Development of a clinical prediction model to guide the selective use of SPECT/CT in papillary thyroid cancer

Introduction

The most prevalent endocrine system cancer is thyroid cancer, and its incidence is increasing worldwide (1,2). Papillary thyroid cancer (PTC) comprises the majority of differentiated thyroid cancers (DTCs) and is characterized by radioiodine uptake ability. After the initial thyroidectomy, radioiodine therapy can be performed to decrease recurrence and improve survival in patients with PTC, especially for those belonging to intermediate- or high-risk groups (3,4). It is also true for patients in the low-risk group to eliminate normal thyroid tissue remnant, thereby ensuring undetectable or minimal serum thyroglobulin (TG) levels (in the absence of neoplastic tissue), which facilitates follow-up (remnant ablation) (5). Radioiodine scintigraphy is a widely used, clinically applicable imaging modality for detecting metastatic DTC, and it is routinely performed as a whole-body scan (WBS) of the anterior and posterior projections with or without spot planar imaging (6,7). Although WBS is a very useful diagnostic procedure after radioiodine therapy, inherent weaknesses in the two-dimensional features of planar imaging are present, which are related to the lack of reliable anatomical markers and superimposition of tracer uptake in nearby anatomical structures (8-10). Thus, differentiating pathological uptake from physiological uptake using WBS alone may be difficult. However, the addition of single-photon emission computed tomography/computed tomography (SPECT/CT) scan to WBS can provide both three-dimensional (3D) radioiodine distribution and anatomical CT information, thus providing an accurate diagnosis of radioiodine uptake (11-15). Furthermore, many studies have suggested that SPECT/CT could improve diagnostic accuracy by detecting hidden lymph nodes (LNs) and distant metastases, accurately localizing metastases, and differentiating between physiologic and pathologic radioiodine uptake in DTC, including PTC (12-17). A recent meta-analysis revealed that the addition of SPECT/CT helped to accurately localize lesions and distinguish them from nonspecific uptake, leading to more definitive readings (8). Thus, almost all studies have recommended the addition of SPECT/CT to WBS to improve lesion localization, diagnostic performance, and therapeutic strategies for patients with thyroid cancer, including the 2025 American Thyroid Association (ATA) guidelines, the 2022 European Thyroid Association Consensus Statement, and the Joint Recommendation by the American and European Societies of Nuclear Medicine in 2022 (18-20).

Although previous findings have underscored the significant benefit of adding SPECT/CT to WBS for DTC patients (8,13,21,22), the clinical challenge remains how to selectively apply this imaging modality to patients most likely to benefit, while avoiding unnecessary radiation exposure, prolonged imaging time, increased cost and may lead to patient discomfort (23,24). Therefore, rather than identifying novel imaging biomarkers, the present study focuses on the practical integration of established variables into a predictive framework to guide individualized SPECT/CT utilization after WBS in DTC patients who underwent radioiodine therapy (Figure 1). To the best of our knowledge, this is the first study to provide a predictive model for reducing unnecessary exposure to SPECT/CT for patients with PTC. We present this article in accordance with the TRIPOD+AI reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1-2659/rc).

Figure 1 Diagram of the study and workflow visualization of the model. Patients with papillary thyroid cancer underwent routine postoperative radioiodine therapy and documented WBS and SPECT/CT imaging. Based on the diagnosis from the WBS and the addition of SPECT/CT, patients were divided into groups of patients with consistent and inconsistent diagnoses. At last, predictive modeling was generated using multifactorial clinical factors to determine whether patients needed additional SPECT/CT assistance. The predictive model will have a significant impact on early intervention and targeted treatment planning. Moreover, the collaboration between physicians and artificial intelligence could lead to more accurate and cost-effective cancer treatments in future studies. Created in BioRender.com. SPECT/CT, single-photon emission computed tomography/computed tomography; WBS, whole-body scan.

Methods

Ethical declaration

This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Kyungpook National University Hospital (IRB No. 2022-05-022-002). Informed consent was waived as the study was retrospective, and the analyses used anonymized clinical data.

Participants

In this retrospective study, 265 patients between May 2014 and December 2017 were included. The inclusion criteria were as follows: (I) patients who underwent radioiodine therapy after thyroidectomy; (II) patients above 18 years of age; and (III) pathologically confirmed diagnosis of PTC with complete surgical report and clinical serology record. Of the patients, 225 were female, and 40 were male (age range, 21–82 years), with a mean age of 49.38±12.51 years. WBS and SPECT/CT were performed in all patients 4–5 days after the administration of the therapeutic dose of radioiodine. All images were independently reviewed by a nuclear medicine physician. Clinical information was collected, including age, sex, histopathological findings, and serological data such as thyroid-stimulating hormone (TSH), TG, and anti-thyroglobulin (antiTG) levels. Primary tumor size was recorded from the final histopathological report. Tumor staging was performed according to the 8th edition of the American Joint Committee on Cancer (AJCC) staging system, which was the contemporary standard during the patient enrollment and data collection period for this study.

Protocol for radioiodine therapy

TSH levels (30 µIU/mL) increased following thyroid hormone withdrawal (THW) or injection of recombinant human TSH (rhTSH; Genzyme Corporation, Cambridge, MA, USA). For THW, levothyroxine (T4) was discontinued five weeks prior to radioiodine therapy. Patients were switched to liothyronine (T3) replacement for three weeks, which was then discontinued two weeks before therapy. Instead of THW, 0.9 mg rhTSH was injected intramuscularly for two consecutive days before radioiodine therapy. All patients were instructed to follow a low-iodine diet for 2 weeks before radioiodine therapy. Prior to the administration of radioiodine, serum TG and antiTG levels of the patients were assessed using TG immunoradiometric assay (IRMA) (Cis Bio International, Gif-sur-Yvette, France) and TG commercial radioimmunoassay kits (BRAHMS anti-Tg_n RIA, BRAHMS GmbH, Henningsdorf, Germany). At the same time, the TSH level was measured by an immunoradiometric commercial kit (TSH IRMA; BRAHMS, Hennigsdorf, Germany).

