Early stimulated thyroglobulin for response prediction after recombinant human thyrotropin-aided radioiodine therapy

Total thyroidectomy followed by radioiodine therapy (RIT) is the treatment of choice for differentiated thyroid carcinoma (DTC) [1, 2]. Although DTC shows a relatively good response to RIT, unsuccessful ablation is reported in about 6–30% of patients [3‐8], so additional treatments, such as surgical procedures or repeated RIT, and life-long follow-up are required. Serum thyroglobulin (Tg) levels measured during hypothyroidism, just before RIT, have good prognostic values for predicting successful ablation [9‐12].

Recombinant human thyrotropin (rhTSH) represents a safe and effective means of increasing serum TSH in preparation for RIT, avoiding signs and symptoms of hypothyroidism associated with thyroid hormone withdrawal (THW) [13, 14]. I-131 is usually administered on the day after the 2nd injection of rhTSH according to the conventional schedule [5, 6]. On the other hand, serum Tg should be measured 72 h after final injection of rhTSH, because this day corresponds to the peak serum Tg levels [15, 16]. However, due to the radioiodine-induced damage to normal thyroid cells and release of stored Tg in thyroid follicles, the Tg level measured after RIT would not be reliably predictive [17, 18].

The present study was undertaken to determine the value of Tg measured before and after RIT that predicts a therapeutic response in DTC patients with rhTSH-aided RIT.

Patients’ characteristics are summarized in Table 1. The study population comprised 114 women (76.0%) and 36 men (24.0%), aged 20–79 years (mean 48.8) at the time of surgery. The histological DTC subtype was papillary in 148 cases (98.7%) and follicular in 2 cases (1.3%). At the time of RIT, early Tg ranged from 0.1 to 117.9 ng/mL (mean 3.2) and delayed Tg ranged from 0.1 to 198.7 ng/mL (mean 9.6).

At the follow-up examination, an excellent response was observed in 85 (56.7%) patients. In the 65 (43.3%) patients with non-excellent responses, 10 (6.7%) patients had a biochemical incomplete response, 10 (6.7%) had a structural incomplete response, and 45 (30.0%) had an indeterminate responses.

Variables significantly associated with an excellent response in the univariate analysis are reported in Table 2. Older age (p = 0.041), smaller tumors (p = 0.009), and low level of early Tg (p < 0.001) significantly predicted an excellent response. In the multivariate logistic regression analysis, early Tg [odds ratio (OR) 2.012; 95% confidence interval (CI) 1.384–2.925; p < 0.001] and tumor size (OR 1.714; 95% CI 1.019–2.882; p = 0.042) significantly predicted an excellent response (Table 3). However, delayed Tg did not show statistical significance for response prediction after RIT (OR 0.992; 95% CI 0.969–1.015; p = 0.492).

An ROC curve analysis showed that the optimal cut-off value for early Tg was 2.0 ng/mL; it had a sensitivity of 46.2%, specificity of 95.3%, PPV of 88.2%, and NPV of 69.8% for predicting a non-excellent response (AUC 0.704; p < 0.001) (Fig. 1). In 116 patients with an early Tg level ≤2.0 ng/mL, an excellent response was observed in 81 patients (69.8%), whereas only 4/34 (11.8%) patients with early Tg levels >2.0 ng/mL showed an excellent response (Table 4).

In patients with DTC, the stimulated Tg level measured just before RIT proved to be a biochemical tumor marker that could predict persistent or recurrent disease after THW [9‐12]. In a recent meta-analysis involving 3947 patients across a broad spectrum of disease, it was demonstrated that the pre-ablation Tg level could be a useful negative predictor of persistent or recurrent DTC. The overall NPV was 94% when the pre-ablation Tg level was <10 ng/mL [12]. In addition, the stimulated Tg measured after THW could be used to predict the volume of the thyroid remnants or the residual tumor burden [22]. However, there is no consensus available with respect to the clinical implications or testing protocol for stimulated Tg at rhTSH-aided RIT, although rhTSH is now widely used as a method for RIT preparation.

