Predictive Value for Disease Progression of Serum Thyroglobulin Levels Measured in the Postoperative Period and After ¹³¹I Ablation Therapy in Patients with Differentiated Thyroid Cancer

MATERIALS AND METHODS

RESULTS

At the time of IAT, thyroid remnants were present on posttherapy scans in all patients except one. The median Tg1 value was 9 ng/mL (range, 0–2,470 ng/mL). In one patient, a very high level of Tg1 (2,470 ng/mL) was observed and controlled, raising the possibility of initial distant metastases. The posttherapeutic WBS showed no foci of uptake outside the thyroid bed but only a large thyroid remnant, confirmed by ultrasonography. No pathologic lymph node was evident. The chest radiograph was also normal. As the Tg level and the scintigraphic image normalized progressively with time, distant metastases were ruled out. In 4 other patients, cervical node metastases were immediately suspected on posttherapeutic scintigraphy, evident on ultrasonography and confirmed a few months later by surgery. These 4 patients, in whom node metastases were associated with high Tg1 values (range, 138–448 ng/mL), were excluded from further analysis.

Among the 208 patients clinically free of disease at the time of IAT, 30 patients had a Tg1 level > 30 ng/mL. At the time of control, 6–12 mo later, 30 patients had an elevated Tg2 level, 19 of whom had initially a Tg1 level > 30 ng/mL. After initial therapy, considering the whole follow-up, 176 patients (85%) were progressively in complete remission, with an undetectable Tg level off therapy. One patient, after complete remission, presented a significant increase of the Tg level off therapy 2 y later, with evidence of bone metastases. Four patients, despite a low and stable Tg level, presented evidence of disease progression (3 local recurrences, 1 lung metastases). In 27 patients, the Tg level never normalized but remained high or increased; 15 of them presented a clinical evolution (12 nodal metastases, 1 local recurrence, 2 distant metastases). In the other 12 patients, despite a persistent elevated Tg level, a complete work-up showed no clinical evidence of disease. Evolution of thyroid disease according to Tg1 and Tg2 levels is presented in Table 2.

Finally, only 20 patients (9%) presented a disease progression. The median time to diagnosis was 2.4 y (range, 0.7–7.2 y). Eighty percent of the clinical events occurred before 5 y. Ultrasonography was contributory in 12, CT in 5, and WBS in only 3 patients. The initial lymph node status in these patients (pN0 = 5%, pNx = 30%, pN1 = 65%) was significantly different (P < 0.0001) compared with that of disease-free patients (pN0 = 32%, pNx = 47%, pN1 = 21%). The median Tg1 level also differed significantly compared with that of patients clinically free of disease (41.5 and 6.7 ng/mL, respectively; P < 0.002).

Predictive variables for disease progression were studied. The lymph node status was among the studied variables. As illustrated in Figure 1, progression-free survival curves were not significantly different between pNx and pN0 patients (P = 0.18), and the 2 groups of patients were combined together, versus the pN1 patients. With univariate analysis, 5 variables were significantly associated with disease progression (Table 3). Ranking in decreasing order of odds ratio (OR) values were Tg2 level > 10 ng/mL, Tg1 level > 30 ng/mL, lymph node invasion, extrathyroidal extension, and tumor size > 4 cm. The most important were Tg2 and Tg1 with very high ORs. When we reexamined our data by multivariate analysis (Table 4), only Tg2 (OR = 16.4, 95% confidence interval [95% CI] = 5.7–47.4; P < 0.001) and node invasion (OR = 2.7; 95% CI = 1.0–7.2; P = 0.048) remained as independent prognostic factors. When only parameters known just before IAT were considered, Tg1 and node invasion were the only 2 independent predictive factors (Table 5). Despite the presence of thyroid remnants, the Tg1 OR was significant (OR = 10.1; 95% CI = 4.0–25.7; P < 0.001) but was lower than the Tg2 OR in the previous analysis. In contrast, the OR increased for the lymph node invasion (OR = 4.4; 95% CI = 1.7–11.2; P = 0.002).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Progression-free survival curves according to initial lymph node status. There was no significant difference (P = 0.18) between pN0 and pNx patients. A significant difference was found between pN0 and pN1 (P = 0.0001) and between pNx and pN1 (P < 0.0001).

As Tg1 threshold was arbitrarily chosen at 30 ng/mL, this cutoff value was checked later on with a receiver-operating-characteristic analysis. Taking into account the clinical events that occurred during the follow-up, the accuracy was calculated from 5 to 100 ng/mL, in steps of 5 ng/mL, and ranged from 48% to 90%. The best accuracy was obtained for Tg1 values between 30 and 100 ng/mL.