WBS and SPECT/CT acquisition

On day 4 (or day 5) after 1.85–7.4 gigabecquerel (GBq) I-131 administration, a dual-head gamma camera fitted with a 5/8-inch sodium iodide (NaI) crystal equipped with a high-energy collimator (NM670; General Electric Medical Systems, Milwaukee, WI, USA) was used to obtain the WBS using 364-keV light max at 20% range. WBS was performed at a scanning speed of 13.3 cm/min. The pixel exposure time was 180 seconds, and the field of view was 40 cm.

SPECT/CT was performed after WBS as the previous study protocol (25). The axial viewing field was 53.3 cm × 38.7 cm. A matrix size of 128×128 pixels was used for SPECT slice image acquisition, and 120 projections (30 s/projection) were acquired at 360°. SPECT data were replicated using 3D iterative algorithms (ordered-subsets expectation maximization with four iterations and eight subsets), and a Gaussian 3D spatial filter was utilized to smooth out the image. Immediately after SPECT, spiral CT was conducted in a volume session close to that used during SPECT. CT was performed at a tube voltage of 120 kV, altering the tube current according to body weight. The restored image had a width and spacing of 3.75 mm. CT data were used for attenuation correction and anatomical details. During this procedure, no contrast agent was administered. Xeleris software (General Electric Medical Systems, Milwaukee, WI, USA) was used to generate WBS and SPECT/CT images. The program provided axial, sagittal, and coronal SPECT, CT, and fusion SPECT/CT images. In addition, when contamination (e.g., urine, saliva) is suspected on WBS, the affected area should be promptly washed and reimaged for verification of contamination. Lateral or oblique planar scan can be helpful to discriminate the contamination by its superficial localization. SPECT/CT application is not frequently required to recognize contaminations.

Image interpretation

WBS and SPECT/CT images were reviewed by an experienced nuclear medicine physician. Lesions with uptake above the surrounding background were interpreted as positive foci. WBS was interpreted as follows: (I) positive foci located in the thyroid bed or the median region of the neck corresponding to the thyroglossal duct were considered residual lesions. At the same time, the positive foci located in the nasal cavity, salivary glands, gastrointestinal tract, etc., were considered as physiologic uptake lesions. Both residual lesions and physiologic uptake were categorized in the non-metastatic group. (II) Positive foci in the lateral neck were considered as LNs metastasis. (III) Positive foci outside the neck (such as lung and bone, except the physiologic uptake) were considered for distant metastasis. SPECT/CT was considered as follows: (I) the patient was grouped in the LNs metastasis group when the positive foci located at LNs on CT; (II) grouped in the distant metastasis when the positive foci located at corresponding non-LN lesions on CT; (III) positive foci localized in the thyroid bed, a tract of the thyroglossal duct; or physiologic iodine uptake or retention organs (such as salivary gland, nasal mucosa, esophagus, intestine, etc.) or non-malignant looking CT image (skin infection, etc.) were categorized as non-metastasis. Finally, true-negative, false-negative, true-positive, and false-positive subgroups were determined based on the diagnosis by WBS alone and by WBS with SPECT/CT (Figure 2A). Confirmation of positive or negative findings was based on either histopathological result, when available, or a composite clinical reference standard. This clinical reference standard included longitudinal serum TG trends (with consideration of antiTG antibody levels) and concordant findings on follow-up imaging studies, such as ultrasonography (USG), CT, magnetic resonance imaging (MRI), and 18F-fluorodeoxyglucose (FDG)-positron emission tomography (PET)/CT. This approach reflects routine clinical practice after radioiodine therapy, in which pathological confirmation is not feasible for suspected lesions. To determine which patients could obtain diagnostic benefits through the addition of SPECT/CT, the clinicopathological parameters between the true-/false-negative subgroups in non-metastasis/metastasis groups were compared.

Figure 2 Classification of diagnosis by WBS and reclassification through the addition of SPECT/CT on a patient basis. (A) Classification and reclassification of patients based on WBS alone versus WBS with SPECT/CT. Patients were initially classified as non-metastasis or metastasis by WBS. After addition of SPECT/CT, patients in the non-metastasis group were further divided into true-negative and false-negative subgroups, while patients in the metastasis group were divided into true-positive and false-positive subgroups. The consistent group comprised patients with concordant diagnoses between WBS and WBS + SPECT/CT, and the inconsistent group comprised those with discordant diagnoses. Finally, a predictive model was established based on the LASSO logistic regression and clinical factors. (B) Changes in the diagnostic results in patient case-based imaging with the addition of SPECT/CT are summarized. Horizontal lines with numbers between WBS and SPECT/CT indicate consistent groups and diagonal lines indicate patients with inconsistent diagnoses. LASSO, least absolute shrinkage and selection operator; LN, lymph node; SPECT/CT, single-photon emission computed tomography/computed tomography; WBS, whole-body scan.

