Rethinking the Current American Joint Committee on Cancer TNM Staging System for Medullary Thyroid Cancer | Endocrinology | JAMA Surgery | JAMA Network
[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Figure 1.  Overall Survival in the National Cancer Database Cohort
Overall Survival in the National Cancer Database Cohort

AJCC indicates American Joint Committee on Cancer.

Figure 2.  External Validation in the Surveillance, Epidemiology, and End Results Program Cohort
External Validation in the Surveillance, Epidemiology, and End Results Program Cohort

Comparison of overall survival based on the current American Joint Committee on Cancer (AJCC) TNM staging system (A) and proposed TNM groupings (B) and disease-specific survival based on the current AJCC staging system (C) and proposed TNM groupings (D). See Figure 1 for descriptions of the stages in the current AJCC staging system and the proposed TNM groupings.

Figure 3.  Correlation Between Overall Survival and Disease-Specific Survival Based on Pathologic TNM Stages
Correlation Between Overall Survival and Disease-Specific Survival Based on Pathologic TNM Stages
Table 1.  Demographic and Clinicopathologic Characteristics of Patients With Medullary Thyroid Cancer in the National Cancer Database, 1998-2012a
Demographic and Clinicopathologic Characteristics of Patients With Medullary Thyroid Cancer in the National Cancer Database, 1998-2012a
Table 2.  Adjusted Overall Survival Based on Proposed TNM Groupings and Current AJCC TNM Staging System
Adjusted Overall Survival Based on Proposed TNM Groupings and Current AJCC TNM Staging System
1.
Williams  ED.  Histogenesis of medullary carcinoma of the thyroid.  J Clin Pathol. 1966;19(2):114-118.PubMedGoogle ScholarCrossref
2.
Wells  SA  Jr, Asa  SL, Dralle  H,  et al; American Thyroid Association Guidelines Task Force on Medullary Thyroid Carcinoma.  Revised American Thyroid Association guidelines for the management of medullary thyroid carcinoma.  Thyroid. 2015;25(6):567-610.PubMedGoogle ScholarCrossref
3.
Tuttle  RM, Ball  DW, Byrd  D,  et al; National Comprehensive Cancer Network.  Medullary carcinoma.  J Natl Compr Canc Netw. 2010;8(5):512-530.PubMedGoogle ScholarCrossref
4.
Jin  LX, Moley  JF.  Surgery for lymph node metastases of medullary thyroid carcinoma: A review.  Cancer. 2016;122(3):358-366.PubMedGoogle ScholarCrossref
5.
Roman  S, Lin  R, Sosa  JA.  Prognosis of medullary thyroid carcinoma: demographic, clinical, and pathologic predictors of survival in 1252 cases.  Cancer. 2006;107(9):2134-2142.PubMedGoogle ScholarCrossref
6.
Boostrom  SY, Grant  CS, Thompson  GB,  et al.  Need for a revised staging consensus in medullary thyroid carcinoma.  Arch Surg. 2009;144(7):663-669.PubMedGoogle ScholarCrossref
7.
Tuttle  RM, Ganly  I.  Risk stratification in medullary thyroid cancer: moving beyond static anatomic staging.  Oral Oncol. 2013;49(7):695-701.PubMedGoogle ScholarCrossref
8.
Beahrs  O, Carr  DT, Rubin  P; American Joint Committee for Cancer Staging and End Results Reporting.  Manual for Staging of Cancer. Philadelphia, PA: J B Lippincott Co; 1977.
9.
Beahrs  O, Henson  DE, Hutter  RVP, Kennedy  B; American Joint Committee on Cancer.  Manual of Staging of Cancer. 4th ed. Philadelphia, PA: J B Lippincott Co; 1992.
10.
Beahrs  O, Henson  D, Hutter  RVP, Myers  MH, eds; American Joint Committee on Cancer.  Manual of Staging of Cancer. 3rd ed. Philadelphia: J B Lippincott Co; 1988.
11.
Beahrs  O, Myers  M; American Joint Committee on Cancer.  Manual of Staging of Cancer. 2nd ed. Philadelphia: J B Lippincott Co; 1983.
12.
