a diagnosis of TED for less than 12 months, 49 concurrent episodes of TRAb and TSI were evaluated with CAS. The average TRAb level was 20.00 ± 30.00 IU/L (range, 0.3–105.6 IU/L) and the average associated CAS was 2.12 ± 1.73 (range, 0–7). The average TSI level was 9.11 ± 13.15 IU/L (range, 0.17–40.00 IU/L) and the average associated CAS was 2.12 ± 1.73 (range, 0–7).
In those patients who had a diagnosis of TED of greater than or equal to 12 months, 74 concurrent episodes of TRAb and TSI were evaluated with CAS. The average TRAb level was 6.64 ± 10.00 IU/L (range, 0.1–236.2 IU/L), and the average associated CAS was 0.68 ± 1.18 (range, 0–6). The average TSI level was 5.21 ± 9.28 IU/L (range, 0.1–40.0 IU/L), and the average associated CAS was 0.68 ± 1.18 (range, 0–6).
TRAb and TSI levels were compared with each other in 163 concurrent episodes (Fig. 1). There was a strong positive correlation between TRAb and TSI levels with a Spearman rank correlation coefficient of 0.828 (p < 0.01), although most of the patients assessed during the cohort of review had TRAb and TSI levels below 15. When TRAb levels and CAS results were compared with each other, there was a positive correlation between TRAb levels and CAS with a Spearman rank correlation coefficient of 0.439 (p < 0.01) (Fig. 2A). When TSI levels and CAS were correlated with each other, there was a much weaker positive correlation than for TRAb levels, with a Spearman rank correlation coefficient of 0.357 (p < 0.01) (Fig. 2A).
A subgroup of patients was evaluated. Those who had a diagnosis of TED of less than 12 months and those with a diagnosis of TED for 12 months or more. In those patients who had a diagnosis of TED of under 12 months (representing the active phase), there was a strong positive correlation between TRAb and CAS (rho = 0.503, p < 0.01) but a weaker positive correlation between TSI and CAS (rho = 0.329, p < 0.01) (Fig. 2B).
A weaker positive correlation was found in those patients who had diagnosis of TED of 12 months or more (representing the inactive phase). In the 74 episodes, a Spearman rank correlation coefficient of 0.347 (p < 0.01) for TRAb and CAS and 0.327 (p < 0.01) for TSI and CAS were identified (Fig. 2C). For the purposes of graphical demonstration, all values >40 for TRAb have been standardized to 40, the maximal value for TSI, in all charts.
Analysis of TRAb levels and TSI levels using a linear regression model found TSI values to be predictive of TRAb levels (r2 = 0.517, p < 0.01). TRAb levels were shown to be weakly predictive of CAS score (r2 = 0.100, p < 0.01) whilst TSI levels were also shown to be predictive, though to a lesser degree than TRAb levels (r2 = 0.075, p < 0.01).
Discussion
The use of thyroid receptor antibodies in the diagnosis and management of TED is now well established [1,12,31]. Nevertheless, the reporting of antibodies and TED activity is inconsistent in the literature with regards to the nomenclature and the type and generation of assay used, making direct comparisons difficult [25,31]. The manual nature of the second-generation TRAb assay (detecting a combination of TSI and thyroid blocking immunoglobulin), has led it to being phased out at some units citing operational and financial unsustainability. We wished to determine if clinical information regarding TED would be lost in abandoning the second-generation TRAb assay in favor of the next generation TSI assay employing bridging technology. We evaluated this by measuring both antibodies concurrently in patients and correlating this to CAS over approximately 4 years.
The results of this study show a strong positive correlation between both biomarkers: TRAb and TSI (rho = 0.828, p < 0.01), although most of the patients assessed during the cohort of review had TRAb and TSI levels below 15. However, this strong correlation continues to be displayed with higher TRAb and TSI levels. In addition, linear regression analysis supported these results, with TSI values being predictive of TRAb levels (r2 = 0.517, p < 0.01). The reason for this strong correlation is likely to be due to the majority of patients having predominantly stimulatory antibodies. It therefore equates that the TBII assay, which measures both stimulatory and inhibitory antibodies, produces a result akin to that of the TSI assay in most patients.
