Abstract
Background
Many countries have national guidelines for the management of differentiated thyroid cancer (DTC), including a risk stratification system to predict recurrence of disease. Studies whether these guidelines could also have relevance, beyond their original design, in predicting survival are lacking. Additionally, no studies evaluated these international guidelines in the same population, nor compared them with the TNM system. Therefore, we investigated the prognostic value of 6 stratification systems used by 10 international guidelines, and the TNM system with respect to predicting disease-specific survival (DSS).
Methods
We retrospectively studied adult patients with DTC from a Dutch university hospital. Patients were classified using the risk classification described in the British, Dutch, French, Italian, Polish, Spanish, European Society of Medical Oncology, European Thyroid Association, the 2009 and 2015 American Thyroid Association (ATA) guidelines, and the latest TNM system. DSS was analyzed using the Kaplan-Meier method, and the statistical model performance using the C-index, Akaike information criterion, Bayesian information criterion, and proportion of variance explained.
Results
We included 857 patients with DTC (79% papillary thyroid cancer, 21% follicular thyroid cancer). Median follow-up was 9 years, and 67 (7.8%) died because of DTC. The Dutch guideline had the worst statistical model performance, whereas the 2009 ATA/2014 British guideline had the best. However, the (adapted) TNM system outperformed all stratification systems.
Conclusions
In a European population of patients with DTC, of 10 international guidelines using 6 risk of recurrence stratification systems and 1 mortality-based stratification system, our optimized age-adjusted TNM system (8th edition) outperformed all other systems.
Keywords: papillary thyroid cancer, follicular thyroid cancer, survival, prognosis, ATA guideline, TNM
The worldwide incidence of differentiated thyroid cancer (DTC) has been steadily increasing over the past 2 decades (1, 2). Because mortality has remained stable to slightly increasing, a less aggressive therapeutic approach seems appropriate (1-4). To optimize the need for deescalation of therapy and follow-up strategies, different systems that predict the risk of recurrence and survival in patients with DTC have been proposed (4-7). Originally, the joint Union International Contre le Cancer and American Joint Committee on Cancer Tumor, Node, Metastasis (TNM) staging system has been designed to predict disease-specific survival (DSS) (6-8), whereas the American Thyroid Association (ATA) and European Thyroid Association (ETA) risk stratification systems have been designed to estimate the risk of disease recurrence (4, 5).
Although ETA and European Society for Medical Oncology consensus guidelines from 2006 and 2019, respectively, exist (5, 9), several European countries have their own guidelines for treating patients with DTC. The current British Thyroid Association guidelines use the 2009 ATA risk stratification system (10, 11), whereas the latest Polish, Spanish, and Italian thyroid cancer guidelines use the 2015 ATA risk stratification system (4, 12-14). The Dutch and French thyroid cancer guidelines use their own risk stratification system (15, 16). The risk stratification systems in these guidelines were originally designed to predict recurrence of disease.
Several studies evaluated whether the 2009 and/or 2015 ATA risk stratification systems could also have relevance, beyond their original design (ie, recurrence of disease), in predicting survival (17-27). Moreover, few studies also investigated overall survival (OS) based on the ATA initial risk stratification (28-30) and showed a clear separation between high-risk and low/intermediate-risk patients (28, 30). Next to this, also in clinical practice recurrence and mortality outcomes, might get mixed, leading to the suggestion that the initial risk stratification can be used to predict DSS. Currently, no studies exist that evaluated all the earlier mentioned risk stratification systems in the same population with respect to predicting DSS, nor compared these with the current 8th edition of the TNM system, which in several studies has come out as the most accurate system for predicting DSS (31-33). Moreover, separate staging of papillary thyroid cancer (PTC) vs follicular thyroid cancer (FTC) is infrequently done despite their different behavior (34-36). Consequently, it is worthwhile investigating to what extent the aforementioned risk stratification systems are also able to predict DSS, and how this compares with the TNM system in patients with DTC, and for PTC and FTC separately.
The aim of the current study is to investigate the prognostic value of 6 different stratification systems used by 10 International guidelines and the TNM system to predict DSS for DTC, and PTC vs FTC separately.
Materials and Methods
Study Population
We retrospectively included all patients, aged 18 years or older, who were diagnosed and/or treated for either PTC or FTC (including Hürthle cell carcinoma) between January 2002 and December 2017 at the Erasmus Medical Center, Rotterdam, The Netherlands. All the included patients underwent thyroid surgery and were treated according to the Dutch guidelines valid at the time of treatment (15). Briefly, this meant that, as a rule, all patients with thyroid cancer exceeding a diameter of 1 cm received a total thyroidectomy and subsequent radioactive iodine therapy. Patients with unifocal, nonmetastasized disease with a diameter of maximum 1 cm were treated by surgery only, which was limited to a hemithyroidectomy unless surgery was performed for other thyroid pathologies such as, but not limited to, goiter or Graves’ disease.
This database was previously used to evaluate and investigate potential improvements of the 8th edition of the TNM system as well as the risk stratification system of the 2015 ATA guidelines (35-40). Demographic, disease, treatment, and mortality characteristics were obtained from patient records. Time from diagnosis to last known follow-up, vital status, and, if applicable, date and cause of death were recorded. Cause of death was obtained from hospital or general practitioner records. Survival was defined as the time from the date of the initial diagnosis to either the date of the last-known follow-up, death, or end of study (December 2020), whichever occurred first. The study protocol was approved by the institutional review board of the Erasmus Medical Center.