The prognostic role of Tg1 and Tg2 levels was also illustrated with the progression-free survival curves (Figs. 2 and 3). Disease progression was more frequent among patients with Tg1 levels > 30 ng/mL compared with that among patients with Tg1 levels ≤ 30 ng/mL. At 5 y, progression-free rates were 66% and 96%, respectively (P < 0.0001). The difference was more important with Tg2. Progression-free rates were 55% and 97% (P < 0.0001).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Progression-free survival curves according to Tg1 level measured in postoperative period just before IAT.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Progression-free survival curves according to Tg2 level measured 6–12 mo after IAT.

Possible factors to explain the importance of high Tg1 levels were analyzed. No correlation was found with the TSH level or the time interval from surgery. In the group with Tg1 levels > 30 ng/mL and in the group with Tg1 levels ≤ 30 ng/mL, the medians of TSH values did not differ significantly. They were 64.9 and 74.8 mIU/mL, respectively (not significant). The time interval between surgery and Tg1 measurement was also similar in the 2 groups, 2.4 ± 1.5 mo (mean ± SD) versus 2.6 ± 2.0 mo (not significant). Statistical associations were sought between high Tg1 levels and other histopathologic features. With a logistic regression model, extrathyroidal extension was found as the only independent predictive factor of high Tg1 values (OR = 2.7; 95% CI = 1.2–6.3; P = 0.02).

Finally, at the time of IAT and at the time of control, node invasion and high Tg levels under TSH stimulation were the 2 prognostic variables outlined by multivariate analysis. Patients at risk of disease progression were patients with an elevated Tg1 or Tg2 level or a nodal invasion. Taking into account the 2 prognostic variables combined together, the 5-y progression rate was estimated in a subgroup of 109 patients followed-up during 5 y or more. Fourteen of 16 clinical evolutions (87.5%) were reported in the patients with a node invasion or an elevated Tg1 or Tg2 level (Tables 6 and 7). The positive predictive value (PPV) and the negative predictive value (NPV) of the 2 prognostic variables were calculated. Considering the node invasion combined with Tg1, these values were 37% and 97%, respectively. Using Tg2, similar values were obtained, 40% and 97%, respectively.

DISCUSSION

In this study of 212 patients with differentiated thyroid carcinoma, all patients had a total or near-total thyroidectomy followed by IAT. The follow-up schedule was the same in all cases, providing a homogeneous clinical management of the studied population. Only 9% of the patients presented clinical recurrences with a median follow-up of 5 y. This small number is similar to that of other reports (14–17). With the help of ultrasonography, lymph node metastases were often found, even in two thirds of the patients who had a cervical node dissection. Diagnostic WBS was of little help; many recent studies have shown its poor usefulness, in contrast to ultrasonography (18–22).

Among the 5 predictive factors for disease progression found by univariate analysis, size > 4 cm, extrathyroidal extension, and lymph node invasion are well-known prognostic factors (15-17,23–25). With multivariate analysis, lymph node invasion was the only morphologic predictive factor. In this series, pN0 patients were pooled with pNx patients in whom node dissection was not performed because of a normal aspect at the time of cervicotomy. It was previously determined that recurrence-free survival curves did not differ significantly between these 2 groups.

In this study, age did not appear as a predictive variable of recurrence. This was unexpected as age is an important prognostic factor in thyroid cancer patients (14–15,23–25). This is probably related to the small number of recurrences reported and also to the high incidence of nodal recurrences observed in this series in younger patients. Seven of 12 nodal recurrences were found in patients <45 y old, whereas all local and distant recurrences were observed in patients >45 y old.

Above all, the 2 major factors to predict the clinical evolution were the biologic factors, the Tg1 and Tg2 levels measured before and after IAT. In univariate analysis, the Tg2 level measured 6–12 mo after therapy was the most important predictive factor among all parameters. When Tg1 and Tg2 were entered into a multivariate model, only Tg2 emerged as a stronger predictive factor, in association with the lymph node invasion. The Tg level measured several months after IAT is a well-known marker of the disease course. Undetectable levels under TSH stimulation are good indicators of the disease-free status, whereas elevated levels correlate with persistent disease (6,19). For Tg2, a threshold of 10 ng/mL was used in the analysis. A Tg2 level above 10 ng/mL is considered pathologic by many investigators because such levels are highly predictive of recurrences (4,7,26). This has been determined even when WBS performed at the same time is negative (2,3,6). As observed recently by Mazzaferri et al., this cutoff, albeit arbitrary, is often used for empiric ¹³¹I therapy (27).

