Development and validation of an US-based clinical prediction model to estimate the probability of metastasis in lymph nodes posterior to the right recurrent laryngeal nerve in papillary thyroid cancer

Background The dissection of lymph nodes posterior to the right recurrent laryngeal nerve (LN-prRLN) in papillary thyroid carcinoma (PTC) remains controversial due to the postoperative complications. Estimating the clinical probability of metastasis in LN-prRLN can facilitate the selective dissection. Objective To develop a clinical prediction model based on clinical and sonographic features for estimating the probability of metastasis of LN-prRLN in PTCs. This model was then validated. Methods A total of 601 patients with surgical-proved PTCs in the First Affiliated Hospital of Nanjing Medical University were evaluated retrospectively. Clinical data, preoperative sonographic features based on ACT TI-RADS categories, and pathological data were used to develop an US-based clinical prediction model. Performance of the model was calculated, compared and then validated by using a test set of 144 patients. Results LN-prRLN metastasis was confirmed in 133 (22.1%) patients. The US-based clinical prediction model identified age < 45 years (odds ratio [OR]:1.752), horizontal diameter > 5mm (OR:1.061), dorsal side (OR:2.185), US-suspicious RLLNM (OR:2.488), pathological size > 5mm (OR:3.697), mulifocality in the right robe (OR:2.584), other CLN metastasis (OR:2.185) and LN-arRLN metastasis ≥ 1 (OR:3.915) were independent risk factors for LN-prRLN metastasis. The AUCs of the clinical prediction model was 0.835 (95% CI: 0.803–0.864). The sensitivity, specificity, positive predictive value, and negative predictive value were 76.69%, 78.42%, 50.20%, and 92.22%, respectively. The clinical prediction model applied to the test dataset demonstrated an AUC of 0.804 (95% CI:0.722–0.886), sensitivity, specificity, positive predictive value, and negative predictive value of 78.79%, 72.07%, 43.88% and 92.46%. Conclusions The clinical prediction model based on clinical and sonographic characteristics can effectively assess LN-prRLN metastasis preoperatively and do help in selective LN-prRLN dissection of PTC.