Comparison of Serum Thyroglobulin Levels in Differentiated Thyroid Cancer Patients Using In-House Developed Radioimmunoassay and Immunoradiometric Procedures

Abstract

Thyroglobulin (Tg) is a proven tumor marker in the follow-up and post-operative management of patients with differentiated thyroid cancer (DTC). All assays for serum thyroglobulin (s-Tg) are based on immunoassays, however, the assay technique has a bearing on the variations seen in the estimations. We studied this using four in-house developed radioimmunoassays (RIA) and immunoradiometric assays (IRMA). Limit of detection, working range, recovery, dilution test, precision profiles and method comparison were evaluated. All four methods were used for the estimation of s-Tg in DTC patients and also compared for their performance using commercially available Tg IRMA kits from DiaSorin and Izotop. The s-Tg values measured by six different immunoassays showed very significant inter-method correlation (0.84–0.99, p < 0.001). However, among the in-house developed assays; the coated tube IRMA showed a better sensitivity and precision at the lower concentration range and hence, is preferable for the routine measurement of s-Tg in patients negative for Tg autoantibodies (TgAb). Although the second generation IRMAs offer practical benefits of having higher sensitivity, shorter turn-around time and convenience of automation, they, unfortunately, also have higher tendency for interference from both TgAb and heterophilic antibodies, if present in the sample. On the contrary, RIA is less prone to such interference and, hence, can be used in patients with TgAb. In order to effectively use this test, it is important that nuclear medicine physicians and endocrinologists understand these intrinsic technical limitations encountered during s-Tg measurement.

Introduction

Measurement of thyroglobulin (Tg) serves as a tumor marker for patients with a diagnosis of differentiated thyroid cancer (DTC) for the detection of disease persistence or recurrence after initial treatment (thyroidectomy and/or radioiodine therapy). Patients with DTC usually have an excellent prognosis, following surgical and radioiodine treatment to remove the cancer cells, suppressive doses of levothyroxine, and long-term follow-up, including measurement of serum thyroglobulin (Tg) using a sensitive assay to detect recurrence [1]. Over the last four decades, Tg has been measured essentially by immunoassay methods: radioimmunoassay (RIA), immunometric assay (IMA) [2]. Liquid chromatography–tandem mass spectrometry (LC–MS/MS), as an absolute, free from interference from autoantibodies against Tg (TgAb) is being explored but suffers from poor sensitivity, extensive sample preparation, longer turn-around time (TAT) and expensive instrumentation [2]. Use of multi-analyte immunoassays for thyroid related hormones including s-Tg has been recently reported, but is in a conceptual stage [3–5]. Regardless of several advances in immunoassays, s-Tg measurement still faces problems regarding its clinical interpretation in various contexts. The three main technical limitations of current Tg assays are (1) suboptimal functional sensitivity (FS) and precision, (2) between-method discordance and specificity differences and (3) interferences from autoantibodies against Tg and heterophilic antibodies, mainly human anti-mouse antibodies (HAMA). Despite universal acceptance of Certified Reference Material (CRM-457) from the Community Bureau of Reference (BCR) in Brussels, Belgium, for standardization, inter-laboratory variability between Tg immunoassays still exists. This could be ascribed to poor standardization, heterogeneity of circulating Tg and differences in epitope recognition by the antibodies used in different assays that detect Tg isoforms with variable avidity. These implies that a change in Tg detection method can cause unexplained variations in the serial Tg measurements, which is crucial in the long-term monitoring of patients with DTC. Therefore, interchangeable use of different Tg assays for follow-up of patients with DTC is not recommended. Interference caused by TgAbs in s-Tg measurements is the most serious problem since they are detected in approximately 25% of patients compared to HAMA interference (1.5–3%). TgAbs cause underestimation of s-Tg in IMA class of Tg methods, thereby, masking the presence of the disease, whereas, HAMA typically causes overestimation. On the contrary, RIA class of Tg methods are still in use as they are less affected by TgAbs although some interfering TgAbs may result in falsely high or low s-Tg value. The latest LC–MS/MS class of immunoassays state to be free from TgAb and HAMA interferences as Tg-TgAb complexes are trypsin digested to release conserved Tg pepetide(s) for LC–MS/MS analysis. Nonetheless, suboptimal sensitivity, longer TAT, high instrumentation cost, limited availability are some of the practical problems which confines the clinical utility of the LC–MS/MS assays [2]. Among the IMA-class tests, the isotopic Immunoradiometric assays (IRMA) are favoured over RIA because of higher sensitivity potential, shorter TAT, wider working range, longer shelf-life of the labeled antibody reagent and the convenience of automation. Since, Tg IRMA assay results are compromised by the presence of mainly TgAbs, several approaches are suggested to overcome these, including (1) a recovery test, (2) adoption of RIA assay in TgAb positive patients, and (3) serial measurement of TgAb during the follow-up [6].