Patients were classified using the risk classification described in the 2006 ETA guidelines (ETCG) (5), 2009 ATA guidelines (11), 2014 British Thyroid Cancer guidelines (BTCG) (10), 2015 Dutch Thyroid Cancer guidelines (DTCG) (15), 2015 ATA guidelines (ATCG) (4), 2017 French Thyroid Cancer guidelines (FTCG) (16), 2018 Italian Thyroid Cancer guidelines (ITCG) (13), 2018 Polish Thyroid Cancer guidelines (PTCG) (14), 2019 European Society for Medical Oncology guidelines (ESTCG) (9), and the 2019 Spanish Thyroid Cancer guidelines (STCG) (12). The ITCG, PTCG, and STCG use the risk classification of the 2015 ATA guidelines (4), and therefore the ATCG, ITCG, PTCG, and STCG will be referred to as ATCG in the remainder of this manuscript. Further, the BTCG uses the risk classification of the 2009 ATA guidelines (11), which are in the remainder are both referred to as BTCG. Further, patients were also classified according to the original 8th edition of the TNM classification system (8), and also by our recently adapted version of the TNM system employing different age cutoffs for PTC (50 years) and FTC (40 years) (36)). See Supplementary Tables S1 to S8 for a description of the mentioned risk stratification and TNM systems (41).
Statistical Analysis
For continuous variables, means and SDs, or medians with interquartile ranges were calculated. For categorical variables, absolute numbers with percentages were recorded. Differences in characteristics between PTC and FTC were assessed using the Student t test or χ2 test as appropriate.
Using the previously described risk stratification and TNM systems, DSS was analyzed using the Kaplan-Meier method and compared across risk categories/stages using the log-rank test and Cox proportional hazards models. To assess the statistical model performance of the risk stratification and TNM systems, we used the concordance index (Harrell C-index) (42, 43), proportion of variance explained (PVE) (44), Akaike information criterion (AIC) (45), and the Bayesian information criterion (BIC) (46). The C-index measures the discriminative power of a model and is a measure of goodness of fit. It ranges from 0.5 to 1.0, with 0.5 meaning the model predicts no better than random chance, and 1.0 being the perfect prediction model. The PVE measures the relative predictive accuracy of a Cox proportional hazards model. Further, the AIC and BIC measure the relative quality of a statistical model, and they provide the relative information lost when a statistical model is used to represent the true model. The model with the highest C-index and PVE and the lowest AIC and BIC is considered to be the best model for predicting outcomes. Therefore, using these 4 criteria, we aimed to find the risk stratification and/or TNM system that optimizes the statistical performance.
P values below 0.05 were considered significant. All analyses were performed using either SPSS Statistics for Windows (version 25.0) or the open source statistical software R (version 4.0.4) with package “survC1” for estimating the C-index and “surev” for estimating the PVE (47).
Results
Population Characteristics
A total of 897 patients fulfilled the inclusion criteria, of which 40 (4.6%) had insufficient information to determine their risk category. Table 1 and Supplementary Table S9 list the characteristics of the finally included 857 patients (41). Mean age was 48.7 years and 603 (70%) were women. PTC was present in 678 patients (79%), whereas the remaining 179 patients (21%) had FTC. The median available follow-up time was 106 months, and during follow-up, 134 patients (16%) died, of which 67 (7.8%) died from thyroid cancer. Furthermore, at baseline, 276 patients (32%) had lymph node metastases and 94 (11%) had distant metastases. Patients with FTC were significantly older (55.5 years vs 47.0 years; P < 0.001), had more often distant metastases (21% vs 8%; P < 0.001), and less often had lymph node metastases (12% vs 38%; P < 0.001).
Table 1.
Characteristics of the study population
| DTC (n = 857)a | PTC (n = 678)a | FTC (n = 179)a | P valueb | |
|---|---|---|---|---|
| Age at baseline (y) | 48.7 ± 16.2 | 47.0 ± 15.4 | 55.5 ± 17.2 | <0.001 |
| Sex | 0.363 | |||
| ȃMale | 254 (29.6) | 196 (28.9) | 58 (32.4) | |
| ȃFemale | 603 (70.4) | 482 (71.1) | 121 (67.6) | |
| T-stage | <0.001 | |||
| ȃT0 | 10 (1.2) | 8 (1.2) | 2 (1.1) | |
| ȃT1 | 326 (38.0) | 300 (44.2) | 26 (14.5) | |
| ȃT2 | 225 (26.3) | 167 (24.6) | 58 (32.4) | |
| ȃT3 | 221 (25.8) | 149 (22.0) | 72 (40.3) | |
| ȃT4 | 69 (8.1) | 51 (7.5) | 18 (10.1) | |
| Lymph node metastases | <0.001 | |||
| ȃNot present | 581 (67.8) | 423 (62.4) | 158 (88.3) | |
| ȃPresent | 276 (32.2) | 255 (37.6) | 21 (11.7) | |
| Distant metastases | <0.001 | |||
| ȃNot present | 763 (89.0) | 621 (91.6) | 141 (78.8) | |
| ȃPresent | 94 (11.0) | 56 (8.3) | 38 (21.2) | |
| Follow-up (mo) | 106 (61–164) | 106 (62–164) | 107 (59–164) | 0.178 |
| Vital status at end of follow-up | 723 (84.4) | 590 (87.0) | 133 (74.3) | |
| ȃAlive | 134 (15.6) | 88 (13.0) | 46 (25.7) | <0.001 |
| ȃDied (all causes) | <0.001 | |||
| ȃDied (thyroid cancer) | 67 (7.8) | 43 (6.3) | 24 (13.4) | 0.002 |
| Survival | ||||
| ȃ10-year OS (%) | 83.7 ± 1.5 | 86.2 ± 1.6 | 74.8 ± 3.6 | <0.001 |
| ȃ10-year DSS (%) | 90.8 ± 1.2 | 92.4 ± 1.2 | 84.8 ± 3.0 | 0.002 |
Significant P-values displayed in bold.
Abbreviations: DSS, disease-specific survival; DTC, differentiated thyroid cancer; FTC, follicular thyroid cancer; OS, overall survival; PTC, papillary thyroid cancer.