Though TRAb and TSI values generally correlated well, notably a small subset, approximately 10% of patients, had TRAb levels that were disproportionately high, compared to their TSI levels (39.41 ± 52.84 IU/L vs. 9.53 ± 12.10 IU/L) with a higher clinical activity as measured by CAS (2.47 ± 1.78). The reason for this discrepancy is not fully understood but may indicate that the more inflammatory TED phenotypes (with higher CAS) may be driven by both stimulatory and inhibitory antibodies. A study by Jang et al. [32], found when TSI (Mc4-TSI bioassay) and TRAb levels (third-generation TBII) were at or above median levels, this was associated with a higher CAS. This was also observed in our study. However, Jang et al. [32] also identified a significantly higher CAS in those patients who had a TSI level higher than or at the median but a lower TRAb than the median. We did not observe this in our study. Jang et al. [32] felt this was due to TSI being better at identifying active soft tissue inflammation and muscle restriction in TED patients and thus better at reflecting ocular manifestations of Graves orbitopathy (GO). Whilst the study certainly has its merits, it may not be directly applicable to our study. In the first instance, the study was performed on a different ethnic group than our patients who are predominantly white. In the second instance, the assays used in the study were a second-generation TRAb and Mc4-TSI bioassay, whilst our study utilized a second-generation TRAb and a next generation TSI bridging assay which may help explain some of the differences in our findings.
We and others have previously discussed the limitations of CAS as an objective measure of disease activity: being binary, subjective, lacking sensitivity to inflammation, and being slow to reflect change. However, until more sensitive quantitative assessments of activity are widely available, CAS remains the main outcome measure used in clinical studies [30]. We used CAS to be consistent and allow comparison with other similar studies.
The correlation between TRAb and CAS (rho = 0.439, p < 0.01), and TSI with CAS (rho = 0.363, p < 0.01) was weaker than that of TRAb and TSI. The lack of a stronger correlation may be due to the inherent limitations of the CAS. Our patient population studied had a range of clinical activity, but the majority were well-controlled with mild disease and low activity score, creating a smaller spread of results by which to detect correlations. This proved to be the case in a similar study [33] where the patients selected for study had chronic inactive disease; no correlation between either biomarker or the CAS was found. Our study benefited from having a spectrum of activity, however, with some patients having high activity scores. In fact, the average duration of TED since the onset of symptoms was 20.48 months compared to 46 months in the other study. This difference likely explains why our study was able to detect a correlation. Likewise, Jang et al. [7], reported antibody and CAS results from a cohort where 14.2% of patients had not achieved endocrine control and had active disease and found a significant correlation.
Our results suggest that the TSI could be cost-efficient and replace TRAb at our center, but with the caveat that clinically valuable information might be lost. We were interested to find a slightly stronger correlation between TRAb levels and CAS (rho = 0.439, p < 0.01), compared to TSI and CAS (rho = 0.363, p < 0.01). Conversely, Jang et al. [7] found that Mc4-TSI bioassay, using genetically engineered Chinese hamster ovary cells, tends to have a stronger positive correlation with CAS compared to inhibitory immunoglobulins (first and third-generation TBII assay). It is questionable as to how comparable the results are to our study, as Jang et al. [7] used a TSI bioassay, whilst our study utilized an automated TSI bridged assay. Nevertheless, the reason for the differences may signify different contributions of these antibodies at different stages of the disease or in different ethnic backgrounds. Jang et al. [32]’s patients were predominantly Asian, and TSI has been reported to be less prevalent in White patients [34].
