Diagnostic, treatment, and follow-up DTC guidelines have recently evolved towards a less aggressive approach that ensures adequate monitoring, especially in less aggressive forms [2]. In fact, the definition of ER in DRS is based exclusively on the absence of imaging findings (mainly cervical ultrasound) and hs-Tg levels under levothyroxine suppressive treatment < 0.2 ng/mL or rhTSH-Tg < 1 ng/mL [2]. However, the absence of imaging finding but elevated Tg levels would classify these patients as IR (hs-Tg 0.2-1.0 ng/mL or rhTSH-Tg 1–10 ng/mL) or BIR (hs-Tg > 1.0 ng/mL or rhTSH-Tg > 10 ng/mL). Therefore, Tg is a key marker of recurrence or persistence disease in the long term evolution of DTC [11]. Nevertheless, these DRS cut-off points for Tg are arbitrary, mainly based on retrospective studies, and may be inconsistent when Tg is measured after stimulation (rhTSH-Tg) or under suppressive therapy (hs-Tg) for the same patient [7, 11].
The DRS allows reclassification of the DTC status throughout its evolution independently of its initial recurrence risk. Those with structural incomplete response (SIR) have the highest risk of morbidity and mortality in DTC. Hence the importance of Tg as a predictive marker of SIR [2].
During the last two decades, significant progress has been made in the measurement of hs-Tg. The hs-Tg assays have a 10-fold higher sensitivity than first-generation assays, which could theoretically obviate the need for Tg stimulation (either by thyroid hormone withdrawal or rhTSH) [12]. However, the implications of DRS reclassification by rhTSH-Tg levels are unclear, especially in those patients with BIR and IR [13, 14].
In the present study we evaluated both questions: 1) the long-term predictive value of hs-Tg and rhTSH-Tg 12 months after completion of initial treatment (total thyroidectomy and I131 ablation) in a cohort with mean follow-up of 7.9 years; and 2) the divergences in DRS according to hs-Tg or rhTSH-Tg levels and their implications on the follow-up.
In the present article, the predictive value of ER at the last follow-up visit for Tg levels 12 months after completing initial treatment showed a high NPV. In fact, a hs-Tg cutoff point of 0.110 ng/mL showed a NPV of 100%. That is, all patients with an undetectable hs-Tg level under levothyroxine treatment 12 months after completing treatment with total thyroidectomy and I131 ablation remained in ER throughout follow-up. These results agree with those previously published in which the NPV is estimated to be around 97–99%. Thus, the risk of a false negative at undetectable hs-Tg levels would be no more than 1% [7, 11].
Surprisingly, the results obtained for rhTSH-Tg did not improve the overall test performance with respect to the disease status at the last follow-up visit. In fact, for a rhTSH-Tg cutoff point of 0.815 ng/dl the NPV and PPV of the test worsened slightly with respect to the predictive ability of hs-Tg. These worse results were due to a single 44-year-old female patient with follicular variant CDT and IR after a follow-up longer than 9 years (last follow-up visit hs-Tg = 0.22 ng/dl) and rhTSH-Tg from 0.19 to 0.37 ng/dl. Despite the borderline positivity in hs-Tg levels, it did not reach the cut-off point established by International Guidelines for rhTSH-Tg DRS, not even the one with the highest sensitivity/specificity calculated for the present study. Although it is true that the risk of developing Structural Disease in this patient is very low, it is undeniable to recognize that the cut-off points established for the definition of the DRS are arbitrary and may even cause divergences in DRS classification depending on whether rhTSH-Tg or hs-Tg is assessed, even though there is a clear correlation between both values (Fig. 1). In any case, our results support the inefficacy of rhTSH-Tg in predicting long-term response to DTC, making this costly procedure unnecessary, with a negative impact on patients´ quality of life, especially in those with an ER after completing initial treatment (certainly the most common subgroup in CDT) [11, 15, 16].
At the same time, special attention should be paid to the lower PPV independent of the type of Tg measurement. Tg levels 12 months after completing initial treatment are not exempt from the risk of false positive results. In fact, 15.7% of patients identified as ER at the end of follow-up were classified as (false) positives in the initial assessment. The persistence of healthy thyroid tissue due to the temporal proximity of the I131 ablative dose, the greater sensitivity of ultrasensitive tests, and even the evolution towards spontaneous resolution in a high proportion of DTC could explain this fact [17]. In any case, our results are similar and even slightly superior to those previously described [11]. It supports the usefulness of hs-Tg as a discrimination tool for those patients with initial ER whose frequency and intensity of follow-up could be reduced, with the consequent cost savings, optimization of resources and reduction of patient anxiety during follow-up [18]
Finally, the DRS assess the risk of recurrence at any time during the DTC evolution in relation to the treatment response (ATA2015). In this sense, those patients without structural disease by imaging will define their DRS, and therefore the risk of recurrence, exclusively based on Tg levels [4, 5]. These Tg cut-off points are arbitrary, based on retrospective studies, and may differ depending on whether they are classified by hs-Tg or rhTSH-Tg [11].
In the present study, DRS classification using rhTSH-Tg 12 months after completing initial treatment overestimated by more than twofold the number of patients initially identified as IR in both ER and BIR groups at the last follow-up visit. However, it provided a slightly better predictive value in those patients with an initial IR using rhTSH-Tg (93.3%) than hs-Tg (73.3%). Despite these apparent discrepancies between DRS classifications, the risk of recurrence associated with BIR or IR remains nowadays a subject of debate and depends on numerous variables (progressive increase in Tg levels, doubling time, etc.) and may evolve towards spontaneous resolution without further treatment in a large number of patients [17, 19]. Moreover, as shown in Table 3, 13.3% of patients with IR by rhTSH-Tg spontaneously progress to ER during follow-up. In fact, the clinical relevance of detectable but only slightly elevated hs-Tg levels is still unclear.
Although there is a recent study aimed at evaluating the usefulness of rhTSH-Tg 5 years after completion of initial treatment to reclassify DRS in those cases with IR or BIR [14], a regular hs-Tg monitoring strategy would probably have a similar result with a lower cost and patients´ anxiety due to the overall good prognosis [20].
The present study has certain limitations to be considered. First, classic studies attribute to DTC the risk of recurrence even several decades after diagnosis [21], which would reduce the predictive value of hs-Tg after completing the initial treatment even in a cohort with an average follow-up of almost 8 years. However, more recent studies show that more than 80% of recurrences are identified in the first 5 years [22], which would support the clinical relevance of our results. This improvement in the early detection of recurrences is essentially due to the technical improvement on cervical ultrasound imaging and hs-Tg during the last decade. In this sense, the high NPV of undetectable hs-Tg and the absence findings on imaging evaluation, added to the higher risk of recurrence within the first 5 years might support the possibility of extending follow-up and even definitively discharging those patients with low-risk DTC after this period [18]. Finally, the smaller number of patients with IR/BIR compared to those with ER could reduce the statistical significance in those patients, however the prospective and longitudinal design of the present study as well as the similarity with other current cohorts ensure its representativeness and reflect the reality of DTC follow-up [23].
In conclusion, in those patients with ER 12 months after completing initial treatment (total thyroidectomy and I131 ablative treatment) the rhTSH-Tg did not provide relevant information. An ER after initial treatment was associated with a high NPV. In those patients with IR or BIR, rhTSH-Tg identified a greater number of patients with IR, with no apparent practical implication during follow-up.