Values are means (± SD), medians (25-75 interquartile range), or numbers of patients (percentage of the respective population).
P value comparing PTC and FTC.
Risk and TNM Stage Classification
For the FTCG, patients were distributed evenly across all 3 stages, whereas for the BTCG the majority was either classified as low (40%) or intermediate (34%) risk. For the ATCG and ESTCG, the majority were classified as either low (43%) or high (37% and 39%, respectively) risk. The DTCG classified 299 patients (35%) as low risk, whereas the other 558 patients (65%) were high risk. Also, the ETCG classified the majority of the patients as high risk (58%). Further, using the TNM 8th edition, 621 patients (73%) were classified as stage I, 147 (17%) as stage II, 32 (4%) as stage III, and 57 (7%) as stage IV. Using the adapted version of the TNM system, fewer patients were classified as stage I (66% vs 73%); these patients were redistributed over the other 3 stages (see Table 2).
Table 2.
Stratification of the different guidelines
| DTC (n = 857)a | PTC (n = 678)a | FTC (n = 179)a | |
|---|---|---|---|
| ETCG | |||
| ȃLow | 361 (42.1) | 289 (42.6) | 72 (40.2) |
| ȃHigh | 496 (57.9) | 389 (57.4) | 107 (59.8) |
| BTCG | |||
| ȃLow | 339 (39.6) | 289 (42.6) | 50 (27.9) |
| ȃIntermediate | 291 (34.0) | 229 (33.8) | 62 (34.6) |
| ȃHigh | 227 (26.5) | 160 (23.6) | 67 (37.4) |
| ATCG | |||
| ȃLow | 370 (43.2) | 289 (42.6) | 81 (45.3) |
| ȃIntermediate | 167 (19.5) | 167 (24.6) | — |
| ȃHigh | 320 (37.3) | 222 (32.7) | 98 (54.7) |
| DTCG | |||
| ȃLow | 299 (34.9) | 253 (37.3) | 46 (25.7) |
| ȃHigh | 558 (65.1) | 425 (62.7) | 133 (74.3) |
| FTCG | |||
| ȃLow | 286 (33.4) | 253 (37.3) | 33 (16.4) |
| ȃIntermediate | 283 (33.0) | 231 (34.1) | 52 (29.1) |
| ȃHigh | 288 (33.6) | 194 (28.6) | 94 (52.5) |
| ESTCG | |||
| ȃLow | 370 (43.2) | 289 (42.6) | 81 (45.3) |
| ȃIntermediate | 153 (17.9) | 153 (22.6) | — |
| ȃHigh | 334 (39.0) | 236 (34.8) | 98 (54.7) |
| TNM 8th edition | |||
| ȃStage I | 621 (72.5) | 520 (76.7) | 101 (56.4) |
| ȃStage II | 147 (17.2) | 107 (15.8) | 40 (22.3) |
| ȃStage III | 32 (3.7) | 25 (3.7) | 7 (3.9) |
| ȃStage IV | 57 (6.7) | 26 (3.8) | 31 (17.3) |
| Adapted TNM | |||
| ȃStage I | 567 (66.2) | 483 (71.2) | 84 (46.9) |
| ȃStage II | 185 (21.6) | 136 (20.1) | 49 (27.4) |
| ȃStage III | 39 (4.6) | 31 (4.6) | 8 (4.5) |
| ȃStage IV | 66 (7.7) | 28 (4.1) | 38 (21.2) |
Abbreviations: ATCG, American Thyroid Association guidelines; BTCG, British Thyroid Cancer guidelines; DTC, differentiated thyroid cancer; DTCG, Dutch Thyroid Cancer guidelines; ESTCG, European Society for Medical Oncology guidelines; ETCG, European Thyroid Association cancer guidelines; FTC, follicular thyroid cancer; FTCG, French Thyroid Cancer guidelines; PTC, papillary thyroid cancer; TNM, Tumor, Node, Metastasis.
Values are numbers of patients (percentage of the respective population).
Survival Prediction
The 10-year DSS for DTC was 91% and was significantly higher for PTC than it was for FTC (92% vs 85%; P = 0.002). For DTC, and PTC and FTC separately, the BTCG, FTCG, ESTCG, and ATCG high-risk groups had a significant higher disease-specific mortality (DSM) than the low- and intermediate-risk groups. Comparing the low and intermediate groups, no significant differences were seen for all guidelines for either DTC, or PTC and FTC separately. Therefore, we additionally investigated whether combining the low and intermediate groups led to better results. For the BTCG, FTCG, ESTCG, and ATCG, a significant difference between the low/intermediate- and high-risk groups was seen for both PTC and FTC separately, and combined as DTC. Ten-year DSS percentages are shown in Table 3, and corresponding Kaplan-Meier curves in Fig. 1 and Supplemental Figs. 1 and 2 (41).
Table 3.
