
Background: Differentiated thyroid cancer (DTC) generally carries an excellent prognosis, yet recurrence occurs in up to 8–9% of patients. Artificial intelligence (AI) models are increasingly proposed to predict DTC recurrence; however, their methodological quality and clinical readiness are poorly defined. We conducted a systematic review to evaluate the methodological quality, risk of bias, and applicability of these models using the PROBAST + AI framework. Methods: We searched Ovid MEDLINE, Embase, Cochrane databases, Scopus, and Web of Science from inception to July 13, 2026. Eligible studies included original research developing or validating AI models for recurrence in DTC. Model quality (for development), risk of bias (for evaluation), and applicability were assessed using the PROBAST + AI framework, with a focus on discrimination, calibration, and validation strategies (PROSPERO; registration number [CRD420251004644]). Results: Of 673 records screened, 34 studies were included (21 unique datasets and 13 UCI-derived publications), encompassing 21,267 patients with a recurrence rate of 7.8%. Most were developed in Asian cohorts (17 of 21 datasets) and were structured clinical/tabular variables (82.4%), including demographic, clinical, laboratory, and pathological data, whereas the remaining models incorporated multimodal inputs such as imaging-derived or radiomic features. Notably, nearly one quarter of included studies (13 of 34, 38%) represented reanalysis of a single dataset. Discrimination was reported in 22 of 34 studies (64.7%) (area under the receiver operating characteristic curve [AUC] 0.67–1.00), though only 10 studies assessed calibration, none with standardized metrics. The PROBAST + AI assessment revealed high overall quality concerns in 26 of 34 (76.5%) development studies due to reliance on single-center cohorts, inadequate handling of missing data, and absence of external validation. Only 6 of 34 included studies (17.6%) were judged to have low overall quality concern, whereas 17 of 34 (50.0%) were rated as having low applicability concern. Conclusions: AI models for predicting DTC recurrence show promising discriminatory performance but have substantial methodological shortcomings in development, and a lack of calibration and external validation necessary for clinical use. Given the guidelines’ emphasis on risk-adapted de-escalation of therapy, the development of robust, validated, and clinically calibrated AI tools is a priority. Current evidence remains insufficient to support routine clinical use.
Background: Thyroid diseases can be evaluated using multiple modalities, including thyroid function tests, thyroid autoantibodies, ultrasonography, and 18 F-fluorodeoxyglucose-positron emission tomography (FDG-PET). However, comprehensive data integrating these modalities in large cohorts remain limited. We aimed to characterize the prevalence and patterns of abnormal findings on thyroid examinations in a medical checkup setting. Methods: This was a cross-sectional baseline analysis of a prospective cohort of adults who underwent their initial comprehensive medical checkups between April 1, 2016, and December 31, 2021, at the Preemptive Medicine and Lifestyle Disease Research Center, Kyoto University Hospital. All participants underwent thyroid function tests, ultrasonography, and FDG-PET; anti-thyroperoxidase antibody (TPOAb) titers were measured in a subset. Results: Among 4407 participants without known thyroid disease (2643 males and 1764 females), the prevalence of thyroid dysfunction, increased blood flow on ultrasonography, diffuse thyroid FDG uptake, and thyroid nodules was 5.8%, 2.5%, 3.4%, and 39.7%, respectively; all were more frequent in females. Among 2420 participants with TPOAb measurements, TPOAb positivity was 7.2%. On the Roche Elecsys platform, TPOAb titers ≥128 IU/mL were associated with concurrent thyroid dysfunction, increased blood flow, and diffuse thyroid FDG uptake. Multivariate analyses identified age, sex, and thyroid volume as major determinants of thyroid function. Using data from 1840 participants without any thyroid abnormalities, we established sex-specific reference ranges for thyroid dimensions and found their correlations with age and body size. Conclusions: This analysis provides epidemiological and physiological insights into thyroid health by integrating findings from thyroid function tests, ultrasonography, FDG-PET, and TPOAb measurements. Notably, higher TPOAb titers were associated with concurrent thyroid dysfunction and abnormalities, and age, sex, and thyroid volume were identified as major determinants of thyroid function. The rigorously defined healthy cohort enabled the establishment of sex-specific reference ranges for thyroid dimensions and confirmed associations between thyroid size, age, and body size. This study provides valuable evidence from a Japanese population, for which data have been limited to date. Furthermore, the present study highlights the associations among abnormal findings, relationships within thyroid physiology, and the concurrent associations of high-titer TPOAb with thyroid dysfunction and abnormalities.