At our Centre, till date, more than 12,000 patients have been treated with ¹³¹I for thyroid cancer and approximately 5000 samples are analyzed annually for s-Tg on these patients, who are on regular follow-up. Hence, our aim was to develop a sensitive, specific, reliable, cost-effective and robust isotopic immunoassay for the measurement of s-Tg, and compare it with well-established and commercially available human Tg IRMA kits (DiaSorin and Izotop). In the context of clinical decision making, we wanted to maintain consistency in our results by using the in-house developed Tg assays. The TgAb levels of all patients were negative, as per the reference range of the institute [7].

Materials and Methods

Reagents

All chemicals and reagents were of analytical grade and purchased locally. Polystyrene ‘star’ tubes were procured from M/s Tarsons India Ltd. Calcutta. Carrier free Na¹²⁵I was supplied by Radiopharmaceuticals Division, BARC. Low density magnetisable cellulose iron-oxide particles were obtained from Scipac, Broad Oak, Sitingbourne, UK (Product Code M-104). Anti-hTg IRMA kit from Immunotech, France was used for measuring TgAb in accord with the established normal range of our Centre (30 mIU/L) in order to exclude TgAb positive samples. TgAb assay was referenced to the WHO TgAb First International Reference Preparation (WHO 65/93).

Instrumentation and data analysis

1. Packard RIASTAR multi well gamma counter, USA for counting and plotting dose response curves.

2. RIA-data analysis was performed using WHO-Program for RIA, by Prof. Ray Edwards, Dept of Molecular Endocrinology, Middlesex Hosp Med School, London, UK.

Methods

Following in-house assays for s-Tg were standardized, validated and compared

GARS-S. aureus RIA

Goat anti-rabbit serum-S. aureus combination (GARS-S. aureus) was used for the separation of bound and free ¹²⁵I-human Tg [8].

GAR-Ig-MP RIA

Modification in the above RIA method was done using goat anti-rabbit-Ig antiserum coupled to magnetic particle (GAR-Ig-MP) for the separation of bound and free ¹²⁵I-human Tg.

Protocol for GAR-Ig-MP RIA

After the primary incubation of 72 h, as per the protocol mentioned by Kumar et al. [8], 50 μL of appropriately diluted GAR-Ig-MP suspension was added and further incubated for 2 h. After the incubation, PBS-EDTA buffer (500 μL) was added to the tubes and the test-tube racks were placed on magnetic rack. After 15 min, the contents of the tubes were decanted, dried and the pellet settled at the bottom was counted for bound radioactivity.

MP-IRMA

A two-step, solid-phase magnetic particle IRMA (MP-IRMA) was developed using partially purified rabbit anti-Tg antibodies coupled to magnetisable cellulose iron-oxide particles.

Protocol for MP-IRMA

In plain polystyrene tubes, Tg calibrators, quality control samples, test samples (100 μL) and anti-Tg-MP suspension (50 μL) was added and the final volume was made to 250 μL with PBS-EDTA buffer. The tubes were incubated on orbital shaker for 3 h. After the incubation, 2 mL of wash buffer (PBS-EDTA buffer containing 0.05% Tween 20), was added to the tubes and placed on magnetic rack for 15 min. Supernatant was decanted and this step was repeated twice. Further 200 μL of ¹²⁵I-anti-human Tg monoclonal mouse antibody (MAb) with an activity of ~ 3000 Bq was added to each tube and incubated for 3 h. Washing steps were repeated, tubes were blotted and pellet was counted for radioactivity.