From the previous studies, serum Tg should be measured in the 3 days after the 2nd injection of rhTSH [15, 16]. In 1998, Genzyme showed the serial Tg data that resulted in FDA approval. In that study protocol, 24 h after the 2nd rhTSH injection, 148 MBq radioiodine were administered orally and Tg levels were measured on days 1, 2, 3, and 7. They found that maximum serum Tg levels were observed on day 3 in most patients. However, the administered dose of radioiodine was not a therapeutic dose that could cause acute thyroid injury [16]. This timing of Tg measurement has raised concerns about its reliability as a tumor marker, because acute thyroid injuries would contribute to Tg production. Taieb et al. [17] showed sequential elevation of median serum Tg levels during the 48 h after I-131 administration (3.7 GBq) in DTC patients with hypothyroidism. Tg levels increased due to an acute effect of I-131 on Tg release. Furthermore, the Tg increments at 48 h were not correlated with Tg levels measured before I-131 administration.

There have been inconsistent studies to show the prognostic value of early Tg and delayed Tg in DTC patients prepared with rhTSH. Ciappuccini et al. [23] showed a significant correlation between the rhTSH-stimulated Tg levels measured immediately before RIT and persistent/recurrent disease (PRD), defined as evidence of tumor burden confirmed by histology or radiological modalities on follow-up. The early Tg levels were significantly lower in nonstructural (microscopic) than in structural (macroscopic) residual disease. The reported Tg cut-off value for predicting PRD was 2.8 ng/mL, closed to the threshold defined in our study, with a sensitivity of 86%, specificity of 64%, PPV of 24%, and NPV of 97%. On the other hand, Melo et al. [24] demonstrated that rhTSH-stimulated Tg measured 3 days after RIT had a predictive value for disease persistence or recurrence 1 year later in 131 consecutive patients. From this study, Tg levels ranged from 0 to 1927 ng/mL and were significantly lower in the disease-free group (17.9 ± 49.1 vs 136.3 ± 341.0 ng/mL, p < 0.001), with an optimal Tg cut-off level of 7.2 ng/mL. However, there was no previous report to compare the prognostic value of early Tg and delayed Tg.

Giovanella et al. [25] suggested that even Tg measured before TSH stimulation can reflect expression of thyroid remnants by showing a significant relationship between neck radioiodine uptake measured just before ablation and Tg level measured before rhTSH stimulation and under T4 treatment. Our study showed that Tg cutoff at 2.0 ng/mL provided high specificity (95.3%) and PPV (88.2%) for predicting a non-excellent response, which could be very helpful to decide follow-up strategy based on early Tg. However, the early Tg still has several limitations as predictive factor due to the short TSH stimulation time. In our study, Tg cutoff at 2.0 ng/mL provided insufficient sensitivity (46.2%) and NPV (69.8%). Tg cut-off value under short TSH stimulation could affect diagnostic accuracy. Our results also indicated that 30.2% of patients with Tg levels ≤2.0 ng/mL could not achieve an excellent response. Most of them belong to an indeterminate response group (Table 4), which of 13–20% are reclassified as having PRD over approximately 10 years of follow-up [26]. Further investigations are necessary to overcome the physiologic drawbacks of early Tg as a reliable predictive marker.

There are several limitations in our study. First, this study was retrospective, so a selection bias was inevitable. Especially, our protocol for I-131 dose selection has changed over the years based on recent studies and guidelines, which the choice of administered dose was heterogeneous, although there was no significant difference in therapeutic response among I-131 dose like previous studies (Table 2) [5, 6, 19]. Second, delayed Tg was measured 96 h after the 2nd injection of rhTSH in some patients due to their admission schedules. However, between the two patient groups undergoing Tg testing in 72 or 96 h, there was no significant difference in the delayed Tg (9.8 ± 29.7 vs 8.8 ± 21.0 ng/mL, p = 0.832) and delayed TSH (34.1 ± 27.1 vs 28.0 ± 28.1 mU/L, p = 0.202) levels. Third, although assessment of therapeutic responses was based on the latest ATA guidelines, follow-up protocols were not consistent for all patients. Some patients underwent DxWBS with stimulated Tg, and some patients underwent neck US with suppressed Tg. Further large-scale prospective studies are required to better determine the significance of early Tg.

In conclusion, the early stimulated serum Tg just before rhTSH-aided RIT significantly predicted therapeutic responses after RIT in patients with DTC. Therefore, physicians should consider measuring the serum Tg level before RIT to predict the therapeutic response, although the early Tg does not precisely reflect the remnant thyroid tissue burden in DTC patients prepared with rhTSH.