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

Background: Radioiodine whole-body scan (WBS) is routinely performed after radioiodine therapy in papillary thyroid cancer (PTC), but its two-dimensional nature limits anatomical localization. The addition of single-photon emission computed tomography/computed tomography (SPECT/CT) improves diagnostic accuracy; however, routine use exposes patients to additional radiation, longer imaging time, and higher costs. This study aimed to establish a predictive model to identify which patients with PTC are more likely to benefit by adding SPECT/CT imaging after WBS, enabling selective rather than routine use. Methods: This retrospective study included 265 patients with pathologically confirmed PTC who underwent postoperative radioiodine therapy followed by WBS and SPECT/CT. Patients were classified into non-metastasis and metastasis groups according to WBS diagnosis. The WBS results were also compared with the SPECT/CT results with WBS, and accordingly, the patients were categorized into true-negative, false-negative, true-positive, and false-positive subgroups. Additionally, the least absolute shrinkage and selection operator (LASSO) logistic regression analysis was used to screen for significant clinical factors that could be applied to develop predictive models for the addition of SPECT/CT imaging. Finally, the performance of the predictive models was evaluated using nomograms and receiver operating characteristic (ROC) curves. Results: Of the 232 patients diagnosed as non-metastasis by WBS, 17 false-negative patients were identified via SPECT/CT. A comparison of the clinicopathological characteristics of the patients revealed statistically significant differences in tumor size and thyroglobulin (TG) levels between the true-negative and false-negative diagnosis subgroups. Furthermore, of the patients diagnosed with metastases by WBS, SPECT/CT identified 22 true-positive patients and excluded 11 false-positive patients. Comparing the true-positive and false-positive patients of the metastasis group, we found that the American Joint Committee on Cancer (AJCC) stage was statistically low in the false-positive subgroup. Finally, the LASSO logistic regression model (LL-model) and the LASSO logistic regression model with clinical factors (Clin-model) were established to predict the utilization of SPECT/CT imaging, which produces an area under the curve (AUC) of 0.814 [95% confidence interval (CI): 0.725-0.902] and 0.855 (95% CI: 0.778-0.932). Conclusions: Our predictive algorithm integrated standard clinical variables (sex, tumor size, TG levels), WBS and SPECT/CT results through optimized LASSO regression. Additionally, evaluation of the model indicated optimal results that can be used to facilitate individualized prediction of PTC patients requiring additional SPECT/CT imaging.