Statistical analysis

Statistical analysis was conducted using IBM SPSS Statistics 25.0 and R Studio (version 4.0.2 at www.r-project.org) as follows:

• Continuous measures are reported as mean ± standard deviation (SD), and categorical measures as n (%). The Kolmogorov-Smirnov test was used to determine the normal distribution of the data. Student’s t-test was conducted to analyze differences between groups if normality was assumed; otherwise, the Mann-Whitney U-test was performed. Chi-squared analysis was used for categorical data, and statistical significance was set at P<0.05.
• LASSO regression was applied to select optimal predictive variables for SPECT/CT diagnostic utility, with a regularization parameter (λ) chosen via 10-fold cross-validation to address variable collinearity. Moreover, multivariate logistic regression was performed to screen for statistically significant variables, and the odds ratio (OR) and 95% confidence interval (CI) were calculated for each variable. The related variables of P<0.05 were used to build a LASSO logistic predictive model (LL-model). Additionally, clinical factors related to inconsistent diagnostic imaging were also used to develop the clinical predictive model (Clin-model). Finally, the nomogram and receiver operating characteristic (ROC) curve were built, and the area under the curve (AUC) was calculated to evaluate the efficacy of the models in predicting the patient requirements of SPECT/CT imaging.

Results

Clinical characteristics and pathological results of patients in true- and false-negative subgroups of the non-metastases group

WBS revealed 232 patients as non-metastasis, out of which 215 were true negative and 17 were classified as LN/distant metastasis, that is, a false-negative subgroup after adding the SPECT/CT imaging. Comparing the clinicopathological characteristics of the two subgroups, we found that the tumor size in the false-negative group was 2.51±2.13 cm, whereas in the true-negative group, it was 1.46±1.24 cm (P<0.05). Additionally, the TG level of the false-negative group was 48.86±133.76 ng/mL, which was significantly higher than that of the true-negative group (3.99±25.43 ng/mL, P<0.05). Moreover, the age, proportion of males, T stage, M stage, and stage of AJCC were higher in the false-negative subgroup than in the true-negative subgroup but without statistical significance (Table 1).

Clinical characteristics and pathological results of patients in the true- and false-positive subgroups of the metastases group

WBS diagnosed 33 patients as the metastasis group, of which 22 were confirmed to have a true-positive diagnosis after the addition of SPECT/CT. Another 11 were determined by SPECT/CT as false positive, that is, patients without metastases. By comparing the clinicopathologic features of the two subgroups, patients in the false-positive group had a lower AJCC staging than those in the true-positive group (stage I: 54.5% vs. 13.6%), and P<0.05. Except for the AJCC stage, age, percentage of intrathyroidal invasion, T stage, N stage, M stage (from preoperative scan), and TG level were lower in the false-positive subgroup but were not statistically significant (Table 2).

Reclassification of diagnosis by the addition of SPECT/CT on a patient basis and analysis of clinical factors that may be applied in the Clin-model

By comparing the diagnostic results of the WBS and WBS with SPECT/CT, consistent results were obtained in 237 out of 265 patients, and discordant results were obtained in 28 (Figure 2A). When only WBS was used as the diagnostic image, 232 patients were identified as non-metastasis, 19 patients as LN metastasis, and 14 patients as distant metastasis. Figure 2B demonstrates the change in diagnostic results with the addition of SPECT/CT to WBS. Compared with imaging interpreted by WBS alone, the addition of SPECT/CT reclassified 14 and three patients from the non-metastasis group to the LN metastasis group or distant metastasis, respectively. Four patients diagnosed with LN metastasis by WBS were recategorized as non-metastasis by the addition of SPECT/CT (Figure 3). Moreover, CT of the SPECT/CT detected three patients with pulmonary metastatic lesions in the non-metastasis group diagnosed with WBS (Figure 4). Meanwhile, seven patients identified as distant metastasis by WBS were recategorized as benign pathologic uptake through the addition of SPECT/CT, such as cutaneous inflammatory lesions in Figure 5.

Figure 3 Precise interpretation of the residual thyroid tissue by SPECT/CT. Multiple cervical tracer uptakes (four positive foci were marked by different colored arrows) suspected as lymph node metastases on the whole-body scan were diagnosed as uptakes by residual thyroid tissue through SPECT/CT. (A) Anterior planar image. (B) Coronal SPECT/CT fusion image. (C-E) Transverse fusion SPECT/CT images. SPECT/CT, single-photon emission computed tomography/computed tomography.

Figure 4 SPECT/CT interpretation of the metastatic pulmonary nodule. WBS and SPECT images demonstrated negative results; however, CT showed a metastatic pulmonary nodule (red arrows). (A,B) Anterior and posterior WBS. (C) Transverse SPECT/CT fusion image. (D) CT image. CT, computed tomography; SPECT/CT, single-photon emission computed tomography/computed tomography; WBS, whole-body scan.

Figure 5 SPECT/CT reclassification of benign pathologic uptake. Tracer uptake suspected as pulmonary or osseous metastasis on the whole-body scan (red arrow in A) was diagnosed as benign pathologic uptake by cutaneous inflammatory lesions (red arrow in B). (A) Posterior planar image. (B) Upper back showing dermatitis. (C) Transverse fusion SPECT image. (D) CT image. CT, computed tomography; SPECT/CT, single-photon emission computed tomography/computed tomography.

Furthermore, the number of patients with a tumor size of >2 cm was 46 (19.4%) in the consistent group and 10 (35.7%) in the inconsistent group, which demonstrated a statistical difference. Also, TG levels were significantly higher in the inconsistent group compared to the consistent group. The results implied that SPECT/CT diagnosis should be added in patients with bigger tumors (P<0.005) and those with high TG levels (P<0.05). Moreover, 32 (13.5%) men were in the consistent group, whereas eight men (28.6%) were in the inconsistent group. Although a statistical difference was not reached, the percentage of males in the inconsistent group was higher than that in the consistent group. No statistically significant differences were found in the remaining factors, including age, degree of tumor invasion, pathological factors, TSH, and antiTG levels (Table 3).