Edge  S, Byrd  D, Compton  C, Fritz  A, Greene  F, Trotti  A; American Joint Committee on Cancer.  AJCC Cancer Staging Manual. 7th ed. New York, NY: Springer; 2010.
13.
Fleming  I, Cooper  J, Henson  D,  et al; American Joint Committee on Cancer.  AJCC Cancer Staging Manual. 5th ed. Philadelphia, PA: Lippincott-Raven; 1997.
14.
Greene  F, Page  D, Fleming  I,  et al; American Joint Committee on Cancer.  AJCC Cancer Staging Manual. 6th ed. New York, NY: Springer; 2002.Crossref
15.
Amin  M, Edge  S, Greene  F,  et al.  AJCC Cancer Staging Manual. 8th ed. Cham, Switzerland: Springer International Publishing; 2017.Crossref
16.
Raval  MV, Bilimoria  KY, Stewart  AK, Bentrem  DJ, Ko  CY.  Using the NCDB for cancer care improvement: an introduction to available quality assessment tools.  J Surg Oncol. 2009;99(8):488-490.PubMedGoogle ScholarCrossref
17.
Fritz  A, Percy  C, Jack  A,  et al, eds.  International Classification of Diseases for Oncology. 3rd ed. Geneva, Switzerland: World Health Organization; 2000.
18.
Adam  MA, Pura  J, Gu  L,  et al.  Extent of surgery for papillary thyroid cancer is not associated with survival: an analysis of 61,775 patients.  Ann Surg. 2014;260(4):601-605.PubMedGoogle ScholarCrossref
19.
Ehrisman  J, Secord  AA, Berchuck  A,  et al.  Performance of sentinel lymph node biopsy in high-risk endometrial cancer.  Gynecol Oncol Rep. 2016;17:69-71.PubMedGoogle ScholarCrossref
20.
Cook  EF, Goldman  L.  Empiric comparison of multivariate analytic techniques: advantages and disadvantages of recursive partitioning analysis.  J Chronic Dis. 1984;37(9-10):721-731.PubMedGoogle ScholarCrossref
21.
Goldstraw  P, Crowley  J, Chansky  K,  et al; International Association for the Study of Lung Cancer International Staging Committee; Participating Institutions.  The IASLC Lung Cancer Staging Project: proposals for the revision of the TNM stage groupings in the forthcoming (seventh) edition of the TNM Classification of malignant tumours.  J Thorac Oncol. 2007;2(8):706-714.PubMedGoogle ScholarCrossref
22.
Davidson  R, MacKinnon  JG.  Bootstrap tests: How many bootstraps?  Econom Rev. 2000;19(1):55-68.Google ScholarCrossref
23.
Strobl  C, Malley  J, Tutz  G.  An introduction to recursive partitioning: rationale, application, and characteristics of classification and regression trees, bagging, and random forests.  Psychol Methods. 2009;14(4):323-348.PubMedGoogle ScholarCrossref
24.
Van der Laan  MJ, Pollard  KS.  A new algorithm for hybrid hierarchical clustering with visualization and the bootstrap.  J Stat Plan Inference. 2003;117(2):275-303.Google ScholarCrossref
25.
Breiman  L.  Bagging predictors.  Mach Learn. 1996;24(2):123-140. doi:10.1023/A:1018054314350Google Scholar
26.
Bauer  E, Kohavi  R.  An empirical comparison of voting classification algorithms: bagging, boosting, and variants.  Mach Learn. 1999;36(1):105. doi:10.1023/A:1007515423169Google ScholarCrossref
27.
Kebebew  E, Ituarte  PH, Siperstein  AE, Duh  QY, Clark  OH.  Medullary thyroid carcinoma: clinical characteristics, treatment, prognostic factors, and a comparison of staging systems.  Cancer. 2000;88(5):1139-1148.PubMedGoogle ScholarCrossref
28.
Yen  TW, Shapiro  SE, Gagel  RF, Sherman  SI, Lee  JE, Evans  DB.  Medullary thyroid carcinoma: results of a standardized surgical approach in a contemporary series of 80 consecutive patients.  Surgery. 2003;134(6):890-899.PubMedGoogle ScholarCrossref
29.
Yang  JH, Lindsey  SC, Camacho  CP,  et al.  Integration of a postoperative calcitonin measurement into an anatomical staging system improves initial risk stratification in medullary thyroid cancer.  Clin Endocrinol (Oxf). 2015;83(6):938-942.PubMedGoogle ScholarCrossref
Original Investigation
September 2017