10-year DSS and mortality for the different guidelines
| DTC | PTC | FTC | ||||
|---|---|---|---|---|---|---|
| Deaths | 10-y DSS | Deaths | 10-y DSS | Deaths | 10-y DSS | |
| ETCG | ||||||
| ȃLow | 3/357 | 99.7 ± 0.3% | 1/287 | 100.0 ± 0.0% | 2/70 | 98.6 ± 1.4% |
| ȃHigh | 64/494 | 83.8 ± 2.0% | 42/387 | 86.4 ± 2.2% | 22/107 | 74.6 ± 5.0% |
| BTCG | ||||||
| ȃLow | 2/336 | 99.6 ± 0.4% | 1/287 | 100.0 ± 0.0% | 1/49 | 97.4 ± 2.5% |
| ȃIntermediate | 4/289 | 98.4 ± 0.9% | 2/228 | 98.5 ± 1.1% | 2/61 | 95.2 ± 3.4% |
| ȃHigh | 61/226 | 67.1 ± 3.9% | 40/159 | 69.9 ± 4.5% | 21/67 | 60.2 ± 7.5% |
| ATCG | ||||||
| ȃLow | 0/367 | 100.0 ± 0.0% | 0/287 | 100.0 ± 0.0% | 0/80 | 100.0 ± 0.0% |
| ȃIntermediate | 2/166 | 99.4 ± 0.6% | 2/166 | 99.4 ± 0.6% | — | — |
| ȃHigh | 65/318 | 74.8 ± 3.0% | 41/221 | 76.8 ± 3.6% | 24/97 | 70.7 ± 5.5% |
| DTCG | ||||||
| ȃLow | 1/295 | 100.0 ± 0.0% | 1/251 | 100.0 ± 0.0% | 0/44 | 100.0 ± 0.0% |
| ȃHigh | 66/556 | 85.6 ± 1.8% | 42/423 | 87.6 ± 2.0% | 24/133 | 79.6 ± 4.0% |
| FTCG | ||||||
| ȃLow | 1/283 | 100.0 ± 0.0% | 1/251 | 100.0 ± 0.0% | 0/32 | 100.0 ± 0.0% |
| ȃIntermediate | 2/281 | 99.2 ± 0.5% | 1/230 | 99.6 ± 0.4% | 1/51 | 97.9 ± 2.1% |
| ȃHigh | 64/287 | 72.5 ± 3.3% | 41/193 | 73.1 ± 4.1% | 23/94 | 71.3 ± 5.5% |
| ESTCG | ||||||
| ȃLow | 0/367 | 100.0 ± 0.0% | 0/287 | 100.0 ± 0.0% | 0/80 | 100.0 ± 0.0% |
| ȃIntermediate | 1/152 | 100.0 ± 0.0% | 1/152 | 100.0 ± 0.0% | — | — |
| ȃHigh | 66/332 | 75.6 ± 2.9% | 42/235 | 77.8 ± 3.4% | 24/97 | 70.7 ± 5.5% |
| TNM 8th edition | ||||||
| ȃStage I | 8/617 | 99.0 ± 0.5% | 6/518 | 98.9 ± 0.5% | 2/99 | 97.3 ± 1.9% |
| ȃStage II | 16/145 | 82.9 ± 4.1% | 10/105 | 84.0 ± 4.9% | 6/40 | 79.4 ± 7.6% |
| ȃStage III | 12/32 | 55.2 ± 11.8% | 11/25 | 49.2 ± 13.2% | 1/7 | 83.3 ± 15.2% |
| ȃStage IV | 31/57 | 35.7 ± 8.0% | 16/26 | 30.5 ± 11.2% | 15/31 | 41.0 ± 11.1% |
| Adapted TNM | ||||||
| ȃStage I | 4/563 | 99.6 ± 0.3% | 2/481 | 99.7 ± 0.3% | 2/82 | 97.0 ± 2.2% |
| ȃStage II | 14/183 | 88.4 ± 3.1% | 11/134 | 86.7 ± 4.0% | 3/49 | 92.2 ± 4.4% |
| ȃStage III | 15/39 | 55.2 ± 10.6% | 14/31 | 50.1 ± 11.6% | 1/8 | 85.7 ± 13.2% |
| ȃStage IV | 34/66 | 37.4 ± 7.7% | 16/28 | 35.8 ± 11.1% | 18/38 | 39.2 ± 10.3% |
Abbreviations: ATCG, American Thyroid Association guidelines; BTCG, British Thyroid Cancer guidelines; DSS, disease-specific survival; DTC, differentiated thyroid cancer; DTCG, Dutch Thyroid Cancer guidelines; ESTCG, European Society for Medical Oncology guidelines; ETCG, European Thyroid Association cancer guidelines; FTC, follicular thyroid cancer; FTCG, French Thyroid Cancer guidelines; PTC, papillary thyroid cancer; DSS, disease-specific survival; TNM, Tumor, Node, Metastasis.
Figure 1.
Kaplan-Meier curves for disease specific survival of the different guidelines and TNM systems for DTC. (A) ETCG, (B) BCTG, (C) ATCG, (D) DTCG, (E), FTCG, (F) ESTCG, (G) TNM 8th edition, and (H) adapted TNM.
With respect to the different guidelines, the BTCG had the best statistical model performance for DTC, and for PTC and FTC separately. On the other side of the spectrum, the DTCG had the worst statistical performance. Additionally, combining the low/intermediate-risk groups did not lead to a consistent improvement for the 8th guidelines involved. When comparing the different guidelines with the TNM system, the latter performed substantially better, especially our age-adapted TNM system (see Table 4 and Supplementary Table S10 (41)). This was the case for both PTC and FTC, and combined as DTC. Further, because the DTCG is the only guideline incorporating the presence of thyroglobulin antibodies into the risk classification system, we also evaluated the DTCG when omitting thyroglobulin antibodies as a factor. The latter resulted in an improved statistical performance, which was still worse than the other guidelines, however (see Supplementary Table S10 (41)).
Table 4.