Background: Pazopanib showed clinical activity in progressive radioiodine-refractory differentiated thyroid carcinoma (RAIR-DTC), although its use is limited by substantial toxicity not reduced by intermittent treatment. We aimed to identify biomarkers associated with outcome and treatment-related toxicity in these patients. Methods: This study reported exploratory results related to the Pazothyr trial (NCT01813136) investigating pazopanib efficacy in RAIR-DTC patients. We evaluated lymphopenia and neutrophil-to-lymphocyte ratio (NLR) from the Pazothyr clinical database ( n = 168). Dedicated informed consent for additional plasma and tumor sample collection allowed quantification of plasma angiogenic factors and cytokines at baseline and after one month of pazopanib ( n = 111). Immune markers (CD8, FOXP3, CD163, and PDL1) analyzed by immunohistochemistry in archived primary tumors from patients ( n = 68) with response or stable disease after 6 months of pazopanib were scored semiquantitatively. Associations with response, progression-free survival (PFS), overall survival (OS), and substantial treatment-related adverse events (TRAEs) were evaluated. Given the exploratory nature of this study, no adjustment for multiple comparisons was performed. Results: At baseline, moderate and severe lymphopenia were associated with shorter PFS, unlike high NLR. Elevated Vascular Endothelial Growth Factor A (VEGF-A), Interleukin 6 (IL-6), and CXCL8 correlated with poorer PFS. High VEGF-A, hepatocyte growth factor (HGF), IL-6, CXCL8, and CXCL9 were associated with shorter OS. X Early changes in placental growth factor (PlGF), HGF, interleukin 12 subunit p40, and CXCL10 were associated with substantial TRAEs. Multivariable analyses identified baseline severe lymphopenia (hazard ratio [HR] = 2.39 [confidence interval or CI 1.27–4.48], p = 0.007) and high CXCL8 (HR = 2.01 [CI 1.26–3.21], p = 0.004) as independent factors associated with progression. High baseline NLR (HR = 2.26 [CI 1.21–4.22], p = 0.01), HGF (HR = 2.37 [CI 1.33–4.20], p = 0.003), and CXCL8 (HR = 2.20 [CI 1.25–3.87], p = 0.006) were similarly associated with death. One-month increases in PlGF (odds ratio [OR] = 2.49 [CI 1.13–5.50], p = 0.024) and HGF (OR = 2.91 [CI 1.32–6.43], p = 0.008) were independently associated with substantial TRAEs. In patients with tumor samples, Programmed cell Death-Ligand 1 (PDL1)-immune cell score (ICS) >1% (HR = 2.05 [CI 1.15–3.66], p = 0.015) was also associated with progression. Conclusions: Baseline lymphopenia and inflammatory plasma biomarkers were associated with clinical outcome, whereas early changes in circulating factors were associated with substantial TRAEs. These exploratory results require prospective validation in RAIR-DTC patients treated with pazopanib or other VEGF receptor-targeting multikinase inhibitors. The relevance of PDL1-ICS expression for anti-PD1/PDL1-based therapies requires dedicated prospective trials.
Background: Thyroid cancer is the second most common cancer among individuals <40 years. The young diagnosis age and slow tumor growth suggest that early-life factors may contribute to risk. Methods: We conducted a population-based nested case–control study within a cohort of individuals <40 years old born in California, United States (1982–2017), using linked birth and cancer registry data. The study included 5670 patients diagnosed with their first primary thyroid cancer at age 0–39 years (pediatric: age 0–19; early-onset: age 20–39) and 283,500 birth-year-matched controls (1:50). Multivariable conditional logistic regression models estimated odds ratios (ORs) and 95% confidence intervals (CIs) for associations between birth and parental characteristics and thyroid cancer incidence overall and stratified by diagnosis age (0–19, 20–39 years), histological subtypes (papillary, follicular, medullary), and ethnicity (non-Hispanic White, Hispanic). Results: In the overall population, females had four times higher odds of thyroid cancer compared with males (OR: 4.00, 95% CI: 3.75–4.27), consistent with prior literature. Higher birth weight was associated with increased odds of thyroid cancer overall (OR per 500 g: 1.05, 95% CI: 1.02–1.08) and across age groups, histological subtypes, and ethnicities. Increased odds were also seen in relation to higher maternal educational attainment, higher parental ages, and lower neighborhood social vulnerability, factors linked to higher socioeconomic status. Hispanic ethnicity and maternal foreign-born status were associated with increased risk, factors sometimes correlated with greater social disadvantage. Inverse associations were observed for higher birth order and Black (vs. non-Hispanic White) individuals. Age-stratified analyses were similar except for social vulnerability, which was positively associated among pediatric cases (OR: 1.28; 95% CI: 1.09–1.50) but inversely associated among early-onset cases (OR: 0.83; 95% CI: 0.76–0.90). Results stratified by histological subtype were generally consistent with the overall results or null. Most risk factors, including maternal foreign-born status, remained in ethnicity-based strata. Conclusions: Multiple perinatal factors were associated with pediatric and early-onset thyroid cancer, underscoring the importance of the early-life time window. Consistent relationships between birth weight and thyroid cancer support a role for in utero growth-related processes. Associations with socioeconomic-related factors suggest competing mechanisms that could reflect differential detection, hormonal, or environmental factors, warranting further investigation.
Background: Papillary thyroid carcinoma (PTC) carries an overall excellent prognosis; however, up to 30% of patients experience locoregional persistent or recurrent disease. Management of central compartment recurrent/persistent disease remains challenging, as revision surgery is associated with increased morbidity. To date, no study has specifically evaluated the prognostic factors of outcomes after revision surgery that included central neck compartments in patients initially classified as low or intermediate risk of recurrence. Methods: We conducted a multicenter retrospective study of 86 patients who underwent reoperation for either recurrent or persistent central neck PTC in three French centers between 2000 and 2024. The primary endpoint was therapeutic response at the time of the patients’ last follow-up. Secondary endpoints included postoperative morbidity, mortality, and distant metastasis. Results: 49/86 patients (57%), with a median follow-up of 6 years, achieved a complete response (CR) after reoperation. Men aged 55 years and over with a recurrent/persistent central neck lesion >15 mm were associated with an incomplete response ( p = 0.044) in univariate analysis. Twenty-six of 86 (30%) patients developed a second structural recurrence after reoperation; a recurrence/persistence size >30 mm was associated with an increased risk of subsequent recurrence (hazard ratio: 3.68; [confidence interval: 1.34–10.08]; p = 0.011). Postoperative persistent complications included recurrent laryngeal nerve palsy in 5.8% of cases, half of which were due to oncologic reasons, and 3.5% resulted from permanent hypoparathyroidism. Conclusions: Reoperation for recurrent/persistent central neck disease in PTC leads to a CR in nearly two-thirds of patients but carries significant morbidity. Lesion size at the first reoperation appears to be a major prognostic factor. These findings support a cautious and individualized surgical approach to optimize outcomes while minimizing risks.