A proposed grading scheme for predicting recurrence in medullary thyroid cancer based on the Ki67 index and metastatic lymph node ratio

The Ki67 index and lymph node ratio (LNR) have been proposed as components of alternative pathological classification schemes, but the most appropriate classification for patients with medullary thyroid cancer (MTC) remains unknown. The aim of the present study was to examine the usefulness of a new grading system combining the Ki67 index and LNR as a predictor of prognostic and disease-free survival (DFS) in MTC. We conducted a retrospective study of patients with MTC who were registered at Sun Yat-sen University Cancer Center, Guangzhou, P. R. China from June 2003 to October 2021. The DFS rates were assessed using risk-adjusted Cox proportional hazard regression modeling to explore the relationship among pathological features, nutritional status and DFS. The Ki67 index (cutoff value: 5% and 10%) and LNR (cutoff value: 0.2 and 0.3) were combined to create a new grading system. In total, 101 matched patients were assessed. The integrated grading system showed better separation of Kaplan Meier (KM) curves for DFS. As the grading stage progressed, there was a significant stepwise decrease in DFS, which was better than Ki67, LNR and N staging alone. According to the grading system, the high-risk group had a worse prognosis. The proposed grading scheme demonstrated a better prognostic performance in MTC patients than the Ki67, LNR and N staging alone. However, larger scale studies are needed to further verify our findings.


Introduction
Medullary thyroid cancer (MTC) originates from thyroid parafollicular cells and accounts for 5 to 10% of thyroid cancers [1]. As a neuroendocrine tumor (NET), MTC is different from other types of thyroid cancer in clinical symptoms, diagnosis and treatment. Due to the absence of thyrotropin receptors in paracinar cells, MTC is not sensitive to thyroid-stimulating hormone suppression. Furthermore, MTC is refractory to radioiodine treatment after surgery [2]. Because MTC commonly presents with lymph node metastases and has a worse prognosis than differentiated thyroid cancer (DTC), the identification of novel prognostic biomarkers is necessary to predict its biological behavior.
Although clinical and pathological features of MTC influence prognosis in most cases, strong evidence is lacking, and other biological factors may influence disease progression and survival [3]. For example, somatic RET mutations in MTC are negatively correlated with tumor prognosis [4]. However, the specific types of RET mutations and their associations with clinical and pathologic features considerably vary, indicating that alternative biological markers may also be useful for determining prognosis in MTC. A few studies focusing on pathological indexes and patient nutritional status have reported the performance of limited methods in predicting disease-free survival (DFS) of MTC with unsatisfactory effects [5,6].
Ki67 is a marker of cell proliferation, and the Ki67 expression index predicts cancer progression in pulmonary and gastrointestinal neuroendocrine neoplasms; the Ki67 index has been well accepted to be a component of pathological grading [6,7]. As MTC is involved in the NET potential, Ki67 immunohistochemistry has good potential to predict cancer progression and survival in patients with MTC. Two studies have combined Ki67 with the mitotic index (MI), tumor necrosis and other indicators to develop two grading systems for MTC. Both the schemes have been shown to have merit by validation in subsequent independent studies [8,9]. However, there is a lack of a grading system to predict the gradient effect of MTC on DFS.
The metastatic lymph node ratio (LNR), which is the ratio of metastasized lymph nodes to the total number of evaluated lymph nodes, has been reported as a predictor in many other cancers, including MTC [10][11][12], but the superiority of the LNR over N staging has not been demonstrated in MTC.
Recently, the association between nutritional status and prognosis for patients with cancer has attracted increased attention. For example, body mass index (BMI) has been shown to predict morbidity rate and outcomes in several cancers, including thyroid cancer. Overweight individuals are at higher risk of developing related cancers [13,14] but exhibit better cancer prognosis [15], which is a phenomenon called 'obesity paradox'. However, the relationship between nutritional status and thyroid cancer is controversial. Some studies argue that higher BMI is associated with poorer prognosis in thyroid cancer [16,17], while other studies have suggested that higher Controlling Nutritional Status (CONUT) score, an immuno-nutritional screening tool based on serum albumin, total cholesterol and lymphocyte count that represents a worse nutritional status, may indicate worse prognosis of patients with thyroid cancer [18]. However, the association between nutritional status and prognosis of MTC has not been investigated. The aim of the present study was to investigate the potential prognostic value of the Ki67 index, LNR and nutritional status for recurrence of MTC as well as to design a grading system that can be used to risk stratify patients with MTC.