The prognostic value of the initial Tg level measured in the postoperative period just before IAT is often debated because of the presence of thyroid remnants that contribute to the Tg synthesis. In our series, although functional thyroid remnants were observed in all patients except one on the posttherapeutic WBS, Tg1 levels above 30 ng/mL were significantly associated with disease progression. Tg1 had a reliable prognostic value but was less significant than Tg2. Until now, only a few reports have demonstrated the prognostic value of the Tg1 level. Some have shown that the Tg1 measurement could be an early indication of the existence of a metastatic disease (7–9,11), but most metastases were initial metastases discovered on posttherapeutic WBS. The purpose of this study was to analyze the possible role of Tg1 on the course of the disease, in patients clinically free of disease after the initial therapy. This is why 4 patients with very high Tg1 levels, in whom initial node metastases were discovered at the time of IAT, were excluded from the analysis. Others have pointed out that patients with high Tg1 levels were at increased risk of recurrence (10,12). But no hypothesis was suggested to explain the high Tg1 levels in patients in whom recurrence developed. In our study, factors as the TSH stimulation level and the time interval between surgery and Tg1 measurement were eliminated. By logistic regression analysis, Tg1 levels above 30 ng/mL were demonstrated to be significantly associated with the extrathyroidal extension, suggesting a basis for the pejorative evolution of these patients. Interestingly, in the study of Hall et al., the median Tg1 level was particularly high in stage III patients presenting an extrathyroidal extension (10). In all of these reports, no comparison was done with Tg2. In our study, we evaluated and compared the prognostic value of the 2 Tg determinations in the same patients. Despite a lower OR, Tg1 had a significant prognostic value at the time of initial therapy.

For the determination of Tg1 threshold, to our knowledge, no reference exists in the literature about the normal distribution of postoperative Tg values in thyroidectomized patients. Very few investigations have studied the prognostic value of Tg1. Thresholds used vary from 10 to 69.7 ng/mL. In the series of Oyen et al., a cutoff value of 10 ng/mL was chosen before ablation because no initial metastases were found below this value (7). However, this cutoff was too low to discriminate between patients with thyroid remnants and those with initial distant metastases who represented only 16% of the patients. Ruiz-Garcia et al. used a cutoff of 23 ng/mL, chosen at the upper limit of the normal range in his laboratory (12). Hall et al., after examining several different values, selected a threshold of 20 ng/mL as the optimum cutoff point because the hazard of recurrences was greatest at this level (10). Ronga et al. performed a receiver-operating-characteristic curve analysis and showed that a threshold of 69.7 ng/mL gave the best PPV for the presence of metastases (9). This higher value of cutoff was related to the population of patients studied. Ronga et al. included many patients with initial distant metastases. In the absence of available references, the choice of a threshold was necessarily somewhat arbitrary. Any value chosen between 30 and 100 ng/mL would not have changed the accuracy. This could explain the various values used by other authors. The 30 ng/mL value was in the lower range of the best accurate values and yielded the best sensitivity.

Finally, considering all results of multivariate analyses, patients at risk of disease progression were patients with a high Tg1 or Tg2 level under TSH stimulation or a node invasion. The PPV of these 2 variables combined together was low. Baudin et al. reported recently a low PPV of the Tg level measured during the first year of follow-up (28). This was related to the low recurrence rate in his series and to a progressive decrease of Tg levels in a significant percentage of patients. A low progression rate was also observed in our series. Furthermore, only clinical events proven by imaging modalities or surgery were considered. Twelve patients with a high Tg2 level maintained a Tg level above 10 ng/mL during the whole follow-up without evidence of disease.

Conversely, it was worth noting the high NPV of a low Tg level associated with the absence of node invasion. This concerned two thirds of our patients for whom the likelihood of complete remission was very high. Interestingly, these PPVs and NPVs were the same, using either Tg1 or Tg2. These results reinforce the idea that the Tg1 threshold of 30 ng/mL, arbitrarily chosen at the upper limit of the normal range, was relevant.

CONCLUSION

Although Tg2 appears more significant than Tg1, both Tg levels, measured under TSH stimulation in the postoperative period and after ablative therapy, have a predictive value for disease progression. Taking into account the lymph node status, patients at risk are the same before and after IAT. In clinical practice, they can be selected as soon as Tg1 is known. In patients with a low Tg1 level and no node invasion, who represent the majority of our population, the risk of a clinical evolution is very low. In these patients, the need for periodic scintigraphic controls and Tg determinations under TSH stimulation seems questionable and clinical management should be as minimal as possible. Conversely, patients with a node invasion or a high Tg1 level should benefit from a more intensive follow-up.