Predictive model construction and evaluation

The LASSO logistic regression model (LL-model) was employed to examine the relationship between 14 variables, including age, sex, invasion stage, tumor-node-metastasis (TNM) stage, AJCC stage, tumor size, LN metastasis number/ratio, TSH, TG, antiTG, and WBS diagnosis result. After the LASSO analysis, five variables (invasion stage, age, LN metastasis number/ratio, and TSH) that affected the fitting were excluded (Figure 6). Four of the nine variables were identified as powerful factors by logistic regression, and the results are visualized in Table 4. Predictive models were established using two methods: (I) LASSO logistic model (LL-model): the LASSO logistic regression analysis with TG and WBS; and (II) Clin-model: TG, tumor size, and sex are clinically related factors with WBS. The nomogram prediction model was constructed based on these factors (Figure 7A). We then modeled the data and tested the specificity and sensitivity of the predictive model with ROC curves (Figure 7B). The results showed that the AUCs of the two models were as follows: 0.855 (95% CI: 0.778–0.932) and 0.814 (95% CI: 0.725–0.902), indicating a promising grade.

Figure 6 LASSO regression analysis for variable selection. (A) LASSO coefficient profiles for 14 properties: age, sex, invasion stage, TNM stage, AJCC stage, tumor size, LN metastasis number/ratio, TSH, TG, antiTG, and WBS diagnosis. Coefficient profiles were plotted based on the log λ (λ) series. (B) Plotting the relationship between biased likelihood bias (binomial bias) and log(λ). AJCC, American Joint Committee on Cancer; LASSO, least absolute shrinkage and selection operator; LN, lymph node; TG, thyroglobulin; TNM, tumor-node-metastasis; TSH, thyroid-stimulating hormone; WBS, whole-body scan.

Figure 7 Performance evaluation of the predictive models. (A) Nomogram for the predictive Clin-model. In the nomogram prediction model, a vertical line was drawn from the axis corresponding to each factor until the top line labeled “points” was reached. The points were then summed up and then a line was drawn down from the axis labeled “total points” intercepted to predict the probability of SPECT/CT necessity. (B) Receiver operating characteristic curve of the predicted models. AUC, area under the curve; CI, confidence interval; SPECT/CT, single-photon emission computed tomography/computed tomography; TG, thyroglobulin; WBS, whole-body scan.

Discussion

In this study, patients with PTC who underwent both WBS and SPECT/CT after their first radioiodine therapy were included. Furthermore, clinicopathological characteristics were used for selecting patients who may need SPECT/CT imaging to accurately determine disease status. Finally, predictive models of the need for additional SPECT/CT based on the LL-model and related clinical factors were established. Our findings fill an important gap in the medical literature and overcome some limitations of previous publications (8,26). In a recent meta-analysis pooling 30 studies published between 2004 and 2022, patients with thyroid cancer with SPECT/CT were found to have increased odds of scan-based conclusive reading rate [relative risk (RR): 1.34; 95% CI: 1.29–1.39] and lesion-based conclusive reading rate (RR: 1.18; 95% CI: 1.16–1.21) than WBS. Additionally, this study mentioned that the quality of evidence and strength of recommendations were low. In this case, identifying the targeted patients and adapting existing protocols to reduce patient exposure is a breakthrough in rationalizing the addition of SPECT/CT to WBS. Every study has reported the benefits of adding SPECT/CT as an adjuvant to WBS; however, few studies have emphasized the side effects of adding SPECT/CT to patients. It is noteworthy that most current studies have focused on adults, with little attention paid to the fact that pediatric patients are more sensitive to radioiodine therapy. While studies from a decade ago emphasized the importance of personalized dose optimization in SPECT/CT imaging to prevent patient radiation overexposure (27), this critical area remains under investigated.

Currently, serum TG is commonly used to detect persistent or recurrent diseases in the clinical follow-up of patients, and it is also very easy to measure (28-31). Moreover, a previous study demonstrated that additional SPECT/CT is warranted in patients diagnosed with residual thyroid tissue by WBS but with high serum TG levels (7,32). Other researchers have also suggested that various imaging modalities, that is, neck USG (28), thorax CT scan (32,33), MRI (32,34), and 18F-FDG-PET/CT (28,35,36), are helpful in patients with DTC and high serum TG levels in detecting iodine nonavid persistent disease. This study also found that the serum TG level of the false-negative subgroup was statistically higher than that in the true-negative subgroup in the non-metastasis group, which indicated the high TG level could be a predictive factor for additional SPECT/CT imaging in patients with PTC. While except for the TG level, the tumor size in the true-/false-negative subgroups also showed statistical differences. Patients with larger tumors are more likely to develop metastasis, and the high sensitivity of SPECT/CT imaging allows for better detection of hidden metastatic lesions (21). In addition, sexual dimorphism strongly influences cancer biology, with most tumors being male-dominated. Uniquely, thyroid cancer is the only non-reproductive cancer that is predominantly female, although men usually have more aggressive disease (37). Until now, most of the previous studies simply analyzed one feature without the incorporation of other pathological and clinical factors (38,39). While two prediction models were established in this study based on the variables of patient characteristics, such as TG, tumor size, sex, and WBS. Although some of the clinical-related factors did not show an independent role in the univariate analysis, all relevant variables were selected using LASSO analysis, which supports the relevance and accuracy of the subsequent logistics model. Finally, we confirmed that the performance evaluation of the models was optimistic using the nomogram and AUC analysis. Encouragingly, the Clin-model showed better predictive results than the LL-model, suggesting that the clinical multifactorial predictive model has great potential to influence the SPECT/CT diagnosis of PTC patients.

