Diagnostic Testing for Graves' or Non-Graves' Hyperthyroidism

A Comparison of Two Thyrotropin Receptor Antibody Immunoassays With Thyroid Scintigraphy and Ultrasonography

Lorenzo Scappaticcio; Pierpaolo Trimboli; Franco Keller; Mauro Imperiali; Arnoldo Piccardo; Luca Giovanella


Clin Endocrinol. 2020;92(2):169-178. 

In This Article


Patient Characteristics

About 124 untreated thyrotoxic patients (93 females and 31 males) were finally included in our retrospective analysis, while 18 other patients were excluded because they lacked the TSI measurement. According to the qualitative scintigraphy, Graves' hyperthyroidism occurred in 86 cases while the remaining cases (n = 38) were diagnosed as having non-Graves' hyperthyroidism (painless thyroiditis in 12 patients; AFTN in 10 patients; iodine-induced thyrotoxicosis in seven patients; amiodarone-induced thyrotoxicosis in four patients; transient subnormal TSH in two patients; toxic multinodular goitre in one patient; iatrogenic thyrotoxicosis due to tyrosine kinase inhibitors in one patient and one case of thyrotoxicosis with unknown origin). The final diagnosis of all the patients was confirmed by one MD (LS) after checking out their outcome and therapy.

As expected, overall baseline characteristics (demographic, laboratory and imaging) significantly differed between the two groups (Graves' hyperthyroidism versus non-Graves' hyperthyroidism), with the exception of the median thyroid volume (consequence of the high prevalence of nodular disease in patients with non-Graves hyperthyroidism; Table 1).

Among the GD group, the ultrasonography pattern 3 was assigned to 60 patients, while the echo patterns 2 and 1 were associated with 24 and two patients, respectively. Yet, among patients with non-Graves' hyperthyroidism in only three cases, US images were read as echo pattern 3, while 14 cases, 12 cases and nine cases corresponded to echo patterns 1, 2 and 0, respectively.

Accuracy and Comparison of Diagnostic Tests

For our data set, ROC curves showed that the optimal cut-off values associated with the highest diagnostic sensitivity and specificity was 0.7 IU/L for TRAb Kryptor® (93 [85.4–97.4] and 86.8 [71.9–95.5]) and 0.1 IU/L for Immulite® TSI (94.2 [86.9–98.1] and 84.2 [68.7–93.9]), respectively (Figure 1A). Accordingly, six false negative (FN) (final diagnosis of Graves' hyperthyroidism along with negative TRAb) and five false positive (FP) (four final diagnosis of AFTN and one of painless thyroiditis, along with positive TRAb) results were recorded using TRAb Kryptor® assay.

Moreover, five FN (GD and TSI <0.1 IU/L) and five FP results (the same four cases of AFTN found positive with TRAb Kryptor® assay and one case of painless thyroiditis other than that positive with TRAb Kryptor® assay, along with TSI >0.1 IU/L) were found by using Immulite® TSI assay. Finally, discordant TRAb/TSI results were found in two cases: one patient with GD (one out of 86, 1.2%) showed TRAb negativity along with TSI positivity; and one case with final diagnosis of painless thyroiditis (one out of 38, 2.6%) was TRAb-negative and TSI-positive, respectively.

Figure 1.

(A) Receiver operating characteristic curves for TRAb Kryptor® and Immulite® TSI with the optimal cut-off values associated with the highest diagnostic sensitivity and specificity. (B) ROC curve for TcTU with the optimal cut-off value associated with the highest diagnostic sensitivity and specificity

As concerns the clinical effectiveness of the echo pattern 3 ('thyroid inferno') in identifying Graves' hyperthyroidism, we found a good sensitivity (92.1%) and a high PPV (95.2%) but a quite low specificity value (69.8%) and a relative low NPV (57.5%). Specifically, 26 patients with GD did not have the echo pattern 3, thus representing the FN cases (Graves' hyperthyroidism without the proper US pattern of 'thyroid inferno'); while three patients were considered as FP cases, since they showed the echo pattern 3, but they were categorized as having non-Graves' hyperthyroidism.

For thyroid scintigraphy, the TcTU cut-off value of 1.3% corresponded to the best limit for sensitivity and specificity in our patients (95.3 [88.5–98.7] and 96.4 [81.6–99.4], Figure 1B). Employing the TcTU cut-off of 1.3%, we found four FN (GD and TcTU <1.3%) and one FP (one transient thyrotoxicosis along with TcTU >1.3%) cases.

As summarized in Table 2, the best accuracy belonged to the quantitative scintigraphy (95.6%, using the threshold of 1.3%), while the lowest value was found for echo pattern 3 (76.6% accuracy).

For both the Passing-Bablok regression and for the Bland-Altman test of the two antibody assays, four patients of our cohort were excluded since their TSI values were reported as >40 IU/L, thus not be comparable to the absolute coupled TRAb values. The Passing-Bablok regression equation (Figure 2A) had as the intercept the value of −0.085 (95% CI, −0,1184 to −0,06 107) and as the slope 0.6167 (95% CI, 0,5369–0,7279), showing that the smallest scattering of points and the best concordance between the assays were present for most of the reported TSHR antibody values, mainly <10 IU/L. Bland-Altman test (Figure 2B) showed a mean bias of 3.3 IU/L with a standard deviation of 8.1 IU/L and its confidence limits of 19.2 and −12.6 IU/L; the scattering of points is diminished, and most points lie relatively close to the line which represents mean bias. Also of note, and evident to both Passing-Bablok regression and the Bland-Altman plot, there was that the concordance between the two assays mostly occurred for low values (and low difference values) of antibodies while it worsened proportionally with increasing results. Thus, a great degree of correlation and agreement were seen between TRAb Kryptor® and Immulite® TSI results.

Figure 2.

(A) Passing-Bablok regression of the two antibody assays (TRAb Kryptor® and Immulite® TSI). (B) Bland-Altman test of the two antibody assays (TRAb Kryptor® and Immulite® TSI)

Table 3 showed that no difference existed between the areas of ROC (AUC) curves for TRAb and TSI, while the AUC of TcTU was slightly larger than the AUCs of TRAb and TSI, respectively.

Correlations Among Variables

Table 4 depicts the correlation analysis of TRAb and TSI levels with thyroid hormones and uptake in patients with Graves' hyperthyroidism. Both TRAb and TSI values showed a significant linear correlation with fT4 (r s = .25, P = .02 and r s = .28, P = .01, respectively), but not with fT3 (r s = .16, P = .13 and r s = .19, P = .08, respectively). Instead, TcTU results were tightly linked with TRAb (r s = .66 and P = .0001, Figure 3A) and TSI (r s = .57 and P = .0001, Figure 3B), respectively.

Figure 3.

(A) Correlation analysis of TRAb levels with thyroid 99mTc-pertechnetate uptake rate (TcTU) in patients with Graves' hyperthyroidism. (B) Correlation analysis of TSI levels with thyroid 99 mTc-pertechnetate uptake rate (TcTU) in patients with Graves' hyperthyroidism