, INTRODUCTION
Cancer staging plays an important role in patient management and in communications about the prognosis of disease among clinicians and between physicians and their patients [1]. Therefore, accurate tumor staging is critical for treatment decisions and ultimately for patient outcomes [1,2]. The American Joint Committee on Cancer (AJCC) tumor-node-metastasis (TNM) staging system has been widely used to predict disease mortality in patients with differentiated thyroid carcinoma (DTC) [1,2,3]. The revised 8th edition of the AJCC TNM staging system (TNM-8) for DTC includes important changes from the 7th edition (TNM-7) [2,3] to minimize over-staging of older patients with low-risk tumors and to provide a more realistic prognosis for patients in high-risk groups [4]. Comparative studies of patients staged by the TNM-7 and TNM-8 showed that the changes incorporated into the TNM-8 could lead to down-staging of approximately 40% of patients with DTC and could better predict patient mortality [5,6]. However, survival discrimination powers were still low in both studies and some aspects of the TNM-8 require improvement [5,6].
Survival outcomes in DTC patients with gross extra-thyroidal extension (ETE) were found to differ according to the direction and extent of ETE [7,8]. Recently, a single center study showed that disease-specific survival (DSS) was better in patients with stage T3b DTC (gross ETE invading only the strap muscles with tumor size ≤4 cm) than for patients with T3a DTC (tumor size >4 cm and limited to the thyroid), with the former having DSS similar to that of patients with T2 DTC [7]. These results suggested the importance of downgrading stage T3b to stage T2 [7]. In addition, modifying TNM-8 by employing N1b classification may better predict survival outcomes [9,10]. The present study investigated whether the TNM-8 classification could be improved by including ETE and N1b in the evaluation of patients with DTC. Specifically, this study compared the prognostic ability of the TNM-8 staging system and a modification of this staging system that included gross ETE and lateral cervical lymph node (LN) metastasis (N1b) to predict DSS in a large, multicenter cohort of patients with DTC.
METHODS
Patients
In this retrospective, multicenter cohort study, 4,878 patients with DTC who underwent thyroid surgery between 1996 and 2005 at five tertiary hospitals in Korea were included: Asan Medical Center (n=3,176), Chonnam National University Hwasun Hospital (n=691), Pusan National University Hospital (n=552), Ulsan University Hospital (n=345), and Chungnam National University Hospital (n=114). All the included patients were pathologically proven to have papillary thyroid carcinoma (PTC) or follicular thyroid carcinoma (FTC). The research protocol was approved by the Institutional Review Board of each institution: Asan Medical Center, 2016-1301; Chonnam National University Hwasun Hospital, CNUHH-2017-053; Pusan National University Hospital, 1701-014-051; Ulsan University Hospital, 2016-12-031; and Chungnam National University Hospital, CNUH 2017-01-018. Informed consent was waived due to the retrospective design of this study [10].
The clinical and pathological data obtained from electronic medical records included age at the time of diagnosis, sex, primary tumor size, pathologic subtype, extent of ETE, presence and location of cervical LN metastases, presence of distant metastases, extent of surgery, and use of radioactive iodine ablation. The survival status, last follow-up time, and causes of death were also collected.
Definition of modified-TNM
Patients categorized according to the TNM-8 [3] were subsequently reclassified using a modified-TNM staging system (Table 1). Patients were reclassified into different T and TNM stages by the modified-TNM staging system. Patients classified as T3b (gross ETE invading only strap muscles with tumor size ≤4 cm) according to the TNM-8 were reclassified as T2 according to the modified-TNM, and stage II on the TNM-8 was subdivided into stages IIA and IIB on the modified-TNM. Older patients (≥55 years) with N1b disease were reclassified as stage IIB on the modified-TNM, whereas the remaining patients were classified into stage IIA.
Primary outcome
The TNM staging system is useful for predicting disease mortality and therefore recommended for staging of patients with DTC [3]. Therefore, the primary outcome of this study was DSS, defined as the time interval from the date of the initial thyroidectomy to the date of last censor or death due to DTC.