Rethinking the Current American Joint Committee on Cancer TNM Staging System for Medullary Thyroid Cancer

Author Affiliations
  • 1Department of Surgery, Duke University Medical Center, Durham, North Carolina
  • 2Department of Biostatistics and Bioinformatics, Duke University, Durham, North Carolina
  • 3Duke Cancer Institute, Durham, North Carolina
  • 4Duke Clinical Research Institute, Durham, North Carolina
  • 5Deputy Editor, JAMA Surgery
JAMA Surg. 2017;152(9):869-876. doi:10.1001/jamasurg.2017.1665
Key Points

Question  How can the current American Joint Committee on Cancer (AJCC) TNM staging system for medullary thyroid cancer be improved to more accurately correlate with survival?

Findings  In a cohort study involving 3315 patients with medullary thyroid cancer, new TNM groupings were proposed that showed more distinct survival differences across TNM groups than was possible with the current AJCC TNM staging system.

Meaning  The current AJCC TNM staging system for medullary thyroid cancer appears to be less than optimal in distinguishing risk of mortality among stage groups, upstaging a significant number of patients to stage IV.

Abstract

Importance  Controversy exists around the American Joint Committee on Cancer (AJCC) TNM staging system for medullary thyroid cancer (MTC). Because of the rarity of the disease and limited available data, the staging system for MTC has been largely extrapolated from staging for differentiated thyroid cancer.

Objectives  To evaluate how well the current (seventh and eighth editions) AJCC TNM staging system correlates with survival for patients with MTC and to suggest a possible revision.

Design, Setting, and Participants  This population-based retrospective cohort analysis used the National Cancer Database to select patients aged 18 years or older diagnosed with MTC in the United States between 1998 and 2012. Patient demographic and tumor characteristics were assessed, and pathologic tumor stages were provided as T, N, and M categories. Recursive partitioning with bootstrapping was used to divide patients by TNM stages into groups with similar survival. The newly identified groupings were validated in a Surveillance, Epidemiology, and End Results (SEER) cohort. Data analysis was conducted between July 17, 2016, and November 11, 2016.

Main Outcomes and Measures  Overall survival and disease-specific survival.

Results  Of the 3315 patients with MTC included in the analysis, 1941 (58.6%) were women. The median (interquartile range) age was 54 (42-65) years, and 2839 (85.6%) self-reported their race/ethnicity as white. The current AJCC TNM staging system classified 941 of these patients (28.4%) as stage I, 907 (27.4%) as stage II, 424 (12.9%) as stage III, and 1043 (31.5%) as stage IV. Recursive partitioning analysis yielded 4 TNM groups: stage I (T1N0-1aM0, T2N0M0), stage II (T1N1bM0, T2N1a-bM0, and T3N0M0), stage III (T3N1a-bM0, T4N0-1bM0), and stage IV (T1-4N0-1bM1). Based on these proposed TNM groupings, 1797 of the 3315 patients (54.2%) were classified as stage I, 684 (20.6%) as stage II, 669 (20.2%) as stage III, and 165 (5.0%) as stage IV. Under the proposed TNM groupings, survival differences across TNM groups were more distinct than under the current AJCC TNM staging system. With stage I as the reference, the hazard ratios of the proposed TNM groupings and the current AJCC TNM staging system were 2.19 (95% CI, 1.37-3.12) vs 1.45 (95% CI, 1.09-1.92) for stage II, 4.20 (95% CI, 2.75-5.75) vs 2.17 (95% CI, 1.59-2.89) for stage III, and 10.97 (95% CI, 5.52-18.57) vs 5.33 (95% CI, 4.13-6.86) for stage IV. In a SEER cohort, the proposed TNM groupings were better at discriminating survival than was the current AJCC TNM staging system.

Conclusions and Relevance  The current AJCC TNM staging system for MTC appears to be less than optimal in distinguishing risk of mortality among stage groups, upstaging a significant number of patients to stage IV. The current AJCC TNM staging system could be improved with the new TNM groupings proposed here for more accurate risk stratification and potential treatment selection.

Introduction

Medullary thyroid cancer (MTC) is a neuroendocrine tumor of the parafollicular cells of the thyroid gland.1 Most MTC cases (75%) are sporadic; the remaining cases present as part of genetically inherited syndromes, such as multiple endocrine neoplasia types 2A and 2B and familial MTC.2 Medullary thyroid cancer is rare, accounting for only 2% to 4% of all new thyroid cancer cases.3,4

Medullary thyroid cancer is a clinically heterogeneous entity with marked differences in presentation, ranging from an indolent disease with survival approaching 100% to a very aggressive tumor with a mortal prognosis.5 Thus, accurate tumor staging is critical for the purposes of accurate risk stratification and treatment selection.

Considerable controversy exists regarding the current staging system for MTC.6,7Quiz Ref ID Because of the rarity of the disease and an overall paucity of data, the American Joint Committee on Cancer (AJCC) TNM staging system for MTC has largely been extrapolated from staging for differentiated thyroid cancer (DTC).8-14 However, MTC inherently differs from DTC in clinical presentation, treatment recommendations, and prognosis.1,2,4 Therefore, a TNM staging system that is based on cases of DTC may not be generalizable to patients with MTC.

We hypothesized that the current (seventh and eighth editions) AJCC TNM staging system for MTC might be improved by data from a large national cohort of patients.12,15 Our aim was to use a statistically sophisticated methodology to identify granular survival differences among patients with MTC to define more accurate staging groups.

Methods

The National Cancer Database (NCDB) was used as the main source of data for analysis because of its large sample size. Specifically, we selected patients 18 years or older diagnosed with MTC in the United States between 1998 and 2012. Because data on disease-specific survival were not available in the NCDB, the Surveillance, Epidemiology, and End Results (SEER) Program data set on a similar population of patients diagnosed between 2004 and 2012 was used to validate the NCDB data set. Data analysis was conducted between July 17, 2016, and November 11, 2016. This study was granted exempt status by the Duke University Medical Center Institutional Review Board because patients were deidentified.

