Differentiated thyroid cancer (DTC) is the most prevalent endocrine malignancy, comprising approximately 90% of all thyroid malignancies, and has demonstrated a sustained rise in global incidence in recent decades (1–3). The management of DTC primarily involves a multimodal approach, including total thyroidectomy, radioiodine (RAI) therapy, and subsequent thyroid-stimulating hormone (TSH) suppression therapy (4–8). Among these, postoperative RAI therapy is pivotal for eradicating residual disease and has been associated with improved disease-specific survival (1, 9, 10).

Dynamic risk stratification (DRS) serves as a key tool for assessing the risk of persistent or recurrent disease in DTC (1, 11). This system classifies patients into four response categories based on their initial therapeutic outcome: excellent response (ER), indeterminate response (IDR), biochemical incomplete response (BIR), and structural incomplete response (SIR) (9). BIR is specifically defined by abnormal suppressed thyroglobulin (sup-Tg) or stimulated thyroglobulin (sTg) levels, or rising anti-Tg antibody (TgAb) titers, in the absence of identifiable structural disease on imaging (2, 9, 11). Patients classified with BIR demonstrate a significantly elevated risk for both structural recurrence and disease persistence (12, 13).

Fortunately, a substantial proportion of patients with an initial BIR attain favorable long-term outcomes. Spontaneous reclassification to no evidence of disease status occurs in 33.7% to 38.0% of these patients without further intervention (14, 15). However, the clinical course is heterogeneous; while approximately one-third experience remission, an estimated 17–20% will progress to structurally identifiable disease, necessitating further intervention (13, 16). The specific risk factors predictive of disease progression within the BIR population have not been definitively established. Therefore, this study was designed to evaluate the clinical outcomes and to identify predictors of progressive disease (PD) in patients who exhibited a BIR following initial RAI therapy.

RAI therapy represents a cornerstone in the postoperative management of DTC, aimed at eradicating residual disease and reducing recurrence risk (8, 9). Despite the generally favorable prognosis associated with DTC, a notable subset of patients, approximately 11–22%, exhibit a BIR following initial RAI therapy (2, 22). Among these BIR patients, an estimated 17–20% will eventually progress to structurally identifiable disease, underscoring the clinical challenge of risk stratification within this heterogeneous population (13, 16).

In a retrospective analysis by Ahn et al. (13), longitudinal surveillance outcomes were evaluated in a cohort of 266 BIR patients who were managed without additional RAI intervention. Unfortunately, after a median follow-up of 12 years, structural recurrence was identified in 43 patients, representing a substantial proportion of the cohort. In the present study, our findings reveal a distinct prognostic dichotomy: while a majority (59.9%) maintained NPD status over a median follow-up of 48.6 months, a substantial proportion (40.1%) experienced disease progression. The 5-year PFS rate of BIR patients was 60.4%. Additionally, the clinical factors associated with PFS in BIR cohort were assessed, and it was observed that patients with older age at diagnosis, presence of the BRAF^V600E mutation, advanced tumor staging (T3b-T4), high recurrence risk stratification, and sTg levels≥50.0 ng/mL might have a poorer PFS than other patients. However, other clinical factors such as gender, multifocality, histological type, and N stage were not associated with PFS. Multivariate Cox regression analysis identified sTg ≥50.0 ng/mL and advanced primary tumor stage (T3b–T4) as independent predictors of PD. These results highlight the necessity of refining risk stratification beyond the broad BIR classification by incorporating readily available clinicopathological and biochemical markers, thereby facilitating individualized surveillance and informing timely therapeutic intervention.

The observed PD rate of 40.1% aligns with and substantiates the non-indolent nature of a significant BIR subset reported in prior literature (22, 23), reinforcing that biochemical persistence signifies active, albeit radiologically occult, disease burden in many patients. Conversely, the 18.0% rate of spontaneous reclassification to an ER status without additional RAI therapy validates the current conservative, observative approach for a portion of BIR patients, as recommended by 2025 ATA paradigms (9).

Indeed, spontaneous decline in Tg levels, ultimately leading to reclassification as remission without any additional therapy, has been documented in 6–33% of BIR patients (2, 13, 14, 24). The likelihood of this spontaneous biochemical remission is inversely correlated with the initial sTg level. This relationship was quantified by Vaisman et al. (14) in a cohort of 77 BIR patients, where the median sTg was significantly lower in those who achieved spontaneous remission (13 ng/mL) compared to those who did not (33 ng/mL; p = 0.03). Notably, in a retrospective study of 32 BIR patients after initial RAI therapy, Pitoia et al. (25) reported that after a median follow-up of 6 years, no patient with a sTg level >10 ng/mL transitioned spontaneously to remission status. Our data suggest that the sTg level serves as a potent surrogate for residual tumor volume and biologic aggressiveness. The strong predictive performance of the 50.0 ng/mL cutoff (AUC 0.774) provides a quantitative, clinically actionable threshold. sTg levels ≥50.0 ng/mL likely reflect a higher tumor burden not yet detectable by conventional anatomic imaging, portending a greater likelihood of eventual biochemical or structural progression. This finding is physiologically coherent, as Tg secretion correlates with functional thyroid tissue mass, and its persistence after thyroidectomy and RAI ablation indicates incomplete response (9, 26, 27).