CT-IRMA

A two-step, solid-phase coated tube IRMA (CT-IRMA) was developed using partially purified rabbit polyclonal anti-Tg antibodies coated to ‘star’ shape polystyrene tubes [9].

For all the above Tg methods, percent radioactivity bound (%B/T) was plotted against corresponding standard Tg concentrations and Tg values for unknown samples were deduced from it. Approval from BARC Animal Ethics Committee was taken for the production of antisera. Validity of the in-house Tg assays was established by studying various parameters like analytical and functional sensitivity, precision, working range, hook effect, recovery, dilution test, shelf-life of radiolabeled tracers, and stability of anti-Tg MP suspension and anti-Tg CT. Patients with an established diagnosis of DTC were analyzed for the presence of s-Tg using these in-house developed assays. Commercial human Tg IRMA kits viz, DiaSorin from Italy and Izotop from Hungary were used for validation and comparison of the in-house developed s-Tg RIA and IRMA assays. All the commercial assays were performed according to the manufacturer’s instructions.

Patient Samples

The serum samples were collected as a part of routine sampling process in follow-up after treatment of DTC at our Centre. Patient consent was taken.

Statistical Analysis

We used Pearson’s coefficient of correlation and regression analysis for the evaluation of different Tg assays. p value of 0.05 was considered statistically significant.

Results

Salient features of all the in-house developed s-Tg assays are summarized in Table 1 and the standard curves along with their respective precision profile are depicted in Fig. 1a–d.

Table 1.

Salient features of the in-house developed RIA and IRMA assays for serum thyroglobulin estimation

Assay parameters	Assay
	GARS-S. aureus RIA	GAR-Ig-MP RIA	MP-IRMA	CT-IRMA
Sensitivity (ng/mL)
a. Analytical	4.5	4.0	4.0	0.2
b. Functional	6.0	4.5	5.0	1.0
Precision (%CV)
a. Intra-assay	7.75	7.3	13.5	6.95
b. Inter-assay	8.15	8.8	14.6	8.85
Range (ng/mL)	4–800	3–800	5–800	1–800
Tracer stability (weeks)	3	3	8	8
Incubation (h)	90	78	6	18
Recovery (%)	85–110	85–110	85–110	85–110

Fig. 1.

Characteristic standard curve of serum thyroglobulin along with the precision profile obtained by in-house developed assay systems. a GARS-S. aureus RIA, b GAR-Ig-MP RIA, c MP-IRMA and d CT-IRMA

Analytical Sensitivity

The analytical sensitivity for RIA and IRMA assays were calculated at Bo binding with 99% confidence limits. The sensitivity for GARS-S. aureus and GAR-Ig-MP RIAs (Bo-3SD) were 4.5 and 4 ng/mL respectively, whereas for MP-IRMA and CT-IRMA (Bo + 3SD) they were 4 and 0.2 ng/mL respectively.

Functional Sensitivity (FS)

FS was determined on the between-run precision of 6–12 months, which was 6 and 4.5 ng/mL for GARS-S. aureus and GAR-Ig-MP RIA respectively. Whereas for MP-IRMA and CT-IRMA it was 5.0 and 1.0 ng/mL respectively.

Method Precision

The with-in run coefficient of variation (CV) for GARS-S. aureus RIA was 7.75%, for GAR-Ig-MP RIA it was 7.3%, for MP-IRMA it was 13.5% and for CT-IRMA it was 6.95%. The between-run CVs for GARS-S. aureus RIA, GAR-Ig-MP RIA, MP-IRMA and CT-IRMA were 8.15, 8.8, 14.6, 8.85% respectively.

Working Range

Working range of each assay was determined from the precision profile, by plotting %CV against standard Tg concentrations (Fig. 1a–d). With 20% CV as the acceptable assay precision, the working ranges for RIAs and IRMAs were found out (Table 1).

Hook Effect

MP-IRMA and CT-IRMA which has a ‘two-step’ assay design did not show any kind of hook effect up to a concentration of 12,800 ng/mL and thereafter, a plateau was observed.

Recovery

For all the assays analytical recovery varied between 85 and 110% when carried out by adding known amount of Tg (negative for the presence of TgAb).

Dilution Test

On diluting the serum sample with Tg free serum, the values were linearly proportional. The observed concentration to the expected ranged from 82.7 to 105.7%.