Measures of model performance for disease-specific survival for the different guidelines
| DTC | PTC | FTC | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| C-index | AIC | BIC | PVE | C-index | AIC | BIC | PVE | C-index | AIC | BIC | PVE | |
| ETCG | 0.706 | 800.4 | 802.6 | 0.030 | 0.714 | 486.2 | 488.0 | 0.028 | 0.690 | 219.5 | 220.7 | 0.049 |
| BTCG | 0.848 | 716.6 | 718.8 | 0.164 | 0.861 | 425.5 | 427.3 | 0.179 | 0.795 | 202.8 | 204.0 | 0.134 |
| ATCG | 0.831 | 726.7 | 728.9 | 0.112 | 0.842 | 440.8 | 442.6 | 0.116 | 0.769 | 199.6 | 200.8 | 0.098 |
| DTCG | 0.681 | 804.9 | 807.1 | 0.026 | 0.686 | 494.1 | 495.9 | 0.018 | 0.647 | 220.1 | 221.3 | 0.048 |
| FTCG | 0.834 | 726.5 | 728.7 | 0.124 | 0.848 | 434.8 | 436.5 | 0.132 | 0.759 | 206.8 | 207.9 | 0.095 |
| ESTCG | 0.826 | 726.1 | 728.3 | 0.110 | 0.837 | 439.8 | 441.5 | 0.113 | 0.769 | 199.6 | 200.8 | 0.098 |
| TNM 8th edition | 0.865 | 694.7 | 696.9 | 0.290 | 0.866 | 411.7 | 413.5 | 0.319 | 0.834 | 195.9 | 197.1 | 0.211 |
| Adapted TNM | 0.892 | 678.5 | 680.7 | 0.307 | 0.908 | 402.9 | 404.7 | 0.327 | 0.835 | 189.8 | 191.0 | 0.232 |
The model with the highest C-index and PVE, and the lowest AIC and BIC is considered to be the best model for predicting outcomes. The best performing models are displayed in bold.
Abbreviations: AIC, Akaike information criterion; ATCG, American Thyroid Association guidelines; BIC, Bayesian information criterion; BTCG, British Thyroid Cancer guidelines; DTC, differentiated thyroid cancer; DTCG, Dutch Thyroid Cancer guidelines; ESTCG, European Society for Medical Oncology guidelines; ETCG, European Thyroid Association cancer guidelines; FTC, follicular thyroid cancer; FTCG, French Thyroid Cancer guidelines; PTC, papillary thyroid cancer; PVE, proportion of variance explained; TNM, Tumor, Node, Metastasis.
Discussion
This study shows that, in a population of patients with DTC, of 10 international guidelines using 6 different risk stratification systems, the risk stratification system of the 2009 ATA/2014 British guidelines performed best with respect to predicting DSS. However, the (age-adapted) TNM system outperformed all stratification systems of the different guidelines in predicting DSS.
We showed that of 10 different international guidelines, the 2009 ATA/2014 British Thyroid Association guidelines perform the best with respect to predicting DSS for both DTC, and PTC and FTC separately. Two earlier studies comparing the 2009 and 2015 ATA risk stratification systems showed mixed results; using disease-free survival (DFS), Suh et al showed a better performance of the 2009 version (21), whereas Lee et al showed superiority of the 2015 version (20). Both studies comprised South Korean patients with PTC, but the latter study had 5 times more patients. Our study shows better performance of the 2009 ATA risk stratification system, but our results are not comparable with the mentioned 2 studies because we used DSS instead of DFS. Kelly et al showed a significant difference between the low- and intermediate-risk group with respect to OS using the 2015 ATA risk stratification system (30), which is in correspondence with our results regarding DSS.
The DSM percentages for DTC of the BTCG risk groups were 0.3% in the low-, 1.4% in the intermediate-, and 27.0% in the high-risk groups, respectively. These were in line with percentages of other studies evaluating the 2009 ATA risk stratification system in patients with DTC (25, 27). DSM percentages for the 2015 ATA risk groups (low, 0%; intermediate, 1.2%; high, 20.4%) were slightly different from 2 earlier studies (Piccardo et al (28): low, 0.2%; intermediate, 3.0%; high 28%, and Shah et al (24): high, 15%). The study by Shah et al differed because it also included patients with poorly differentiated thyroid cancer and had shorter follow-up. The 2009 ATA risk stratification performed better than the 2015 version, which is probably caused by the higher number of patients classified as high risk using the 2015 version. No earlier studies evaluated the FTCG and ESTCG, in which DSM percentages were in line with the 2015 ATA version, whereas the DTCG and ETCG had lower DSM percentages in the high-risk group. The latter probably reflects misclassification by both the DTCG and ETCG of patients with a relatively low risk of dying as high risk. When omitting thyroglobulin antibodies as a factor from the DTCG, an improved statistical performance was seen, but this was still worse than the other guidelines. We also aimed to improve the other guidelines (except the ETCG) by combining the low/intermediate-risk groups, but this did not lead to consistent improvements.
We showed that the TNM system better predicts DSS than the risk stratification systems of the different guidelines. Our adapted version of the 8th edition of the TNM system with different age cutoffs for PTC and FTC performed best among all stratification systems (36). An earlier study compared the 2009 and 2015 ATA guidelines with the 7th and 8th TNM systems with respect to DFS in Korean patients with PTC, and showed that the 8th edition performed the best (21). Further, Abiri et al recently compared the 2009 ATA guidelines with the 7th edition of the TNM system with respect to OS, and showed superiority of the latter (29). These results are in line with our study and suggest that the TNM system is the optimal system for initial survival prediction. Further research is needed to determine if this also is the case for predicting response to therapy and recurrence. Next to this, because age is a major determinant in the TNM system, and recent studies showed improvement of the ATA risk stratification system when high-risk patients are differently staged based on age (35, 48), incorporating age into the risk stratification systems of the different guidelines may have potential to further improve these systems, possibly also with respect to predicting DSS.
Although the risk stratification systems were designed to predict disease recurrence, as mentioned earlier, both clinical practice and earlier research also suggested that the initial risk stratification can be used to predict DSS (28, 30). Our study indicates the superiority of our age-adjusted TNM system over the studied guidelines for predicting survival, and therefore one should refrain from using the different risk stratification systems to predict DSS.
Earlier studies showed that PTC and FTC should be staged as separate entities (35, 36); therefore, in the current study, we also evaluated PTC and FTC separately. Results for PTC and FTC in the current study were concordant with the results when combined as DTC, and of the investigated guidelines, the 2009 ATA/2014 British guideline performed the best with respect to predicting DSS but was outperformed by the age-adapted TNM system. Our adapted TNM system uses different age cutoffs for PTC and FTC, substantiating the importance of staging of PTC and FTC separately.