Materials and methods
Patients A retrospective analysis was performed using data collected from patients with MTC at Sun Yat-sen University Cancer Center from June 2003 to October 2021. Ethical approval for the study was obtained from the Institutional Ethical Committee. The requirement for informed consent was waived due to the observational, retrospective nature of this study. Participants were patients who underwent total thyroidectomy and central neck dissection with or without lateral neck dissection. All patients underwent surgery in the study hospital or other hospitals and received definite pathological diagnosis in our hospital (including consulted pathology diagnosis). Patients with distant metastasis were excluded. All included patients had the following data: Ki67 index, LNR, BMI, CONUT score, postoperative serum calcitonin (Ctn) and postoperative carcinoembryonic antigen (CEA). (Fig. 1).

Procedure
According to the standard clinical guideline from Chinese expert consensus, all patients had total thyroidectomy and received prophylactic or therapeutic center neck dissection. Neck dissection was performed for patients in which cervical lymph node metastasis was diagnosed by physical exam combined with ultrasound and/or a computed tomography (CT) scan. In general, the neck dissection involved levels II, III, IV and V.

Pathological features
The N staging was determined by the TNM staging system of the American Joint Commission on Cancer (AJCC) 8th edition. The LNR was defined as the proportion of positive neck lymph nodes among the total number of neck lymph nodes removed. According to the LNR cutoff value, patients were classified into low-LNR (LNR < 0.2), medium-LNR (0.2 ≤ LNR < 0.3) and high-LNR (0.3 ≤ LNR) groups. To avoid proportional bias, when the total number Fig. 1 Flow chart of sample selection of neck lymph nodes was less than six, the patient was classified into the low-LNR group regardless of the number of metastatic lymph nodes.
The weight of the patients was defined as underweight (BMI < 18.5), normal weight (18.5 ≤ BMI < 25) and overweight (25 ≤ BMI). The cutoff value of the CONUT score was set to 3 points.

Follow-up
All patients were reexamined 3 months after operation, and thyroid function, serum Ctn and CEA values were measured. The cut-off of normal range of Ctn and CEA following surgery were <18 pg/mL and <5 ng/mL respectively. For patients with undetectable basal Ctn and a CEA value within reference range, the follow-up intervals were 6 to 12 months. For patients with detectable basal Ctn or elevated CEA value, negative imaging and asymptomatic, the follow-up intervals were 3-6 months. However, the frequencies varied based on patient biochemical and clinical status. Ultrasonography of the neck was performed every 6 months, and CT of the neck and chest was performed every year. Lifelong follow-up was recommended for all patients. Local recurrence was diagnosed using sonography and fine-needle aspiration. DFS was defined as the time from the date of total thyroidectomy to structural recurrence or last follow-up. Overall survival (OS) was defined as the survival time after surgery. Structural recurrence was defined as metastatic lymph nodes confirmed by reoperation or distant metastasis confirmed by CT scan or positron emission tomography (PET). The cutoff date for follow-up was Oct 2022.

Statistical analysis
Continuous variables are presented as the mean and standard deviation (SD) or median (interquartile range, IQR). Categorical variables are presented as numbers and percentages. Data were analyzed using SPSS Statistics, version 21.0 (IMB corp, Armonk NY, USA). Both bivariate and multivariate Cox regression models were used in the association analyses. DFS was estimated by the Kaplan Meier (KM) method. Plots were generated with PRISM, version 8.0 (GraphPad, San Diego, CA, USA).

Baseline characteristics of the study population
In total, 101 patients with a mean age of 45.1 years (range 15-74 years) were enrolled, 52 (51.5%) female patients and 49 (48.5%) male patients. Among these, 93 patients were diagnosed with sporadic medullary thyroid carcinoma (sMTC), and eight patients were diagnosed with hereditary medullary thyroid carcinoma (hMTC). The mean maximum tumor diameter was 1.9 cm (range, 0.5-6.5 cm). All patients underwent total thyroidectomy and central lymph node dissection, of which 48 (47.5%) underwent cervical lymph node dissection. Table 1 shows the number of patients in the different weight, CONUT score and Ki67 groups.

Factors affecting DFS after surgery
Structural recurrence was identified in 31 patients at the end of the study period with a median follow-up period of 50 months. The longest follow-up period was 177 months, and the shortest follow-up period was 12 months. At the end of the follow-up, three patients died. Due to only a few deaths, the influence of related indicators on OS could not be analyzed.