NCDB Cohort

The NCDB is a joint program of the American Cancer Society and the American College of Surgeons. It contains data from more than 1500 accredited cancer programs, capturing more than 85% of all new thyroid cancer cases in the United States.16

Patients diagnosed with MTC were identified using International Classification of Diseases for Oncology, third edition17 codes 8345/3 and 8510/3. Variables such as age at diagnosis, race/ethnicity, sex, annual income, insurance status, and year of diagnosis were obtained from the NCDB. Pathologic tumor stages were provided as T, N, and M categories. Data on extent of surgery documented as lobectomy or total thyroidectomy were extracted from the NCDB.18 Overall survival and follow-up data were provided in the NCDB.

SEER Cohort

Similar to the NCDB cohort, the SEER cohort comprised patients with MTC who were identified using International Classification of Diseases for Oncology, third edition, codes 8345/3 and 8510/3. Variables included age at diagnosis, sex, race/ethnicity, and year of diagnosis. Treatment data included extent of surgery. Pathologic characteristics included tumor size and presence or absence of extrathyroidal extension. Pathologic T, N, and M categories were obtained from the SEER data set. Overall and disease-specific survival and follow-up data were obtained from the SEER database.

Cohort Selection

All patients with missing data on pathologic T, N, or M categories or survival were excluded. Patients who underwent removal of less than a thyroid lobe also were excluded.

Statistical Analysis
Recursive Partitioning Analysis

Recursive partitioning is a form of decision tree analysis that can classify a population into homogeneous subpopulations according to the association between a set of independent variables and a dependent variable. Recursive partitioning finds cuts or groupings of the independent values that best predict a dependent variable value. These splits or partitions of the data are done recursively to form a tree of decision rules until the desired fit is reached. The optimum splits are chosen from the set of all possible splits.19,20

Recursive partitioning was applied to the NCDB cohort to divide the patients according to T, N, and M categories into groups with similar overall survival (eFigure 1 in the Supplement).21 The minimum criterion for dividing a group was a 2-sided P ≤ .10. The groups defined by this method were then sorted by the survival proportion at 100 months and amalgamated into 4 homogeneous groups based on these proportions. The group with the highest survival proportion was defined as stage I, the second best survival group was stage II, the third best survival group was stage III, and the most compromised survival group was stage IV. These stages were used to predict survival by using a multivariable Cox proportional hazards model adjusted for age, sex, income level, insurance status, margin status, year of diagnosis, and hospital location and type. The variable year of diagnosis was included in the model to account for the potential effect of improvement in cancer detection and treatment over time. Bootstrapping (with 1000 replications) was used to estimate the hazard ratios and 95% CIs for covariates in the Cox proportional hazards model.22 In bootstrapping, statistical modeling and/or estimation is repeated several times on different data set “replicates,” which are each created by drawing a random sample of the same sample size with replacement from the original data set. This repetition results in an estimated sampling distribution. Use of this sampling distribution, rather than merely the original data set with an assumed distribution, offers a better estimation of the true population effect. In this study, each bootstrap replicate included creation of a partitioning tree, amalgamation into 4 stages, and estimation of hazard ratios. The frequency of assignment to stage I, stage II, stage III, or stage IV for each TNM group was recorded out of the total 1000 replications, and the most frequently assigned stage for each TNM group was used to define final staging (eTable 1 in the Supplement).23,24 Unadjusted and adjusted survival estimates were compared between the 4 newly identified TNM groups using the Kaplan-Meier and Cox proportional hazards methods.

The current AJCC TNM staging system for MTC (eTable 2 in the Supplement) was applied to the NCDB cohort, and unadjusted and adjusted survival were compared with models that included the newly identified TNM groupings proposed here.

External Validation

The proposed TNM groupings and the current AJCC TNM staging system were applied to the SEER cohort to examine overall survival and disease-specific survival estimates. The estimates obtained using the proposed TNM groupings were compared with those obtained using the current AJCC TNM staging system.

Analyses were performed using SAS, version 9.4 (SAS Institute Inc), and R, version 3.1.0 (R Project).

Results

A total of 3315 patients with MTC were included in the study. Of these, 1941 (58.6%) were women, the median (interquartile range) age was 54 (42-65) years, and 2839 (85.6%) self-reported their race/ethnicity as white. Table 1 describes the demographic, clinicopathologic, and treatment characteristics of the cohort. Overall, 1223 patients (36.9%) had pathologic T1 tumors, 1098 patients (33.1%) had T2, 562 patients (17%) had T3, and 432 patients (13%) had T4. In total, 1405 patients (42.4%) had either N1a (560 [16.9%]) or N1b (845 [25.5%]) disease. Few patients (165 [5.0%]) had distant metastatic disease (Table 1). Under the current AJCC TNM staging system, 941 patients (28.4%) had tumors deemed to be pathologic stage I, 907 (27.4%) stage II, 424 (12.9%) stage III, and 1043 (31.5%) stage IV.