Strengths of our study include that it has a substantial proportion of patients with advanced disease stages, and especially those with distant metastases at presentation, and therefore there were also sufficient patients in these stages to be able to perform the analyses. Furthermore, our population included a substantial proportion of patients with FTC, which enabled us to investigate PTC and FTC separately in a robust manner. A possible limitation of our study is its retrospective nature, with patients being reclassified using the available, unfortunately not always complete, (clinical) information of the different criteria of the risk stratification systems. Additionally, it is possible that an (unknown) proportion of patients with DTC would in present times be classified otherwise (eg, noninvasive follicular thyroid neoplasm with papillary like nuclear features). Unfortunately, the retrospective nature of our study precludes any ascertainment in this respect. Further, another possible limitation of the study is that patients were recruited from a single tertiary university hospital, which might attract patients with more aggressive disease (as indicated by the high mortality rate), especially FTC, because of the availability of advanced treatments. Finally, the lack of detailed follow-up information precluded assessment of response to therapy, recurrence, and DFS in the current study, which are, as described earlier, the outcomes the risk stratification systems of the different guidelines were originally designed for. Therefore, further studies are needed regarding the performance of these risk stratification systems with respect to response to therapy and recurrence.
Conclusion
The present study shows that in patients with DTC, including a large set of patients with FTC, our age-adjusted TNM system outperforms the 6 stratification systems used by 10 international guidelines. These results hold true for both PTC and FTC, and combined as DTC. Therefore, in clinical practice, the TNM staging system is most suited to predict DSS, and therefore one should refrain from using the different risk stratification systems to predict DSS. Further research is needed to determine the best risk strategy across the different stratification systems with respect to predicting response to therapy and recurrence. Combining the current study with such future studies may help to advance to a uniform international guideline for DTC.
Abbreviations
- ATA
American Thyroid Association
- AIC
Akaike information criterion
- ATCG
American Thyroid Association guidelines
- BIC
Bayesian information criterion
- BTCG
British Thyroid Cancer guidelines
- DFS
disease-free survival
- DSM
disease-specific mortality
- DSS
disease-specific survival
- DTC
differentiated thyroid cancer
- DTCG
Dutch Thyroid Cancer guidelines
- ESTCG
European Society for Medical Oncology guidelines
- ETA
European Thyroid Association
- ETCG
European Thyroid Association cancer guidelines
- FTC
follicular thyroid cancer
- FTCG
French Thyroid Cancer guidelines
- ITCG
Italian Thyroid Cancer guidelines
- OS
overall survival
- PTC
papillary thyroid cancer
- PTCG
Polish Thyroid Cancer guidelines
- PVE
proportion of variance explained
- STCG
Spanish Thyroid Cancer guidelines
- TNM
Tumor, Node, Metastasis
Contributor Information
Evert F S van Velsen, Academic Center for Thyroid Diseases, Department of Internal Medicine, Erasmus Medical Center, 3015 CE, Rotterdam, The Netherlands.
Robin P Peeters, Academic Center for Thyroid Diseases, Department of Internal Medicine, Erasmus Medical Center, 3015 CE, Rotterdam, The Netherlands.
Merel T Stegenga, Academic Center for Thyroid Diseases, Department of Internal Medicine, Erasmus Medical Center, 3015 CE, Rotterdam, The Netherlands.
Folkert J van Kemenade, Academic Center for Thyroid Diseases, Department of Pathology, Erasmus Medical Center, Rotterdam, 3015 CE, The Netherlands.
Tessa M van Ginhoven, Erasmus MC Cancer Institute, Department of Surgical Oncology and Gastrointestinal Surgery, Erasmus Medical Center, 3015 CE, Rotterdam, The Netherlands.
Mathé van Balkum, Academic Center for Thyroid Diseases, Department of Internal Medicine, Erasmus Medical Center, 3015 CE, Rotterdam, The Netherlands.
Frederik A Verburg, Academic Center for Thyroid Diseases, Department of Radiology and Nuclear Medicine, Erasmus Medical Center, 3015 CE, Rotterdam, The Netherlands.
W Edward Visser, Academic Center for Thyroid Diseases, Department of Internal Medicine, Erasmus Medical Center, 3015 CE, Rotterdam, The Netherlands.
Funding
This research did not receive any specific grant from any funding agency in the public, commercial or not-for-profit sector.
Author Contributions
E.V.V., F.A.V., and W.E.V designed the current study. E.V.V., W.E.V., M.T.S., M.V.B., and R.P.P. created the original databases to collect the clinical data. E.V.V. conducted the statistical analyses and wrote the initial version of the manuscript. All authors reviewed and revised the manuscript to improve its intellectual and technical content.
Disclosures
F.A.V. has received consultancy fees from Sanofi and EISAI as well as speaker honoraria from Sanofi and research support from EISAI. R.P.P. received teaching fees from Sanofi and Bayer. The other authors declare no conflicts of interest and that no competing financial interests exist.
Data Availability
All datasets generated during and analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.