Univariate and multivariate Cox regression analyses
The factors associated with recurrence were analyzed by univariate and multivariate Cox proportional hazard models. Extra nodal extension, high postoperative Ctn, medium Ki-67 proliferative index, high Ki-67 proliferative index, lymph node metastasis, N stage, medium LNR, high LNR, lymph node metastasis in area VI-VII and lymph node metastasis in area V were all significantly associated with worse DFS in univariate analyses (P < 0.05). Multivariable analysis identified medium Ki-67 proliferative index, high Ki-67 proliferative index, medium LNR and high LNR as predictors of DFS (P < 0.05; Table 2).
The difference of DFS between the smaller lymph node size group (lymph node <2 cm) and larger lymph node size group (lymph node ≥2 cm) was not significant (log rank: = 2.417, P = 0.103; Fig. S1). The DFS of the extra nodal extensionnegative group was significantly better than that of the extra nodal extension-positive group (log rank: x 2 = 4.152, P = 0.031; Fig. S2). The DFS of the low postoperative Ctn group (Ctn < 18 pg/mL) was significantly better than that of the high postoperative Ctn group (Ctn ≥ 18 pg/mL), (log rank: x 2 = 9.432, P = 0.011; Fig. S3). The difference of DFS between the low postoperative CEA group (CEA < 5 ng/mL) and high postoperative CEA group (CEA ≥ 5 ng/mL) was not significant (log rank: x 2 = 0.040, P = 0.843; Fig. S4).

Grading system
Using a combination of the Ki67 index and LNR, cases were divided into low-risk grade, medium-risk grade and high-risk grade groups. When the total number of neck lymph nodes was less than six, the patient was classified into the low-LNR group regardless of the number of LNR metastatic lymph node ratio, BMI body mass index, CONUT score Controlling Nutritional Status score metastatic lymph nodes. Cases with low Ki67 index (defined as Ki67 < 5%) and low LNR (defined as LNR < 0.2) were considered low-risk grade. Cases with low Ki67 index (defined as Ki67 < 5%) but with high LNR (defined as LNR ≥ 0.2) or medium Ki67 index (5% ≤ Ki67 < 10%) with low LNR (defined as LNR < 0.2) were defined as mediumrisk grade. Cases with medium Ki67 index (5% ≤ Ki67 < 10%) but with high LNR (defined as LNR ≥ 0.2) or high Ki67 index (defined as Ki67 ≥ 10%) were defined as high-risk grade ( Table 3).