The current guideline recommends that post-treatment WBS (with or without SPECT/CT) be performed after radioiodine residual ablation or treatment of patients with thyroid cancer (4). However, given the radiation exposure, long imaging time, and cost-related issues, a better plan for the addition of SPECT/CT becomes an urgent need. Even clinicians have pointed out that the widespread use of imaging methods for screening has increased the incidence of thyroid cancer, resulting in overdiagnosis and overtreatment as early as 5 years ago (40). To date, only two studies have tested the radiation effects of CT on patients where SPECT/CT is routinely performed. According to a 2016 study, radiation exposure from the CT portion of SPECT/CT ranged from 0.01 to 7.4 millisievert (mSv), depending on the imaging site, the exam protocol, and the weight and age of patients (41). Another 2012 study also showed that the effective per-patient CT dose varied between 0.06 and 11.9 mSv and highlighted the need to appropriately use and optimize SPECT/CT protocols to maximize benefits to patients (26). Although previous studies partly supported our study, they elucidated that adding SPECT/CT could increase the diagnostic accuracy of patients and improve the treatment plan (11,13,16,17,22,42); however, in the previous studies, predicting what type of patients is more suitable for the addition of SPECT/CT examination that is key to ensuring personal treatment was not mentioned. For example, in 11 false-positive patients, SPECT/CT helps differentiate causes of increased uptake, including contamination, benign conditions (e.g., bronchiectasis or lung infections) and artifacts. By accurately identifying the source of uptake, SPECT/CT reduces unnecessary interventions—such as repeat radioiodine ablation or procedures prompted by false-positive WBS. With the predictive models established by this study, the clinicians could separate patients who can benefit from the addition of SPECT/CT from patients who get additional radiation without diagnostic benefit.

From a clinical perspective, the proposed nomogram provides an intuitive tool that may assist physicians in individualized decision-making regarding the addition of SPECT/CT. By estimating the probability of diagnostic discordance between WBS and SPECT/CT, the model will help physicians determine which patients can obtain a valuable diagnostic conclusion by the addition of SPECT/CT. For example, for the patients in the false-negative group, the WBS imaging only was unable to find metastases that already existed, which means that the clinician may lose the optimal time to treat the hidden lesions, resulting in the progression of the disease or even the death of the patient. Therefore, SPECT/CT imaging can provide the most valuable diagnostic advice by visualizing hidden lesions for these patients and allowing patients to receive timely treatment. For the false-positive group, a portion of the patients could be scheduled for more diagnostic studies, such as MRI and USG, which consume more time and money. Worse still, another group of patients may receive the 2nd radioiodine treatment, suffering overtreatment. Personalized medicine led by the predictive models not only reduces unnecessary tests but also leads to better cost and time savings for patients. For example, the male patients, those with larger tumor sizes, and patients with elevated TG levels may benefit further from enhanced diagnostic results using additional imaging of SPECT/CT. In addition, to reduce false positives in SPECT/CT, efforts focus on mitigating physiological uptake and motion artifacts. Key measures include meticulous patient preparation (e.g., a low-iodine diet, adequate hydration, and TSH stimulation), delayed imaging to differentiate pathological from physiological uptake, and immobilization techniques to minimize motion blurring. Even, the use of predictive modeling in oncology is prevalent, most of the previous studies have focused on disease screening (43), recurrence (44), and survival (45). However few studies applied the predictive model in a diagnostic imaging methodology for patients with thyroid cancer (46-48), which provides a novel direction for cancer diagnosis.

Still, this study has several limitations, including being a retrospective single-center study, the relative smallness of the population size, and imaging interpretation by one nuclear medicine physician. In particular, the number of patients with discordant findings between WBS and SPECT/CT was limited, which may affect model stability and increase the risk of overfitting. To mitigate this concern, LASSO regression with 10-fold cross-validation was applied to reduce model complexity and select the most robust predictors. Nevertheless, the absence of external validation restricts the generalizability of the proposed models. Therefore, the current results should be interpreted as exploratory and hypothesis-generating. In addition, although explicit decision thresholds were not predefined in this retrospective study, such thresholds could be prospectively calibrated according to clinician and patient priorities, such as minimizing false-negative findings or reducing unnecessary imaging. Future multicenter studies with larger and more diverse cohorts are warranted to externally validate, determine optimal cutoff values and to evaluate the impact of model-guided SPECT/CT utilization on clinical outcomes. Also, antiTG levels may interfere with the accuracy of serum TG measurements, potentially confounding the interpretation of TG-based risk stratification and model performance. The antiTG levels were recorded, while their modifying effect on TG reliability was not explicitly modeled in this study. Future studies incorporating antibody-adjusted TG measurements are warranted to address this limitation. In addition, longitudinal studies with different populations to evaluate the long-term impact of the predictive models on the prognosis of patients should be considered. The PTCs can have various oncogenic mutations, such as RET/PTC rearrangements, BRAF, RAS, and TERT promoter mutations (49), which are associated with radioiodine-refractory disease development (50). A critical clinical challenge arises in managing radioiodine-refractory metastases that show no appreciable radioiodine avidity on imaging. So, predictive models considering the oncogenic mutations are needed in future studies. Despite these limitations, our study is the onlyto date that has constructed a predictive model to select the additional value of SPECT/CT and evaluated the models in patients with PTC.

Conclusions

This study demonstrated that a multifactorial predictive model integrating TG levels, tumor size, sex, and WBS findings can identify patients with PTC who are more likely to benefit from additional SPECT/CT imaging. By enabling selective rather than routine use of SPECT/CT, the proposed models may help optimize diagnostic accuracy while reducing unnecessary radiation exposure, imaging time, and cost. These findings support a more individualized imaging strategy, pending prospective multicenter validation.

Acknowledgments

We are very grateful to all the participants in this study and wish them all good health.