References
- 1. La Vecchia C, Malvezzi M, Bosetti C, et al. Thyroid cancer mortality and incidence: a global overview. Int J Cancer. 2015;136(9):2187‐2195. [DOI] [PubMed] [Google Scholar]
- 2. Davies L, Welch HG. Increasing incidence of thyroid cancer in the United States, 1973-2002. JAMA. 2006; 295(18):2164‐2167. [DOI] [PubMed] [Google Scholar]
- 3. Lim H, Devesa SS, Sosa JA, Check D, Kitahara CM. Trends in thyroid cancer incidence and mortality in the United States, 1974–2013. JAMA. 2017; 317(13):1338‐1348. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Haugen BR, Alexander EK, Bible KC, et al. 2015 American Thyroid Association management guidelines for adult patients with thyroid nodules and differentiated thyroid cancer: the American Thyroid Association guidelines task force on thyroid nodules and differentiated thyroid cancer. Thyroid. 2016; 26(1):1‐133. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Pacini F, Schlumberger M, Dralle H, et al. European consensus for the management of patients with differentiated thyroid carcinoma of the follicular epithelium. Eur J Endocrinol. 2006;154(6):787‐803. [DOI] [PubMed] [Google Scholar]
- 6. Tuttle RM, Haugen B, Perrier ND. Updated American Joint Committee on cancer/tumor-node-metastasis staging system for differentiated and anaplastic thyroid cancer (eighth edition): what changed and why? Thyroid. 2017;27(6):751‐756. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Edge SB, Compton CC. The American Joint Committee on cancer: the 7th edition of the AJCC cancer staging manual and the future of TNM. Ann Surg Oncol. 2010;17(6):1471‐1474. [DOI] [PubMed] [Google Scholar]
- 8. Amin MB, Greene FL, Edge SB, et al. The Eighth Edition AJCC cancer Staging Manual: continuing to build a bridge from a population-based to a more “personalized” approach to cancer staging. CA Cancer J Clin. 2017;67(2):93‐99. [DOI] [PubMed] [Google Scholar]
- 9. Filetti S, Durante C, Hartl D, et al. Thyroid cancer: ESMO clinical practice guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2019;30(12):1856‐1883. [DOI] [PubMed] [Google Scholar]
- 10. Perros P, Boelaert K, Colley S, et al. Guidelines for the management of thyroid cancer. Clin Endocrinol (Oxf). 2014;81(Suppl 1):1‐122. [DOI] [PubMed] [Google Scholar]
- 11. American Thyroid Association Guidelines Taskforce on Thyroid Nodules and Differentiated Thyroid Cancer, Cooper DS, Doherty GM, Haugen BR, et al. Revised American Thyroid Association management guidelines for patients with thyroid nodules and differentiated thyroid cancer. Thyroid. 2009;19(11):1167‐1214. [DOI] [PubMed] [Google Scholar]
- 12. Gallardo E, Medina J, Sanchez JC, et al. SEOM clinical guideline thyroid cancer (2019). Clin Transl Oncol. 2020;22(2):223‐235. [DOI] [PubMed] [Google Scholar]
- 13. Pacini F, Basolo F, Bellantone R, et al. Italian consensus on diagnosis and treatment of differentiated thyroid cancer: joint statements of six Italian societies. J Endocrinol Invest. 2018;41(7):849‐876. [DOI] [PubMed] [Google Scholar]
- 14. Jarzab B, Dedecjus M, Slowinska-Klencka D, et al. Guidelines of Polish national societies diagnostics and treatment of thyroid carcinoma. 2018 update. Endokrynol Pol. 2018; 69(1):34‐74. [DOI] [PubMed] [Google Scholar]
- 15. 2015 Dutch Thyroid Cancer Guidelines . 2015 [updated 2015-02-16]. https://www.oncoline.nl/schildkliercarcinoom.
- 16. Zerdoud S, Giraudet AL, Leboulleux S, et al. Radioactive iodine therapy, molecular imaging and serum biomarkers for differentiated thyroid cancer: 2017 guidelines of the French societies of nuclear medicine, endocrinology, pathology, biology, endocrine surgery and head and neck surgery. Ann Endocrinol (Paris). 2017;78(3):162‐175. [DOI] [PubMed] [Google Scholar]
- 17. Tuttle RM, Tala H, Shah J, et al. Estimating risk of recurrence in differentiated thyroid cancer after total thyroidectomy and radioactive iodine remnant ablation: using response to therapy variables to modify the initial risk estimates predicted by the new American Thyroid Association staging system. Thyroid. 2010;20(12):1341‐1349. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Shen FC, Hsieh CJ, Huang IC, Chang YH, Wang PW. Dynamic risk estimates of outcome in Chinese patients with well-differentiated thyroid cancer after total thyroidectomy and radioactive iodine remnant ablation. Thyroid. 2017;27(4):531‐536. [DOI] [PubMed] [Google Scholar]
- 19. Pitoia F, Bueno F, Urciuoli C, Abelleira E, Cross G, Tuttle RM. Outcomes of patients with differentiated thyroid cancer risk-stratified according to the American Thyroid Association and Latin American Thyroid Society risk of recurrence classification systems. Thyroid. 2013;23(11):1401‐1407. [DOI] [PubMed] [Google Scholar]
- 20. Lee SG, Lee WK, Lee HS, et al. Practical performance of the 2015 American Thyroid Association guidelines for predicting tumor recurrence in patients with papillary thyroid cancer in South Korea. Thyroid. 2017;27(2):174‐181. [DOI] [PubMed] [Google Scholar]
- 21. Suh S, Kim YH, Goh TS, et al. Outcome prediction with the revised American Joint Committee on Cancer Staging system and American Thyroid Association guidelines for thyroid cancer. Endocrine. 2017;58(3):495‐502. [DOI] [PubMed] [Google Scholar]
- 22. Grani G, Zatelli MC, Alfo M, et al. Real-World performance of the American Thyroid Association risk estimates in predicting 1-year differentiated thyroid cancer outcomes: a prospective multicenter study of 2000 patients. Thyroid. 2021;31(2):264‐271. [DOI] [PubMed] [Google Scholar]
- 23. Vaisman F, Momesso D, Bulzico DA, et al. Spontaneous remission in thyroid cancer patients after biochemical incomplete response to initial therapy. Clin Endocrinol (Oxf). 2012;77(1):132‐138. [DOI] [PubMed] [Google Scholar]