Discussion
Compared to other differentiated thyroid cancers, MTC is more aggressive with higher risk of distant metastasis and poor prognosis [19]. The clinical behaviors of MTC are varied and unpredictable despite close surveillance with imaging, serum calcitonin and CEA testing [20]. Some patients may pursue a slow clinical course even with distant metastases or lymph node metastasis in the lateral cervical region. In contrast, some patients with earlier clinical stages have poor DFS. Therefore, it is of great interest to identify indicators that effectively predict the biological behavior of MTC. Ki67, which is directly proportional to cell proliferation, is one of the most widely used markers of proliferation in oncology. Uncontrolled cell proliferation is a basic feature of malignant tumors [21]. Therefore, the use of Ki67 as a biological behavior predictor of malignant tumors is feasible. There is now broad agreement that neuroendocrine tumors (NETs) of many sites, including pulmonary and gastrointestinal neuroendocrine neoplasms, can be graded on the basis of the simple factors of Ki67 proliferative index [22]. As a type of NET [23], MTC can be graded by the standard Ki67 grading index.
Some studies have revealed that the LNR may be superior to the previous N staging system in predicting the prognosis of a number of cancers [24][25][26]. However, in MTC, none of the investigated novel LNR classification systems has demonstrated clear discriminative superiority in the prediction of prognosis over the currently implemented N staging system [27]. In the present study, we investigated the ability of the Ki67 index combined with the LNR to predict the DFS of MTC.
The results showed that the DFS of the low-Ki67 group was better than that of the medium-and high-Ki67 groups, but there was no statistically significant difference between the latter two groups. Similar results were observed in the different LNR groups and N staging groups. Neither the LNR nor Ki67 showed a stronger association with DFS than   N staging. Based on these findings, we propose a novel grading system that uses a combination of the Ki67 index and the LNR to predict DFS in MTC. The integrated grading system showed better separation of KM curves for DFS. As the grading stage progressed, there was a significant stepwise decrease in DFS. Therefore, for MTC patients with high grading, a more aggressive surgical approach should be considered, and postoperative follow-up of such patients should be more rigorous. In addition to the Ki67 index and the LNR affecting the risk assessment of MTC patients, univariate analysis showed that extra nodal extension, higher postoperative Ctn, higher N staging and lymph node metastasis (VI-VII area and V area) indicated poorer DFS, which was similar to previously published results [28][29][30]. Thus, the Ki67 index combined with the LNR may increase the ability to predict the biological activity of MTC, thereby aiding in the development of more personalized treatment strategies.
Recently, two independent studies have identified the MI, tumor necrosis and Ki67 proliferative index as prognostic histological features in MTC, proposing two different but broadly similar histological grading schemes for MTC [8,9]. Fuchs et al. [8] proposed a three-tiered grading system based on the Ki67 index (<3, 3-20 and >20%) and coagulation necrosis with the highest OS in the low Ki67 group (mean survival:195 months) followed by the intermediate Ki67 group (mean survival: 99 months) and the high Ki67 group (24 months). Similar results have been obtained by combining coagulation necrosis indexes. Xu et al. introduced the MI and used a secondary grading system, and they classified MTC as high grade when it has at least one of the following three features: MI ≥ 5 per 2 mm 2 , Ki67 proliferative index ≥ 5% and/or tumor necrosis. Different from the above two studies, we grouped MTC with truncation values of 5 and 10% combined with the LNR, which made the classification easier and achieved good DFS prediction ability. The fifth edition of the World Health Organization (WHO) Classification of Endocrine and Neuroendocrine Tumors also introduced a pathological staging system based on Ki67 index, MI and tumor necrosis. All above confirm the importance of Ki67 in MTC and the rationality of our research [31]. The difference of DFS among the high-, medium-and low-risk groups in the present study (mean survival: 41.3 months vs. 94.9 months vs. 153.8 months) compared to the study by Fuchs et al. may be due to different grading conditions. The nutritional status, including BMI and CONUT score, have been identified as independent prognostic indicators in several cancers [32][33][34][35][36][37], indicating that their roles in MTC should be investigated. However, the present study showed that there were no differences in DFS among patients with different BMI and CONUT scores, which may be caused by a paucity of patients with high BMI and high CONUT scores.
There was one point that merit discussion: the data showed a 64.4% proportion of patients whose calcitonin is higher than normal 3 months after operation. Two studies showed that the main factors of the failure to normalize postoperative calcitonin in MTC patients 3 months after operation were lymph node positivity, higher preoperative calcitonin level than 500 pg/mL and high AJCC stage [38,39]. For example, Machens' study showed that the percentage of 3-6 months postoperative Ctn normalization in lymph node positive MTC patients was 10% [39], another study showed that MTC patients with positive lymph nodes had a Ctn normalization rate of 28.6% within 3 months after operation [40], and in Choi's study, the percentage of 6 months postoperative Ctn normalization in patients with stage III-IV MTC was 22.2% [41]. Among the patients in our study, the proportion of these three types of patients is high: the proportion of patients with positive lymph node was 68.3%, the proportion of patients with AJCC stage III-IV was 71.3% and the proportion of patients with preoperative calcitonin level higher 500 pg/mL was 73.3%. In the postoperative Ctn non-normalized group, the proportion of III-IV AJCC stage was 87.7%, the proportion of lymph node positive patients was 84.6%, and the proportion of patients with calcitonin level higher than 500 pg/mL was 83.1%. This may be the reason for the high proportion of 3 months postoperative non-normalization of calcitonin in patients enrolled in our study. Moreover, all operating surgeons were head and neck disease specialists and had extensive experience in thyroid gland operations, which could ensure R0 resection achieved. And our data showed that 95% patients had at least Ultrasonography of the neck after operation. Imaging examinations showed no abnormalities. 98% enrolled patients had major reductions in postoperative calcitonin levels. (Mean decrease: 2834.6 pg/mL and mean decreasing proportion: 84%), and it may further decrease in the later period since. Taking together, we believe that in our study, a 64.4% proportion of patients whose calcitonin is higher than normal 3 months after operation, or Ctn may take it nadir in a longer time range is reasonable.
The present study had several limitations, including a small number of enrolled patients, short follow-up duration, no prospective study and lack of OS data, which may be caused by the short follow-up period. However, the present study has certain guiding significance for the prediction of clinical characteristics of MTC and DFS, and additional studies should be performed for verification.
In conclusion, the proposed grading scheme demonstrated a better prognostic performance in MTC patients compared to the Ki67 index, LNR and N staging alone. However, larger scale studies are needed to further verify our findings. The present grading scheme is a simple and feasible indicator to aid clinicians in the decision-making process for the diagnosis and treatment plan of MTC.
Author contributions XPF, WD and LXK designed this experiment and participated in data collection, data analysis, manuscript writing and manuscript revision.