Footnote

Reporting Checklist: The authors have completed the TRIPOD+AI reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1-2659/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1-2659/dss

Funding: This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (grant/award No. NRF-2022R1A2C2005057). The funder had no role in the study design, data collection, analysis, interpretation, or preparation of the manuscript.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1-2659/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Kyungpook National University Hospital (IRB No. 2022-05-022-002). Informed consent was waived as the study was retrospective, and the analyses used anonymized clinical data.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Lim H, Devesa SS, Sosa JA, Check D, Kitahara CM. Trends in Thyroid Cancer Incidence and Mortality in the United States, 1974-2013. JAMA 2017;317:1338-48. [Crossref] [PubMed]
2. Song B, Lin Z, Feng C, Zhao X, Teng W. Global research landscape and trends of papillary thyroid cancer therapy: a bibliometric analysis. Front Endocrinol (Lausanne) 2023;14:1252389. [Crossref] [PubMed]
3. Schlumberger M, Leboulleux S. Current practice in patients with differentiated thyroid cancer. Nat Rev Endocrinol 2021;17:176-88. [Crossref] [PubMed]
4. Haugen BR, Alexander EK, Bible KC, Doherty GM, Mandel SJ, Nikiforov YE, Pacini F, Randolph GW, Sawka AM, Schlumberger M, Schuff KG, Sherman SI, Sosa JA, Steward DL, Tuttle RM, Wartofsky L. 2015 American Thyroid Association Management Guidelines for Adult Patients with Thyroid Nodules and Differentiated Thyroid Cancer: The American Thyroid Association Guidelines Task Force on Thyroid Nodules and Differentiated Thyroid Cancer. Thyroid 2016;26:1-133. [Crossref] [PubMed]
5. Nguyen NC, Anigati EM, Desai NB, Öz OK. Radioactive Iodine Therapy in Differentiated Thyroid Cancer: An Update on Dose Recommendations and Risk of Secondary Primary Malignancies. Semin Nucl Med 2024;54:488-96. [Crossref] [PubMed]
6. Ahn BC. Personalized Medicine Based on Theranostic Radioiodine Molecular Imaging for Differentiated Thyroid Cancer. Biomed Res Int 2016;2016:1680464. [Crossref] [PubMed]
7. Jeong SY, Lee SW, Kim HW, Song BI, Ahn BC, Lee J. Clinical applications of SPECT/CT after first I-131 ablation in patients with differentiated thyroid cancer. Clin Endocrinol (Oxf) 2014;81:445-51. [Crossref] [PubMed]
8. Chong A, Seo Y, Bang JI, Park S, Kim K, Hong CM, Choi M, Oh SW, Lee SW. Clinical Implications of Adding SPECT/CT to Radioiodine Whole-Body Scan in Patients With Differentiated Thyroid Cancer: A Systematic Review and Meta-analysis. Clin Nucl Med 2024;49:215-25. [Crossref] [PubMed]
9. Alkhybari EM, Albeshan SM, Alanazi BM, Alqahtani DM, Abokhater HK, Albakhiti SH, Ghanem IE. The diagnostic incremental value of 131I SPECT-CT scan compared to planar 131I WBS for differentiated thyroid carcinoma: A single institutional experience. Journal of Radiation Research and Applied Sciences. 2021;14:440-8.
10. Sergieva S, Sofiyanski T, Robev B, Dimcheva M, Fakirova A. Retrospective study of post-treatment (131)I SPECT/CT imaging for personalizing therapy of patients with differentiated thyroid cancer. Folia Med (Plovdiv) 2024;66:776-86. [Crossref] [PubMed]
11. Spanu A, Nuvoli S, Gelo I, Mele L, Piras B, Madeddu G. Role of Diagnostic (131)I SPECT/CT in Long-Term Follow-up of Patients with Papillary Thyroid Microcarcinoma. J Nucl Med 2018;59:1510-5. [Crossref] [PubMed]
12. Bashank N, Farghaly H, Hassanein S, Abdel-Tawab M, Wahman M, Mahmoud H. Rare sites of metastases in patients with differentiated thyroid carcinoma and added value of SPECT/CT over planar whole body radioactive iodine scan. Eur J Hybrid Imaging 2022;6:34. [Crossref] [PubMed]
13. Malamitsi JV, Koutsikos JT, Giourgouli SI, Zachaki SF, Pipikos TA, Vlachou FJ, Prassopoulos VK. I-131 Postablation SPECT/CT Predicts Relapse of Papillary Thyroid Carcinoma more Accurately than Whole Body Scan. In Vivo 2019;33:2255-63. [Crossref] [PubMed]
14. Oh JR, Ahn BC. False-positive uptake on radioiodine whole-body scintigraphy: physiologic and pathologic variants unrelated to thyroid cancer. Am J Nucl Med Mol Imaging 2012;2:362-85.
15. Jia X, Wang Y, Yang L, Fan K, Tao R, Liu H, Yao X, Yang A, Zhang G, Gao R. Lymph Node Metastases Identified at the Post-Ablation 131I SPECT/CT Scan Is a Prognostic Factor of Intermediate-Risk Papillary Thyroid Cancer. Diagnostics (Basel) 2022;12:1254. [Crossref] [PubMed]
16. Spanu A, Solinas ME, Chessa F, Sanna D, Nuvoli S, Madeddu G. 131I SPECT/CT in the follow-up of differentiated thyroid carcinoma: incremental value versus planar imaging. J Nucl Med 2009;50:184-90. [Crossref] [PubMed]