This difference is more marked when correlations are adjusted to the duration of GO. For the purposes of assessing if a correlation was more apparent in the early active phase of disease compared to late, patients were divided into two categories: those with a diagnosis of TED for 12 months or less and those with a diagnosis of TED greater than 12 months. The two time periods represent different stages on Rundle’s curve. Based on previous studies, as mentioned earlier, on average the active phase lasts for 12 months in nonsmokers and 2 to 3 years in smokers. In treated nonsmoking patients, Roos et al. [12] have shown the TRAb to normalize at 18.5 ± 6.5 months, whilst Menconi et al. [22] found, in untreated patients, activity tends to peak between 13 and 24 months. The majority of patients in this series were euthyroid at the point of diagnosis and on treatment. Therefore, a 12-month time point was chosen, to represent the transition between activity and inactivity in this patient cohort.
When evaluating patients during the active phase (under 12 months), TRAb has a stronger correlation to the CAS than TSI (rho = 0.503 and rho = 0.329, respectively; p < 0.01), whilst TRAb and TSI show a similarly weak positive correlation (rho = 0.347 and rho = 0.327, respectively; p < 0.01)in those with longstanding disease, i.e., 12 months or longer. The CAS in the former group (2.12) was also much higher, suggesting more active disease, than the latter (0.68), alluding to inactivity. This observation has been noted in a previous study [9], where higher second-generation TRAb titers (TBII) were seen in patients with more active disease. The reasons behind this may lie in stimulatory and blocking antibodies proving key to the more inflammatory changes in TED. Thyroid blocking antibodies are integral to the pathogenesis of many autoimmune inf lammatory conditions such as autoimmune thyroiditis [35]. In our study, we have also observed a weaker correlation for both TSI and TRAb, in patients with less active longstanding disease. It has been suggested that a decline in both TSHR expression and TSI affinity for them results in a weaker correlation between both TRAb and TSI with GO over time [36–38].
In contrast, other studies have shown TBII to have no correlation with GO [9,39]. This may be due to several reasons including the use of different assays. Different studies have utilized varying generations of assays with differing results. The studies by Noh et al. [39] and Eckstein et al. [40], which showed no correlation between TRAb and GO, utilized a first-generation assay. In addition, first- and third-generation assays are highly dependent on whether patients are pretreatement or posttreatment [36]. As alluded earlier, ethnic differences may also explain this discrepancy. Our patient population consisted primarily of White patients. The review by Chng et al. [41] identifies several studies in White populations where both TRAb and TSI have been associated with GO, whilst in Asian populations it is only TSI that has been associated. The reason for this remains unexplained, but genetic, environmental, and anatomical factors may be involved.
This study was borne out of a predetermined financially driven decision, to replace the more expensive batch processed second-generation TRAb, with the more economical, automated next generation TSI in the biochemistry department at our institution. During the transition period, an opportunity arose to assess and evaluate the two types of assays. Our study has demonstrated that for the majority of patients with TED, the second-generation competitive TRAb assay (stimulatory and inhibitory antibodies) and next generation TSI assay employing bridging technology (stimulatory antibodies only) correlate well. The second-generation TRAb assay is more costly and requires slower batch processing unlike the TSI assay which is automated. Therefore, for the majority of patients, the TRAb assay can be replaced by TSI to increase speed and reduce cost. However, removing the TRAb assay, and relying on the TSI alone, could result in the loss of clinically important information that may help us better understand the underlying immunopathogenesis of TED, particularly in the active inf lammatory phase, and monitor those patients with greater disease activity. To better study and unravel the immune pathogenesis of TED and the different phenotypes, TRAb remains an essential test. Further investigations are required with better objective assessment of activity in patients in whom a disparity in antibody levels can shed further light on TED pathogenesis.
Acknowledgements
The authors would like to thank the Department of Biochemistry, Cambridge University Hospitals NHS Foundation Trust (Cambridge, UK) for their contributions in processing and providing technical expertise with the assays described in this article.
Notes
Conflicts of Interest: Blood Sciences Laboratory at Cambridge University Hospitals NHS Foundation Trust (Cambridge, UK) collaborated with Siemens to evaluate the thyroid stimulating immunoglobulin assay prior to its introduction in 2017. No other potential conflicts of interest relevant to this article were reported.
Funding: None.