- 24. Shah S, Boucai L. Effect of age on response to therapy and mortality in patients with thyroid cancer at high risk of recurrence. J Clin Endocrinol Metab. 2018;103(2):689‐697. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Hassan A, Razi M, Riaz S, et al. Survival analysis of papillary thyroid carcinoma in relation to stage and recurrence risk: a 20-year experience in Pakistan. Clin Nucl Med. 2016;41(8):606‐613. [DOI] [PubMed] [Google Scholar]
- 26. Kowalska A, Walczyk A, Palyga I, et al. The delayed risk stratification system in the risk of differentiated thyroid cancer recurrence. PLoS One. 2016;11(4):e0153242. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Cano-Palomares A, Castells I, Capel I, et al. Response to initial therapy of differentiated thyroid cancer predicts the long-term outcome better than classical risk stratification systems. Int J Endocrinol. 2014;2014:591285 . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Piccardo A, Siri G, Raffa S, et al. How to better stratify the risk of differentiated thyroid carcinomas: the key role of radioactive iodine therapy, age, and gender. Eur J Nucl Med Mol Imaging. 2021;48(3):822‐830. [DOI] [PubMed] [Google Scholar]
- 29. Abiri A, Pang J, Prasad KR, et al. Prognostic utility of tumor stage vs American Thyroid Association risk class in thyroid cancer. Laryngoscope. 2023;133(1):205-211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Kelly A, Barres B, Kwiatkowski F, et al. Age, thyroglobulin levels and ATA risk stratification predict 10-year survival rate of differentiated thyroid cancer patients. PLoS One. 2019;14(8):e0221298. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Momesso DP, Tuttle RM. Update on differentiated thyroid cancer staging. Endocrinol Metab Clin North Am. 2014;43(2):401‐421. [DOI] [PubMed] [Google Scholar]
- 32. Tanase K, Thies ED, Mäder U, Reiners C, Verburg FA. The TNM system (version 7) is the most accurate staging system for the prediction of loss of life expectancy in differentiated thyroid cancer. Clin Endocrinol (Oxf). 2016;84(2):284‐291. [DOI] [PubMed] [Google Scholar]
- 33. Verburg FA, Mader U, Kruitwagen CL, Luster M, Reiners C. A comparison of prognostic classification systems for differentiated thyroid carcinoma. Clin Endocrinol (Oxf). 2010;72(6):830‐838. [DOI] [PubMed] [Google Scholar]
- 34. D'Avanzo A, Treseler P, Ituarte PH, et al. Follicular thyroid carcinoma: histology and prognosis. Cancer. 2004;100(6):1123‐1129. [DOI] [PubMed] [Google Scholar]
- 35. van Velsen EFS, Peeters RP, Stegenga MT, et al. The influence of age on disease outcome in 2015 ATA high-risk differentiated thyroid cancer patients. Eur J Endocrinol. 2021;185(3):421‐429. [DOI] [PubMed] [Google Scholar]
- 36. van Velsen EFS, Visser WE, Stegenga MT, et al. Finding the optimal age cutoff for the UICC/AJCC TNM staging system in patients with papillary or follicular thyroid cancer. Thyroid. 2021;31(7):1041‐1049. [DOI] [PubMed] [Google Scholar]
- 37. van Velsen EFS, Stegenga MT, van Kemenade FJ, et al. Comparing the prognostic value of the Eighth Edition of the American Joint Committee on cancer/tumor node metastasis staging system between papillary and follicular thyroid cancer. Thyroid. 2018;28(8):976‐981. [DOI] [PubMed] [Google Scholar]
- 38. van Velsen EFS, Stegenga MT, van Kemenade FJ, et al. Evaluating the 2015 American Thyroid Association risk stratification system in high-risk papillary and follicular thyroid cancer patients. Thyroid. 2019;29(8):1073‐1079. [DOI] [PubMed] [Google Scholar]
- 39. van Velsen EFS, Stegenga MT, van Kemenade FJ, et al. Evaluation of the 2015 ATA guidelines in patients with distant metastatic differentiated thyroid cancer. J Clin Endocrinol Metab. 2020;105(3):e457‐e465. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. van Velsen EFS, Peeters RP, Stegenga MT, et al. Evaluating the use of a two-step age-based cutoff for the UICC/AJCC TNM staging system in patients with papillary or follicular thyroid cancer. Eur J Endocrinol. 2022;186(3):389‐397. [DOI] [PubMed] [Google Scholar]
- 41. van Velsen EFS, Peeters RP, Stegenga MT, et al. Evaluating disease specific survival prediction of risk stratification and TNM systems in differentiated thyroid cancer—Supplemental material. 2022. https://figshare.com/articles/journal_contribution/_/21505254. [DOI] [PMC free article] [PubMed]
- 42. Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med. 1996;15(4):361‐387. [DOI] [PubMed] [Google Scholar]
- 43. Harrell FE Jr, Lee KL, Califf RM, Pryor DB, Rosati RA. Regression modelling strategies for improved prognostic prediction. Stat Med. 1984; 3(2):143‐152. [DOI] [PubMed] [Google Scholar]
- 44. Lusa L, Miceli R, Mariani L. Estimation of predictive accuracy in survival analysis using R and S-PLUS. Comput Methods Programs Biomed. 2007;87(2):132‐137. [DOI] [PubMed] [Google Scholar]
- 45. Akaike H. A new look at the statistical model identification. IEEE Trans Automat Control. 1974;19(6):716‐723. [Google Scholar]
- 46. Schwarz G. Estimating the dimension of a model. Ann. Statist. 1978;2:461‐464. [Google Scholar]
- 47. Team RC . R: A language and environment for statistical computing. R Foundation for Statistical Computing; 2013. [Google Scholar]
- 48. Trimboli P, Piccardo A, Signore A, et al. Patient age is an independent risk factor of relapse of differentiated thyroid carcinoma and improves the performance of the American Thyroid Association stratification system. Thyroid. 2020;30(5):713‐719. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
All datasets generated during and analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.