The independent prognostic value of advanced T stage (T3b-T4) in our multivariate analysis, is particularly noteworthy. It underscores that the intrinsic locoregional invasiveness of the primary tumor, specifically gross extrathyroidal extension. This observation is corroborated by extant literature, which consistently identifies advanced T staging, as a predictor of diminished therapeutic response to RAI therapy (28, 29). The underlying pathophysiological basis for this association lies in the aggressive tumor biology and extensive local invasion characteristic of T3b-T4 lesions, factors which are known to compromise radioiodine avidity and efficacy. These findings are further supported by quantitative analyses, such as that by Wang et al. (30), which established a significant correlation between T4 staging and inferior survival metrics, including overall survival and disease-specific survival. The adverse impact of advanced T stage on prognosis is a consistent theme, as similarly demonstrated by Li et al. (31) in a cohort of stage IVB DTC patients. Clinically, the T4 stage in advanced DTC signifies a greater initial tumor burden and a more invasive phenotype, which collectively contribute to a higher risk of structural recurrence and poorer long-term outcomes. Consequently, these data collectively underscore the clinical imperative to consider more vigilant surveillance and the potential need for personalized, multi-modal therapeutic strategies in this high-risk patient subgroup.

Interestingly, while age, BRAF^V600E mutation status and initial ATA recurrence risk stratification (high vs. intermediate) were significant predictors on univariate analysis, they did not retain independence in the multivariate model. This finding, particularly for BRAF^V600E, can be interpreted in the context of established tumor biology and statistical mediation. The BRAF^V600E mutation is a well-documented driver of aggressive tumor behavior, including local invasion and metastatic potential (32, 33). While its significant univariate association with PD, its loss of independence in the multivariable model suggests that its prognostic effect is largely mediated by its correlation with more advanced T stage and higher sTg levels, which themselves emerged as dominant predictors. This suggests that within the BIR population, the anatomic extent of disease (T stage) and the quantitative biochemical burden (sTg) serve as more proximate and powerful integrators of risk, potentially capturing the downstream clinical consequences of the BRAF^V600E mutation. Furthermore, the sample size of our cohort may have been limited in detecting a modest independent effect of BRAF^V600E status after adjusting for these strong predictors. Therefore, while BRAF^V600E remains an important biological marker, its utility for independent risk stratification in BIR patients may be superseded by the combined assessment of T stage and sTg.

From a clinical nuclear medicine perspective, these findings have direct implications for practice. Patients presenting with BIR and either sTg ≥50.0 ng/mL or a primary T3b-T4 tumor constitute a high-risk subgroup warranting intensified surveillance. This could entail more frequent serial assessments of serum biomarkers (specifically Tg and TgAb), alongside anatomical imaging such as neck ultrasonography and chest CT. Furthermore, a lower threshold for employing advanced functional imaging such as 18F-FDG PET/CT to detect occult metastatic disease. In cases where clinical or biochemical suspicion persists, the feasibility of empiric repeat RAI therapy may warrant earlier discussion, particularly if there is a reasonable expectation of residual RAI-avid tissue. Conversely, BIR patients with sTg <50.0 ng/mL and lower T stage may be reasonably managed with standard-interval follow-up, avoiding unnecessary interventions given their higher probability of stable disease or spontaneous remission.

Several limitations of this study must be acknowledged. Its retrospective, single-center design inherently carries risks of selection and information bias. The exclusion of patients with TgAb positive, while methodologically necessary for accurate sTg interpretation, limits the generalizability of the sTg threshold to the approximately 20-25% of DTC patients with interfering antibodies. The administered RAI activities (100–150 mCi) followed a standardized protocol, but the lack of dosimetric assessment precludes correlation between delivered radiation dose to metastases and biochemical response. Furthermore, the median follow-up of just over 4 years, while informative, is still relatively short for a typically indolent malignancy like DTC; longer-term data are needed to confirm the stability of these predictive relationships and final disease-specific survival outcomes.

In conclusion, DTC patients exhibiting BIR after initial RAI therapy had a high risk of biochemical progression or structural disease recurrence and only a small proportion of patients reverted to ER after long-term follow-up. We propose that sTg ≥50.0 ng/mL and T3b-T4 disease be considered high-risk features within the BIR classification, prompting a more vigilant management pathway. These readily available parameters enhance the granularity of dynamic risk stratification, facilitating personalized follow-up strategies and timely therapeutic decisions.