17. Aide N, Heutte N, Rame JP, Rousseau E, Loiseau C, Henry-Amar M, Bardet S. Clinical relevance of single-photon emission computed tomography/computed tomography of the neck and thorax in postablation (131)I scintigraphy for thyroid cancer. J Clin Endocrinol Metab 2009;94:2075-84. [Crossref] [PubMed]
18. Pacini F, Fuhrer D, Elisei R, Handkiewicz-Junak D, Leboulleux S, Luster M, Schlumberger M, Smit JW. 2022 ETA Consensus Statement: What are the indications for post-surgical radioiodine therapy in differentiated thyroid cancer? Eur Thyroid J 2022;11:e210046. [Crossref] [PubMed]
19. Avram AM, Giovanella L, Greenspan B, Lawson SA, Luster M, Van Nostrand D, Peacock JG, Ovčariček PP, Silberstein E, Tulchinsky M, Verburg FA, Vrachimis A. SNMMI Procedure Standard/EANM Practice Guideline for Nuclear Medicine Evaluation and Therapy of Differentiated Thyroid Cancer: Abbreviated Version. J Nucl Med 2022;63:15N-35N.
20. Ringel MD, Sosa JA, Baloch Z, Bischoff L, Bloom G, Brent GA, Brock PL, Chou R, Flavell RR, Goldner W, Grubbs EG, Haymart M, Larson SM, Leung AM, Osborne JR, Ridge JA, Robinson B, Steward DL, Tufano RP, Wirth LJ. 2025 American Thyroid Association Management Guidelines for Adult Patients with Differentiated Thyroid Cancer. Thyroid 2025;35:841-985. [Crossref] [PubMed]
21. Spanu A, Nuvoli S, Marongiu A, Gelo I, Mele L, De Vito A, Rondini M, Madeddu G. The Diagnostic Usefulness of (131)I-SPECT/CT at Both Radioiodine Ablation and during Long-Term Follow-Up in Patients Thyroidectomized for Differentiated Thyroid Carcinoma: Analysis of Tissue Risk Factors Ascertained at Surgery and Correlated with Metastasis Appearance. Diagnostics (Basel) 2021;11:1504. [Crossref] [PubMed]
22. Zilioli V, Peli A, Panarotto MB, Magri G, Alkraisheh A, Wiefels C, Rodella C, Giubbini R. Differentiated thyroid carcinoma: Incremental diagnostic value of (131)I SPECT/CT over planar whole body scan after radioiodine therapy. Endocrine 2017;56:551-9. [Crossref] [PubMed]
23. Ahmadi S, Coleman A, de Morais NS, Landa I, Pappa T, Kang A, Kim MI, Marqusee E, Alexander EK. Clinical experience following implementation of routine SPECT-CT imaging following 131-iodine administration for thyroid cancer. Endocr Connect 2022;11:e210371. [Crossref] [PubMed]
24. Mihailović J. Pre-Treatment and Post-Treatment I-131 Imaging in Differentiated Thyroid Carcinoma. J Clin Med 2024;13:1984. [Crossref] [PubMed]
25. Hong CM, Kim CY, Son SH, Jung JH, Lee CH, Jeong JH, Jeong SY, Lee SW, Lee J, Ahn BC. I-131 biokinetics of remnant normal thyroid tissue and residual thyroid cancer in patients with differentiated thyroid cancer: comparison between recombinant human TSH administration and thyroid hormone withdrawal. Ann Nucl Med 2017;31:582-9. [Crossref] [PubMed]
26. Sharma P, Sharma S, Ballal S, Bal C, Malhotra A, Kumar R. SPECT-CT in routine clinical practice: increase in patient radiation dose compared with SPECT alone. Nucl Med Commun 2012;33:926-32. [Crossref] [PubMed]
27. Nadel HR. SPECT/CT in pediatric patient management. Eur J Nucl Med Mol Imaging 2014;41:S104-14. [Crossref] [PubMed]
28. Vijayan R, Palaniswamy SS, Vadayath UM, Nair V, Kumar H. Clinicopathological features and outcome of thyroglobulin elevation and negative iodine scintigraphy (TENIS) patients with negative neck ultrasound: Experience from a thyroid carcinoma clinic in India. World J Nucl Med 2021;20:361-8. [Crossref] [PubMed]
29. Fatourechi V, Hay ID. Treating the patient with differentiated thyroid cancer with thyroglobulin-positive iodine-131 diagnostic scan-negative metastases: including comments on the role of serum thyroglobulin monitoring in tumor surveillance. Semin Nucl Med 2000;30:107-14. [Crossref] [PubMed]
30. Cho JS, Kim HK. Thyroglobulin Levels as a Predictor of Papillary Cancer Recurrence After Thyroid Lobectomy. Anticancer Res 2022;42:5619-27. [Crossref] [PubMed]
31. Bellini P, Dondi F, Gatta E, Zilioli V, Albano D, Cappelli C, Bertagna F. Prognostic role and characteristics of the indeterminate response in differentiated thyroid cancer: a systematic review. Endocrine 2024;84:812-21. [Crossref] [PubMed]
32. Ciappuccini R, Heutte N, Trzepla G, Rame JP, Vaur D, Aide N, Bardet S. Postablation (131)I scintigraphy with neck and thorax SPECT-CT and stimulated serum thyroglobulin level predict the outcome of patients with differentiated thyroid cancer. Eur J Endocrinol 2011;164:961-9. [Crossref] [PubMed]
33. Gao R, Zhang GJ, Wang YB, Liu Y, Wang F, Jia X, Liang YQ, Yang AM. Clinical Value of (99m)Tc-3PRGD2 SPECT/CT in Differentiated Thyroid Carcinoma with Negative (131)I Whole-Body Scan and Elevated Thyroglobulin Level. Sci Rep 2018;8:473. [Crossref] [PubMed]
34. Piscopo L, Nappi C, Volpe F, Romeo V, Nicolai E, Gallicchio R, Giordano A, Storto G, Pace L, Cavaliere C, Salvatore M, Cuocolo A, Klain M. Prognostic Value of Hybrid PET/MR Imaging in Patients with Differentiated Thyroid Cancer. Cancers (Basel) 2022;14:2958. [Crossref] [PubMed]