Statistics
Data were analyzed using R version 3.5.1 software and the R libraries (R Foundation for Statistical Computing, Vienna, Austria; http://www.R-project.org). Categorical variables were presented as numbers with percentages and continuous variables were as means with standard deviations. We plotted DSS curves using the Kaplan-Meier method and compared them using log-rank test. We evaluate the impact of TNM stage groupings on DSS based on the Cox-proportional hazard model. Relative risks of survival are reported as hazard ratios (HRs), 95% confidence interval (CI), and P value.
The proportion of variation explained (PVE) in the Cox-proportional hazard model was calculated to estimate how well each of the staging systems predicted the outcome of DSS. PVE, which ranges from 0 to 100, was calculated using the formula: PVE=1−exp(−G2/n), in which G2 is the maximum likelihood ratio determined by analysis using the chi-square test associated with the null hypothesis, and n is the total number of valid patients in the study [11]. All P values were two-sided, and P values less than 0.05 were considered statistically significant.
RESULTS
Baseline characteristics
The baseline clinical and pathological characteristics of the 4,878 patients with DTC are illustrated in Table 2. Mean patient age was 45.9±12.2 years, and 87% were women. Most patients (n=4,704, 96%) were diagnosed with PTC. The mean primary tumor size was 1.7±1.3 cm, and 3,534 (72%), 1,116 (23%), and 228 patients (5%) had primary tumor sizes ≤2, 2 to 4, and >4 cm, respectively. Of the 4,878 patients, 1,407 (29%), 577 (12%), and 267 (5%) showed microscopic ETE, gross ETE invading the strap muscles alone, and gross ETE invading areas other than the strap muscles, respectively. Of the 577 patients with gross ETE invading only the strap muscles, 530 (92%) had primary tumors ≤4 cm in size and 47 (8%) had tumors >4 cm. Cervical LN metastases were identified in 2,116 patients (43%), including 1,624 (33%) with N1a and 492 (10%) with N1b disease. Distant metastasis at diagnosis was found in 74 patients (2%). The median follow-up period until censoring or death was 10.2 years (interquartile range, 7.8 to 12.0).
Distribution of patients per stage according to the TNM-8 and the modified-TNM
Fig. 1 shows the migration of staging from the TNM-8 to the modified-TNM. Of 709 patients classified as having T3 disease on the TNM-8, 530 (75%) with T3b (gross ETE invading only strap muscles with tumor size ≤4 cm) disease were down-graded to the T2 category (Fig. 1A). The TNM-8 classified 4,214 patients (86%) as having stage I tumors, 540 (11%) as having stage II, 97 (2%) as having stage III, and 27 (1%) as having stage VI (Fig. 1B). Application of the modified-TNM to the 540 patients classified as stage II according to the TNM-8 resulted in 75 (14%) of these patients being down-staged to stage I, with the remaining 381 (71%) and 84 patients (16%) reclassified as stages IIA and IIB, respectively. Among the 75 patients being down-staged to stage I, there were 72 classical PTC, two follicular variant PTC, and one FTC. Patients with stages III and IV tumors according to the TNM-8 showed no changes when classified according to the modified-TNM.
DSS according to the TNM-8 and modified-TNM
DSS in patients with DTC was evaluated according to the TNM-8 and modified-TNM staging systems (Table 3, Fig. 2). DSS was associated significantly with stage at diagnosis, as determined by the TNM-8 (P<0.001) (Fig. 2A) and modified-TNM (P<0.001) (Fig. 2B) staging systems. However, differences in DSS between stage groups appeared to be more distinct when patients were staged by the modified-TNM than the TNM-8 (Fig. 2).
When staged according to the TNM-8, the 10-year DSS rates for patients with stages I, II, III, and IV DTCs were 99.8%, 95.9%, 81.0%, and 41.6%, respectively (Table 3). When staged according to the modified-TNM, the 10-year DSS rates for patients with stages I, IIA, IIB, III, and IV DTCs were 99.8%, 96.4%, 93.3%, 81.0%, and 41.6%, respectively. Using modified-TNM stage I as the reference, patients with stage IIA (HR, 16.7; 95% CI, 8.1 to 34.0; P<0.001) and stage IIB (HR, 38.6; 95% CI, 17.4 to 85.4; P<0.001) had significantly poorer DSS. The value of the PVE of the modified-TNM system was 6.5% and that of the TNM-8 was 6.3%.