Stability and Performance of Anti-Tg MP and Anti-Tg CT

Maximum binding (Bmax) and non-specific binding (NSB) for anti-Tg MP over a period of 24 months ranged between 11.7–22.6% and 0.22–0.56% respectively. Up to 1 year, there was neither a significant fall in %Bmax nor increase in %NSB and hence, the CT were utilizable. Anti-Tg CT, when stored for a period of 12 months and more, showed %Bmax ranging between 22 and 33% with %NSB of 0.28–0.47%. Therefore, even the stored CT were satisfactory for use.

Method Comparison and Validation

TgAb positive samples were excluded from the analysis. The cumulative comparison of the in-house developed assays with the commercialized IRMA kits (DiaSorin, Italy and Izotop, Hungary) showed very significant correlation. The coefficient of correlation (r) and the regression equations are summarized in Table 2.

Table 2.

Comparison of Tg levels in DTC patients (TgAb negative) by different assay systems

Assay system	X-axis
	GARS-S. aureus RIA	GAR-Ig-MP RIA	DiaSorin	Izotop
Y-axis
GAR-Ig-MP RIA	y = 0.99x − 2.5 r = 0.99, n = 69, p < 0.001	–	y = 0.92x − 0.39 r = 0.98, n = 28 p < 0.001	y = 2.95x + 20.46 r = 0.96, n = 96 p < 0.001
MP-IRMA	–	y = 0.92x − 6.7 r = 0.94, n = 157 p < 0.001	y = 1.15x − 3.24 r = 0.96, n = 28 p < 0.001	–
CT-IRMA	–	y = 0.68x + 11.3 r = 0.84, n = 398 p < 0.001	y = 0.96x +2.69 r = 0.93, n = 249 p < 0.001	y = 1.26x − 5.80 r = 0.89, n = 100 p < 0.001

Discussion

Measurement of s-Tg level is one of the most useful investigations in the postoperative management of patients with DTC. Owing to the technical limitations inherent in Tg measurement, different approaches are being made by the researchers to effectively measure s-Tg. Besides the competitive RIA methodology and non-competitive IMA methods, the LC–MS/MS and MAIA detection systems have also been proposed as a reference method, since it is not prone to Tg-Ab interference. However, there are several technical and analytical issues related to these procedures which prevent their application as a routine diagnostic tool in the clinical research laboratory [1–5].

At our Centre, a large number of serum samples are routinely investigated for Tg by RIA. However, RIA methods need very long assay-incubation times for maximal sensitivity for clinical efficacy. Further, it is difficult to balance the clinical need for sensitivity with expediency (short incubations) when developing these methods. Therefore, RIAs for Tg, due to this major practical limitation are being replaced by more sensitive IMA with faster TAT, with suggestions for improving between-assay precision and detecting hook effect [10]. It was therefore, considered essential to improve the conventional RIA technique or develop an IRMA system to overcome these limitations. Hence, the present work was also aimed at standardizing a rapid, sensitive and precise, IRMA for the estimation of s-Tg.

The earlier in-house GARS-S. aureus RIA was modified by substituting the GARS-S. aureus with magnetic particles coupled to GAR-Ig (GAR-Ig-MP). This brought about a significant reduction in the incubation period and facilitated an easy centrifugation-free separation system. The correlation coefficient of 0.99 showed that the values obtained using these two systems were comparable. A good laboratory approach to introducing a new assay system is to run it simultaneously with an established test-system, over several assays to identify and remove shortcomings, if any. Hence, the GAR-Ig-MP RIA was compared with commercially available Diasorin and Izotop IRMA kits and a good correlation was seen. However, with increasing sample loads and the practical limitations of RIA, prompted us to standardize two-step IRMA assays using magnetic particles and polystyrene tubes as solid substrates. Among the two solid-phases, CT-IRMA showed better performance than the MP-IRMA in terms of sensitivity and precision. It was found to have better analytical sensitivity (0.2 ng/mL) and FS (1 ng/mL) which is optimal for clinical use and hence, makes it superior to the other in-house developed assays. Nowadays, FS has more importance than the analytical sensitivity for clinical diagnosis. Indirectly, it defines the clinical utility or clinical targeting of Tg assays.