35. Leboulleux S, El Bez I, Borget I, Elleuch M, Déandreis D, Al Ghuzlan A, Chougnet C, Bidault F, Mirghani H, Lumbroso J, Hartl D, Baudin E, Schlumberger M. Postradioiodine treatment whole-body scan in the era of 18-fluorodeoxyglucose positron emission tomography for differentiated thyroid carcinoma with elevated serum thyroglobulin levels. Thyroid 2012;22:832-8. [Crossref] [PubMed]
36. Ozkan E, Aras G, Kucuk NO. Correlation of 18F-FDG PET/CT findings with histopathological results in differentiated thyroid cancer patients who have increased thyroglobulin or antithyroglobulin antibody levels and negative 131I whole-body scan results. Clin Nucl Med 2013;38:326-31. [Crossref] [PubMed]
37. Shobab L, Burman KD, Wartofsky L. Sex Differences in Differentiated Thyroid Cancer. Thyroid 2022;32:224-35. [Crossref] [PubMed]
38. Son J, Hong CM, Kim DH, Jeong SY, Lee SW, Lee J, Ahn BC. Comparison of Prognostic Value Between Stimulated and Nonstimulated Thyroglobulins in Differentiated Thyroid Cancer: A Retrospective Study. Nucl Med Mol Imaging 2023;57:257-64. [Crossref] [PubMed]
39. Lee CH, Jung JH, Son SH, Hong CM, Jeong JH, Jeong SY, Lee SW, Lee J, Ahn BC. Risk factors for radioactive iodine-avid metastatic lymph nodes on post I-131 ablation SPECT/CT in low- or intermediate-risk groups of papillary thyroid cancer. PLoS One 2018;13:e0202644. [Crossref] [PubMed]
40. Li X, Zhang S, Zhang Q, Wei X, Pan Y, Zhao J, et al. Diagnosis of thyroid cancer using deep convolutional neural network models applied to sonographic images: a retrospective, multicohort, diagnostic study. Lancet Oncol 2019;20:193-201. [Crossref] [PubMed]
41. Rausch I, Füchsel FG, Kuderer C, Hentschel M, Beyer T. Radiation exposure levels of routine SPECT/CT imaging protocols. Eur J Radiol 2016;85:1627-36. [Crossref] [PubMed]
42. Sriprapaporn J, Sethanandha C, Yingsa-nga T, Komoltri C, Thongpraparn T, Harnnanthawiwai C. Utility of Adding SPECT/CT Imaging to Post-Therapeutic Radioiodine Whole-Body Scan in Patients with Differentiated Thyroid Cancer. J Med Assoc Thai 2015;98:596-605.
43. Liao W, Coupland CAC, Burchardt J, Baldwin DR. DART initiative, Gleeson FV, Hippisley-Cox J. Predicting the future risk of lung cancer: development, and internal and external validation of the CanPredict (lung) model in 19·67 million people and evaluation of model performance against seven other risk prediction models. Lancet Respir Med 2023;11:685-97. [Crossref] [PubMed]
44. Stassen RC, Maas CCHM, van der Veldt AAM, Lo SN, Saw RPM, Varey AHR, Scolyer RA, Long GV, Thompson JF, Rutkowski P, Keilholz U, van Akkooi ACJ, Verhoef C, van Klaveren D, Grünhagen DJ. Development and validation of a novel model to predict recurrence-free survival and melanoma-specific survival after sentinel lymph node biopsy in patients with melanoma: an international, retrospective, multicentre analysis. Lancet Oncol 2024;25:509-17. [Crossref] [PubMed]
45. Clift AK, Collins GS, Lord S, Petrou S, Dodwell D, Brady M, Hippisley-Cox J. Predicting 10-year breast cancer mortality risk in the general female population in England: a model development and validation study. Lancet Digit Health 2023;5:e571-81. [Crossref] [PubMed]
46. Wang J, Zheng N, Wan H, Yao Q, Jia S, Zhang X, Fu S, Ruan J, He G, Chen X, Li S, Chen R, Lai B, Wang J, Jiang Q, Ouyang N, Zhang Y. Deep learning models for thyroid nodules diagnosis of fine-needle aspiration biopsy: a retrospective, prospective, multicentre study in China. Lancet Digit Health 2024;6:e458-69. [Crossref] [PubMed]
47. Sharma R, Mahanti GK, Panda G, Rath A, Dash S, Mallik S, Hu R. A Framework for Detecting Thyroid Cancer from Ultrasound and Histopathological Images Using Deep Learning, Meta-Heuristics, and MCDM Algorithms. J Imaging 2023;9:173. [Crossref] [PubMed]
48. Habchi Y, Himeur Y, Kheddar H, Boukabou A, Atalla S, Chouchane A, Ouamane A, Mansoor W. AI in Thyroid Cancer Diagnosis: Techniques, Trends, and Future Directions. Systems 2023;11:519.
49. Oh JM, Ahn BC. Molecular mechanisms of radioactive iodine refractoriness in differentiated thyroid cancer: Impaired sodium iodide symporter (NIS) expression owing to altered signaling pathway activity and intracellular localization of NIS. Theranostics 2021;11:6251-77. [Crossref] [PubMed]
50. Leboulleux S, Boucai L, Busaidy N, Durante C, Fagin JA, Fazeli S, Gianoukakis AG, Haugen BR, Kang H, Konda B, Laetsch TW, Locati L, Ryder M, Spitzweg C, Worden FP, Wirth L, Ho A. Redifferentiation therapy in unresectable or metastatic radioactive iodine refractory thyroid cancer: an International Thyroid Oncology Group statement. Lancet Diabetes Endocrinol 2025;13:516-27. [Crossref] [PubMed]