Recursive Partitioning Analysis

Median (interquartile range) follow-up time was 68 (0-189) months. The recursive partitioning analysis and the splitting of the TNM groups on the basis of maximum overall survival differences are demonstrated in eFigure 2 in the Supplement. The initial partitioning analysis identified 10 TNM groups with similar within-group survival: (1) T1-2N0M0, (2) T1N1aM0, (3) T1N1bM0, (4) T2N1a-bM0, (5) T1-2N0-1bM1, (6) T3N0-1aM0, (7) T3N1bM0, (8) T4N0-1bM0, (9) T3-4N0-1aM1, and (10) T3-4N1bM1. Subsequent bootstrapping identified 4 distinct TNM groups: stage I (T1N0-1aM0, T2N0M0), stage II (T1N1bM0, T2N1a-bM0, and T3N0M0), stage III (T3N1a-bM0, T4N0-1bM0), and stage IV (T1-4N0-1bM1) (Figure 1). Under the proposed TNM groupings, of the 3315 patients, 1797 (54.2%) had stage I disease, 684 (20.6%) had stage II, 669 (20.2%) had stage III, and 165 (5.0%) had stage IV.

Survival Estimates

Under the current AJCC TNM staging system, of the 3315 patients, 941 (28.4%) had stage I and 1043 (31.5%) had stage IV disease compared with 1797 (54.2%) who had stage I and only 165 (5.0%) who had stage IV under the newly proposed groupings (eTable 2 in the Supplement). According to recursive partitioning, differences in survival between TNM groups appeared to be more distinct with the proposed TNM groupings than with the current AJCC TNM staging system (Figure 1). Quiz Ref IDBased on the proposed TNM groupings, 5-year overall survival for MTC was 94% for stage I, 86% for stage II, 69% for stage III, and 35% for stage IV. Quiz Ref IDAccording to the current AJCC TNM staging system, overall survival was 95% for stage I, 91% for stage II, 89% for stage III, and 68% for stage IV. After adjustment, TNM groups also were more distinct in overall survival using the proposed TNM groupings compared with the current AJCC TNM staging system. With stage I as the reference, the hazard ratios of the proposed TNM groupings and the current AJCC TNM staging system were 2.19 (95% CI, 1.37-3.12) vs 1.45 (95% CI, 1.09-1.92) for stage II, 4.20 (95% CI, 2.75-5.75) vs 2.17 (95% CI, 1.59-2.89) for stage III, and 10.97 (95% CI, 5.52-18.57) vs 5.33 (95% CI, 4.13-6.86) for stage IV (Table 2). The multivariable survival model in which tumor staging was represented by the proposed TNM groupings had the lowest Akaike information criterion, indicating a better statistical fit than the model that incorporated the current AJCC TNM staging system definition (eTable 3 in the Supplement).

External Validation in SEER

The current AJCC TNM staging system and the proposed TNM groupings were applied to a similar population of patients with MTC from SEER. This comparison demonstrated that the proposed TNM groupings’ algorithm could discriminate between TNM groups better than the current AJCC TNM staging system’s algorithm could. Quiz Ref IDThe proposed TNM groupings vs the current AJCC TNM staging system showed a 5-year overall survival of 92% vs 92% for stage I, 87% vs 91% for stage II, 81% vs 94% for stage III, and 33% vs 73% for stage IV (Figure 2A and B). Similarly, measurement of disease-specific survival between the proposed TNM groupings appeared to separate the survival curves better than the current AJCC TNM staging system could. The proposed TNM groupings compared with the current AJCC TNM staging system were associated with 5-year disease-specific survival rates of 100% vs 100% for stage I, 97% vs 99% for stage II, 89% vs 97% for stage III, and 40% vs 82% for stage IV (Figure 2C and D).

Sensitivity Analysis

Because the main analysis that generated the proposed TNM groupings was based on overall survival, we correlated TNM estimates of 5-year overall survival rates from NCDB with TNM estimates of 5-year disease-specific survival rates from SEER. There was strong correlation between overall survival and disease-specific survival in the cohort of patients with MTC (r = 0.98; P < .001) (Figure 3).

Discussion

This large study examined the appropriateness of the current AJCC TNM staging system to estimate survival and identify homogeneous survival groups in patients with MTC. It also provided possible revisions to sharpen the estimates of prognosis. The current AJCC TNM staging system for MTC can be informed by contemporary nationwide demographic and clinical data to provide an enhanced ability to stratify mortality risk between tumor stage groupings. Using recursive partitioning analysis, we identified 4 distinct TNM groups on the basis of within-group similarities in survival. The combination of recursive partitioning and bootstrapping provides an ensemble approach that acknowledges the inherent instability of individual partition trees. This approach, also referred to as “bagging,” has been shown to produce accurate predictions.25,26 The proposed TNM groupings appeared to be better correlated with survival and to better discriminate among different TNM groups when compared with the current AJCC TNM staging system. The proposed TNM groupings may better inform accurate prognostication and, potentially, treatment selection for patients with MTC, especially as these groupings are informed by national data.

