Thyroid tumors are highly prevalent worldwide, accounting for 3% of the global incidence of malignancies with 586 000 new cases in 2020 [16]. Papillary thyroid cancer (PTC) was the leading cause of total thyroid cancer incidence in the 25 countries for which statistics were available. The annual incidence rate of PTC in Chinese women and men were 25.8 and 8.66/100,000 person-years respectively [3, 17].
It's worth noting that CLNM is an important prognostic factor for PTC. The occurrence of CLNM is often associated with a poor prognosis, including residual and recurrence, and this has been confirmed by several studies [18–20]. The current preoperative evaluation of lymph node metastasis is mainly performed by US and US-guided FNA, but studies have shown that US is not very sensitive for central group lymph nodes in the neck, and the sensitivity of US for CLNM has been reported to be as low as 10 to 30% [21]. Evidence shows prophylactic CLND leads to improvements in disease-specific survival [22], local recurrence [23, 24], and post Tg-levels [24, 25] as well as provides a grading basis for patient prognosis, postoperative treatment, and follow-up of iodine-131 [26], while some studies have argued that prophylactic dissection does not improve the long-term prognosis of patients and increases the likelihood of hypocalcemia and hypothyroidism [27, 28]. Therefore, prophylactic CLND is still controversial.
It has been reported that patients with PTC have a 20–80% chance of regional lymph node involvement, most commonly in the central region of the neck [4–6]. We thought the reason for such a discrepancy of CLNM rate is the lack of a more refined subgroup analyses for patients with PTC. We noticed that currently, many investigators are developing CLNM risk stratification prediction tools incorporating many risk factors to help evaluate the need for CLND. Yet among these, few studies have focused on the likelihood of CLNM in the mono-focal, encapsulated group of PTC, underestimating the risk of CLNM in this group of patients whom have traditionally been considered as low cervical lymph node involvement risk group among all PTC patients [13–15]. Our study innovatively reviewed data from 1014 mono-focal and encapsulated PTC patients to better assist clinicians in making decisions about whether to perform CLND, and the frequency of postoperative follow-up.
Our retrospective cohort study can reveal that patient age, gender, cMTD, pMTD and mTCI are all independent risk predictors of CLNM among PTC patients, which is similar to the results reported in our previous study [29]. Meanwhile, similar results were found in a retrospective study of 1555 and 966 PTC patients by Shuai Xue et al [30] and Chenxi Liu et al [10], respectively, who both recognized age ≤ 45 years, male, tumor size > 1.0 cm, and ETE as independent risk factors for CLNM, among which the risk factor of male was supported by most of the studies [31–33], while in terms of age, researchers hold different opinions: Liu Z [34] et al found that CLNM was not associated with age and other studies reported that age < 45 years was a risk factor for CLNM [32, 33]. In addition, in the study by Chenxi Liu et al [10], US features as microcalcifications were included in the analysis of risk factors. However, their studies only performed a uni- or multivariate analysis without further quantitative prediction modeling whereas the model developed in our study can more accurately guide clinicians' decisions and is more practical. The two pathological variables pMTD, mTCI obtained postoperatively are also contribute to CLMN by multivariate analysis and were therefore included in the construction of the postoperative model.
Interestingly, Hashimoto's thyroiditis has been considered in several studies as a precursor disease to PTC and also as a protective factor for CLNM [35, 36], but in this study, iHT in patients demonstrates no correlation with CLNM was derived (p = 0.456 > 0.05; Table 1). This is actually consistent with the findings of Song E, et al [37]. However, it is noteworthy that in patients with PTC with HT, mono-focal patients were more likely to have CLNM than multifocal patients [38], contrary to the finding that in the total population with PTC, multifocal patients were more likely to have CLNM than mono-focal patients [29, 39]. This suggests that there may be some interaction between HT and multifocality, and further studies can be conducted on this contraction.
Nomogram's ability to generate personalized predictions allows it to be utilized for patient risk stratification. The combination of a user-friendly interface and wide availability has contributed to its popularity among oncologists and patients [40]. Thus, it is well suited to the clinical decision-making problem that this study is trying to address. In our study, two nomograms were constructed based on the three preoperative and four postoperative variables described above to screen for the likelihood of CLNM in mono-focal encapsulated PTC patients preoperatively and postoperatively. Both models were shown to be valid by calculating C-index and plotting calibration curves. There are other models from similar studies [39, 41], but they are basically only preoperative models, which are not as good as our pre-and post-operative combined models, providing complete pre- and post-operative scores to comprehensively evaluate the risk of CLNM. Meanwhile, they are also not as specific and detailed in the PTC patient group of concerns as what we focus on. With our models, clinicians can individualize the preoperative and postoperative scores for each patient and find the corresponding CLNM risk based on the scores in the nomogram.
In addition, a detailed treatment selection flow chart was designed based on risk factor analysis and nomogram conduction. It is used to demonstrate the risk stratification of CLNM in mono-focal encapsulated PTC patients and the relevant steps (Supplementary Figure S1). Before surgery, patients can be evaluated using the preoperative nomogram. The three subgroups formed by the selection of two cut-off points proved to be reasonable and valid for (p < 0.001) risk stratification: low (≤ 80, with CLNM rate of 4.0%), moderate (80–160, with CLNM rate of 15.4%) and high risk (≥ 160, with CLNM rate of 45%) groups. Considering the higher CLNM rate, prophylactic CLND is recommended for patients at high risk in the predictive model. The risk of occult CLNM can be reassessed by the post-operative prediction model for those at moderate and low risk in the preoperative model, as well as those in the high-risk group who do not receive CLND, in which nomogram adds several pathological factors that could not be derived preoperatively to aid in a more accurate risk assessment.
As far as the postoperative evaluation process is concerned, for patients in the high-risk group (Post-model risk score ≥ 160, with CLNM rate of 40.4%) or in the intermediate-risk group (Post-model risk score 80–160, with CLNM rate of 22.1%), it is strongly recommended to perform close follow-up with particular focus on the central neck region to monitor changes in CLN at all times in order to improve survival with timely and appropriate treatment. For the low-risk group (≤ 80, with CLNM rate of 6.7%), regular follow-up is recommended considering the relatively low probability of CLNM. Most of the studies in this topic focus more on the risk factors of CLNM in overall PTC patients [10, 34], and fewer studies have been conducted on mono-focal and encapsulated PTC patients because of their relatively lower risk of CLNM compared with multifocal and gross ETE PTC. However, this does not mean that the risk of CLNM in mono-focal and encapsulated PTC patients can be ignored, so our study originally focus on individualized CLNM risk assessment in this specific group of PTC patients to help clinicians determine the optimal timing of CLND for maximum patient benefit.