Further, the IRMA method developed for s-Tg estimation scores over RIA due to the stability of the tracer. The radiolabeled Tg preparations required for RIA have disadvantage of short shelf-life and inherent instability [11]. The shelf-life of ¹²⁵I-MAb used in IRMA assays was longer compared to the ¹²⁵I-Tg used in RIAs.

CT-IRMA is technically simpler compared to RIA and MP-IRMA due to lesser number of pipetting steps. It showed very good correlation on comparison with the two well-accepted commercial kits. Good correlation was also seen with the in-house GARS-S. aureus RIA, but the regression equation indicated significant method induced differences. These systemic differences between the methods were mainly due to different calibrators, heterogeneity of circulating Tg, interference from TgAb and different specificities of the antibodies used in different assays [12, 13]. In order to minimize these differences, CRM 457 is being adopted either directly or indirectly for the calibration of Tg immunoassays by most of the laboratories. Even then, s-Tg levels that are determined by methods using this standard are reported to show four-fold variations [14]. Recovery and dilution tests were performed to ensure that the calibration is accurate and that the matrix used for making standards is appropriate. On diluting the serum sample containing high Tg with hormone free serum and assaying it, the values were found to be linearly proportional. For all the Tg assays, recovery varied between 85 and 110% when carried out by adding known amount of Tg. This reveals that, the matrix used for preparing Tg standards was compatible to the human serum.

Among all the in-house developed RIA and IRMA methods, CT-IRMA has improved clinical value for the detection of recurrent disease and for serial measurements in long-term follow-up of DTC patients, since it is accepted that serial measurements of s-Tg are more informative than an absolute single value and thus, have additional clinical assistance [1].

In the present study, both RIA and IRMA techniques showed an excellent agreement of s-Tg values in samples which were negative for presence of TgAb. As mentioned in the Introduction, we selected samples, which were negative for TgAb to have an unbiased comparison of the assay methods.

RIAs due to their assay design are less susceptible to interference from TgAb and HAMA, as compared to IRMA-type of assays. This is well documented by several workers [1, 2, 10, 15]. LC–MS/MS is another method for Tg-assay that is not affected by both TgAb and HAMA interference, but is presently not suited for large sample throughputs [2]. The effect of TgAb leading to high underestimation of true s-Tg value by the IRMA methods, is the most serious specificity problem affecting s-Tg measurement. Unfortunately, less progress has been made in precisely detecting, quantifying and eliminating the degree of TgAb interference in the measurement of s-Tg.

In conclusion, among the two solid-phases introduced for immobilizing anti-Tg antibodies, CT-IRMA served to be a robust and superior assay compared to MP-IRMA in terms of sensitivity, precision, stability and automation. Hence, the former can be routinely employed for monitoring low basal s-Tg concentrations (arising from normal thyroid remnants), which have recognized prognostic significance in the low-risk DTC patients who are on TSH-suppression and are no longer considered necessary for radioiodine therapy. Further, it is recommended that for judicious and confident clinical representation, followed by therapeutic decision making; TgAb levels should be determined using sensitive immunoassays that complement with the IMA methods for s-Tg. Whereas RIAs, considered as methods that are TgAb-resistant could be reserved for the patients with the presence of TgAb. For optimal postoperative DTC monitoring, different kind of approaches are needed to cater to the DTC patients with positive or negative TgAb status. The former and latter form ~ 25 and ~ 75% of DTC patients respectively.

In addition, for long term follow-up of cancer patients, it is strongly recommended that serial s-Tg measurements (in a patient) be made in the same laboratory using the same assay method. This is because Tg values are not interchangeable among laboratories, even when using the international Tg standard preparation (CRM 457) which can reduce but cannot eliminate biases between various Tg methods. Physician-laboratory communication is a key for maximizing the clinical utility of s-Tg testing and minimizing unnecessary costly imaging studies.

Compliance with Ethical Standards

Conflict of interest

Chandrakala Gholve, J. Kumarasamy, Archana Damle, Savita Kulkarni, Meera Venkatesh, Sharmila Banerjee and M. G. R. Rajan declare that they have no conflict of interest.

Ethical Approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed Consent

Informed consent was obtained from all individual participants included in the study.