Since the introduction of the AJCC TNM staging system, staging for MTC has been, in large part, aligned with staging algorithms for DTC.8-14 Given that MTC is inherently distinct from DTC, the generalizability of DTC staging to MTC may not be appropriate. Published data have focused on comparing the appropriateness of existing staging systems developed for MTC. In a single-institution study, Boostrom et al6 examined the role of the changes made for MTC from the fifth edition to the sixth edition of the AJCC TNM staging system, including the designation of lateral lymph node metastases (N1b disease) and reallocation of the presence of extrathyroidal extension (T4ab) from being part of stage III to stage IV. Boostrom et al compared the 2 editions of the AJCC TNM staging system as applied to 173 patients with MTC and, with a median follow-up of 82 months, reported that the sixth edition upstaged a significant number of patients to stage IV and that the overall survival for patients with stage IV MTC was 82% with the sixth edition but only 46% with the fifth edition. The authors recommended that all patients with lymph node metastases and without distant metastases be classified as having stage III disease.6 Kebebew et al27 compared different existing staging systems for MTC in 108 patients treated between 1960 and 1998. At a median follow-up of 60 months, there were 11 deaths related to MTC. Among the fifth edition of the AJCC TNM staging system and the European Organisation for Research and Treatment of Cancer, National Thyroid Cancer Treatment Cooperative Study, and SEER staging systems, the European Organisation for Research and Treatment of Cancer and AJCC TNM staging systems had the highest predictive scores. Of note, both the European Organisation for Research and Treatment of Cancer and the fifth edition of the AJCC TNM staging schema included distant metastases as part of their most advanced stage, while the most advanced stage of the National Thyroid Cancer Treatment Cooperative Study staging system included extrathyroidal extension and distant metastases.13 Although these studies provided data about the appropriateness of existing TNM staging systems, our study identified new TNM groupings that were objectively determined by their similarities in survival and then compared these groupings with the current AJCC TNM staging system. The recursive partitioning modeling we used allowed for more objective determination of TNM groups without the application of any preexisting assumptions about the importance of any of the TNM combinations.

Quiz Ref IDIn contrast to stage IV in the current AJCC TNM staging system, the stage IV in our proposed TNM groupings includes only patients with distant metastases; patients with T4 and/or N1b disease are no longer part of stage IV but have moved to stage II or stage III. The proposed TNM groupings emphasize the prognostic significance of distant metastasis in MTC and suggest that it should not be treated as equivalent to T4 or N1b disease. Studies have shown that distant metastases in MTC confer distinctly compromised survival when compared with the absence of distant metastases. Yen et al28 examined data from 80 patients with MTC who were treated at a single institution and found that death from the disease was uncommon in the absence of distant metastases. Disease-specific mortality was 30% for patients with distant metastases, whereas mortality was just 4% for patients without distant metastases. Under the proposed TNM groupings, patients with stage IV disease had a 5-year survival rate of 33% compared with 73% under the current AJCC TNM staging system. According to the current AJCC TNM staging system, 32% of patients were classified as having stage IV disease; however, according to the proposed TNM groupings, only 5% of the patients remained in stage IV. This finding indicates that the current AJCC TNM staging system potentially upstages a significant number of patients.

In the proposed TNM groupings, T4 disease in the absence of distant metastases is considered stage III and N1b disease in the setting of small (T12) tumors is part of stage II. These proposed changes represent a substantial departure from the current AJCC TNM staging system for MTC, but it is important to emphasize that cancer staging should be based on the ability to be accurately correlated with survival, maximizing the similarities in survival within groups and the differences in survival between groups; the proposed TNM groupings appear to do this better.

Strengths and Limitations

There are limitations to our study, including the possibility of coding errors and the retrospective nature of the cohort. Data on calcitonin levels and doubling time were not available for analysis. Some studies have suggested incorporating postoperative calcitonin levels to improve the accuracy of MTC staging7,29; however, it is unclear how to factor in the effect of postoperative calcitonin for the purposes of preoperative risk stratification and possibly treatment decisions. Data also suggest that elevation of calcitonin without clinical evidence of disease does not necessarily portend increased mortality.6,27 The strengths of the study include its large sample size and an adequate number of death events, use of the recursive partitioning method, application of bootstrapping, and external validation in SEER. These methods are reproducible. Any future information about prognostic indicators in MTC could be incorporated with minor changes, making our proposed staging system adaptable for the future.