DSS in patients with stage migration after applying the modified-TNM
Patients classified as stage II according to the TNM-8 were reclassified as stages I, IIA, and IIB according to the modified-TNM. Following stage migration, the DSS curves of these three groups differed significantly (P=0.010) (Fig. 3A), with their 10-year DSS rates being 96.7%, 96.4%, and 93.3%, respectively. DSS rates also differed significantly between stages IIA and IIB (P=0.030) (Fig. 3B), with patients classified as stage IIB having a 2.3-fold poorer DSS than those reclassified as stage IIA (95% CI, 1.1 to 4.8; P=0.035).
DISCUSSION
This multicenter cohort study evaluated whether and how the prognostic predictability of the TNM-8 staging system for patients with DTC could be improved by modifications focusing on both gross ETE and lateral cervical LN metastasis. Application of the modified-TNM staging system resulted in stage migration only in patients classified as stage II according to the TNM-8. About 14% of these patients were down-staged to stage I, whereas 71% and 16% were reclassified as stages IIA and IIB, respectively. The modified-TNM staging system appeared to better discriminate among stages and to better predict DSS than the TNM-8. Potential modifications, including the downgrading of patients with T3b (gross ETE invading only strap muscles with tumor size ≤4 cm) to T2 and the upgrading of older patients with N1b disease from stage II, may better predict DSS in patients with DTC.
In the modified-TNM, tumors with gross ETE measuring ≤4 cm are down-staged to T2 disease. This change is based on the findings of previous studies, which showed that gross ETE invading only the strap muscles did not affect survival outcome in patients with DTC, especially in patients with tumors ≤4 cm in size [7,8,12]. In addition, patients with gross ETE invading only the strap muscles and tumor size ≤4 cm showed no difference in DSS compared to patients with T2 disease, but rather showed significantly better DSS compared to patients with T3a disease [7]. ETE has been considered as a prognostic factor in patients with DTC and is observed in 5% to 45% of these patients [13]. Beginning with the sixth edition of the AJCC TNM staging system, the presence of ETE has been divided into two grades, minimal and massive ETE, affecting the T category [2]. Many studies have shown that DTC patients with massive ETE have a much poorer survival outcome than those with minimal ETE [13,14], and that DTC patients with minimal and no ETE have similar survival outcomes [15,16]. For this reason, the revised TNM-8 ascribed high importance to gross ETE but eliminated minimal ETE in T staging [3]. However, it was unclear whether patients having tumors ≤4 cm with gross ETE invading the strap muscles alone also have a poor prognosis, and whether it is appropriate for these patients to be upgraded to a higher T grade. Although 75% of patients in this study with T3 disease were down-graded to T2, the 10-year DSS rate of patients classified as stage T2 by the modified-TNM was excellent (98.9%, data not shown).
The TNM-8 classification of patients with DTC does not consider the location of metastatic LNs in stage grouping [3]. Consequently, the TNM-8 downstages a large number of patients with N1b disease. However, there is evidence that N1b disease is a significant prognostic factor in patients with DTC, and that survival outcomes are worse for DTC patients with N1b than with N1a [9,17,18]. This study found that older patients with N1b disease were predicted to have a significantly lower 10-year DSS rate than those with N0 or N1a disease, emphasizing the prognostic significance of N1b in older patients with DTC, and suggesting that N1b should not be treated as equivalent to N0 or N1a. Further research is needed to assess the prognostic value of N1b disease according to the modified-TNM staging system for patients with DTC.
This study had limitations because of its retrospective design. There may be variations in the intensity of treatment and the frequency of follow-up. However, our findings are based on data from a large multicenter cohort population with a relevant number of disease-specific death events during a median 10.2 years of follow-up.
In conclusion, this study provided valuable information about the suitability of the TNM-8 for staging DTC and suggested that modifications based on ETE and N1b may improve its accuracy. Differences in DSS between each pair of stages on the modified-TNM appeared greater than those between each pair of stages on the TNM-8. In addition, the modified-TNM grouping appeared to be a better predictor of DSS than the TNM-8. These findings indicate that the modified-TNM grouping proposed here may better predict outcomes and improve treatment selection for patients with DTC. Further researches are needed to valid the prognostic accuracy of these modifications.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
CrossRef TDM
Print