Conclusions

This large study provides valuable information about the appropriateness of the current AJCC TNM staging system for MTC and suggests the potential to improve its applied accuracy. The clinical course of MTC can vary from a very indolent disease with excellent survival to a lethal tumor with dismal prognosis.5 Therefore, accurate tumor staging is critical for more accurate risk stratification and possibly treatment selection. The current AJCC TNM staging system for MTC appears to be less than optimal in discriminating risk of mortality among disease stage groups, upstaging a significant number of patients to stage IV. Our proposed TNM groupings appear to be better at predicting survival than the current AJCC TNM staging system. Cancer staging plays a critical role in cancer management and communication about disease prognosis among clinicians and between physicians and their patients. Therefore, the accuracy of MTC staging could have important implications for treatment decisions and ultimately for patient outcomes.

Back to top
Article Information

Corresponding Author: Julie A. Sosa, MD, MA, Department of Surgery, Duke University Medical Center, PO Box 2945, Durham, NC 27710 (julie.sosa@dm.duke.edu).

Accepted for Publication: March 12, 2017.

Published Online: June 21, 2017. doi:10.1001/jamasurg.2017.1665

Author Contributions: Dr Adam and Mrs Thomas had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: All authors.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Adam.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Adam, Thomas, Hyslop.

Obtained funding: Sosa.

Administrative, technical, or material support: Adam, Roman, Sosa.

Study supervision: Adam, Roman, Hyslop, Sosa.

Conflict of Interest Disclosures: Dr Sosa reported being a member of the Data Monitoring Committee of the Medullary Thyroid Cancer Consortium Registry and supported by Novo Nordisk, GlaxoSmithKline, AstraZeneca, and Eli Lilly. No other disclosures were reported.

Funding/Support: This study was funded by grant P30CA014236 from the National Institutes of Health.

Role of the Funder/Sponsor: The funding source had no role in the collection, management, analysis, and interpretation of the data; design and conduct of the study; preparation, review, and approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: Dr Sosa is a deputy editor of JAMA Surgery, but she was not involved in any of the decisions regarding review of the manuscript or its acceptance. A portion of the data used in the study was derived from a deidentified National Cancer Database file. The American College of Surgeons and the Commission on Cancer have not verified and are not responsible for the analytic or statistical methodology used or for the conclusions drawn from these data by the investigators.

Meeting Presentation: A portion of the data used in the study was presented at the Endocrine Society’s 98th Annual Meeting and Expo; April 2, 2016; Boston, Massachusetts.

Additional Contributions: Kevin L. Anderson Jr, BS, Duke University School of Medicine, helped edit the revised version of the manuscript. He was not compensated for his contribution.

References
1.
Williams  ED.  Histogenesis of medullary carcinoma of the thyroid.  J Clin Pathol. 1966;19(2):114-118.PubMedGoogle ScholarCrossref
2.
Wells  SA  Jr, Asa  SL, Dralle  H,  et al; American Thyroid Association Guidelines Task Force on Medullary Thyroid Carcinoma.  Revised American Thyroid Association guidelines for the management of medullary thyroid carcinoma.  Thyroid. 2015;25(6):567-610.PubMedGoogle ScholarCrossref
3.
Tuttle  RM, Ball  DW, Byrd  D,  et al; National Comprehensive Cancer Network.  Medullary carcinoma.  J Natl Compr Canc Netw. 2010;8(5):512-530.PubMedGoogle ScholarCrossref
4.
Jin  LX, Moley  JF.  Surgery for lymph node metastases of medullary thyroid carcinoma: A review.  Cancer. 2016;122(3):358-366.PubMedGoogle ScholarCrossref
5.
Roman  S, Lin  R, Sosa  JA.  Prognosis of medullary thyroid carcinoma: demographic, clinical, and pathologic predictors of survival in 1252 cases.  Cancer. 2006;107(9):2134-2142.PubMedGoogle ScholarCrossref
6.
Boostrom  SY, Grant  CS, Thompson  GB,  et al.  Need for a revised staging consensus in medullary thyroid carcinoma.  Arch Surg. 2009;144(7):663-669.PubMedGoogle ScholarCrossref
7.
Tuttle  RM, Ganly  I.  Risk stratification in medullary thyroid cancer: moving beyond static anatomic staging.  Oral Oncol. 2013;49(7):695-701.PubMedGoogle ScholarCrossref
8.
Beahrs  O, Carr  DT, Rubin  P; American Joint Committee for Cancer Staging and End Results Reporting.  Manual for Staging of Cancer. Philadelphia, PA: J B Lippincott Co; 1977.
9.
Beahrs  O, Henson  DE, Hutter  RVP, Kennedy  B; American Joint Committee on Cancer.  Manual of Staging of Cancer. 4th ed. Philadelphia, PA: J B Lippincott Co; 1992.
10.
Beahrs  O, Henson  D, Hutter  RVP, Myers  MH, eds; American Joint Committee on Cancer.  Manual of Staging of Cancer. 3rd ed. Philadelphia: J B Lippincott Co; 1988.
11.
Beahrs  O, Myers  M; American Joint Committee on Cancer.  Manual of Staging of Cancer. 2nd ed. Philadelphia: J B Lippincott Co; 1983.
12.
Edge  S, Byrd  D, Compton  C, Fritz  A, Greene  F, Trotti  A; American Joint Committee on Cancer.  AJCC Cancer Staging Manual. 7th ed. New York, NY: Springer; 2010.
13.
Fleming  I, Cooper  J, Henson  D,  et al; American Joint Committee on Cancer.  AJCC Cancer Staging Manual. 5th ed. Philadelphia, PA: Lippincott-Raven; 1997.
14.
Greene  F, Page  D, Fleming  I,  et al; American Joint Committee on Cancer.  AJCC Cancer Staging Manual. 6th ed. New York, NY: Springer; 2002.Crossref
15.
Amin  M, Edge  S, Greene  F,  et al.  AJCC Cancer Staging Manual. 8th ed. Cham, Switzerland: Springer International Publishing; 2017.Crossref
16.
Raval  MV, Bilimoria  KY, Stewart  AK, Bentrem  DJ, Ko  CY.  Using the NCDB for cancer care improvement: an introduction to available quality assessment tools.  J Surg Oncol. 2009;99(8):488-490.PubMedGoogle ScholarCrossref
17.
Fritz  A, Percy  C, Jack  A,  et al, eds.  International Classification of Diseases for Oncology. 3rd ed. Geneva, Switzerland: World Health Organization; 2000.
18.
Adam  MA, Pura  J, Gu  L,  et al.  Extent of surgery for papillary thyroid cancer is not associated with survival: an analysis of 61,775 patients.  Ann Surg. 2014;260(4):601-605.PubMedGoogle ScholarCrossref
19.
Ehrisman  J, Secord  AA, Berchuck  A,  et al.  Performance of sentinel lymph node biopsy in high-risk endometrial cancer.  Gynecol Oncol Rep. 2016;17:69-71.PubMedGoogle ScholarCrossref
20.
Cook  EF, Goldman  L.  Empiric comparison of multivariate analytic techniques: advantages and disadvantages of recursive partitioning analysis.  J Chronic Dis. 1984;37(9-10):721-731.PubMedGoogle ScholarCrossref
21.
Goldstraw  P, Crowley  J, Chansky  K,  et al; International Association for the Study of Lung Cancer International Staging Committee; Participating Institutions.  The IASLC Lung Cancer Staging Project: proposals for the revision of the TNM stage groupings in the forthcoming (seventh) edition of the TNM Classification of malignant tumours.  J Thorac Oncol. 2007;2(8):706-714.PubMedGoogle ScholarCrossref
22.
Davidson  R, MacKinnon  JG.  Bootstrap tests: How many bootstraps?  Econom Rev. 2000;19(1):55-68.Google ScholarCrossref
23.
Strobl  C, Malley  J, Tutz  G.  An introduction to recursive partitioning: rationale, application, and characteristics of classification and regression trees, bagging, and random forests.  Psychol Methods. 2009;14(4):323-348.PubMedGoogle ScholarCrossref
24.
Van der Laan  MJ, Pollard  KS.  A new algorithm for hybrid hierarchical clustering with visualization and the bootstrap.  J Stat Plan Inference. 2003;117(2):275-303.Google ScholarCrossref
25.
Breiman  L.  Bagging predictors.  Mach Learn. 1996;24(2):123-140. doi:10.1023/A:1018054314350Google Scholar
26.
Bauer  E, Kohavi  R.  An empirical comparison of voting classification algorithms: bagging, boosting, and variants.  Mach Learn. 1999;36(1):105. doi:10.1023/A:1007515423169Google ScholarCrossref
27.
Kebebew  E, Ituarte  PH, Siperstein  AE, Duh  QY, Clark  OH.  Medullary thyroid carcinoma: clinical characteristics, treatment, prognostic factors, and a comparison of staging systems.  Cancer. 2000;88(5):1139-1148.PubMedGoogle ScholarCrossref
28.
Yen  TW, Shapiro  SE, Gagel  RF, Sherman  SI, Lee  JE, Evans  DB.  Medullary thyroid carcinoma: results of a standardized surgical approach in a contemporary series of 80 consecutive patients.  Surgery. 2003;134(6):890-899.PubMedGoogle ScholarCrossref
29.
Yang  JH, Lindsey  SC, Camacho  CP,  et al.  Integration of a postoperative calcitonin measurement into an anatomical staging system improves initial risk stratification in medullary thyroid cancer.  Clin Endocrinol (Oxf). 2015;83(6):938-942.PubMedGoogle ScholarCrossref
×