Predicting Outcomes in Sporadic and Hereditary Medullary Thyroid Carcinoma over Two Decades

Introduction

Medullary thyroid cancer (MTC) is a neuroendocrine tumor that accounts for up to 5% of all thyroid cancer cases in the United States (1). MTC is different from the other more common types of thyroid cancers (papillary and follicular) because it originates from the parafollicular C cells of the thyroid gland. MTC often spreads to lymph nodes (LNs) in the neck or metastasizes to other organs (1). In 25% of cases, this cancer is inherited, and in some of the inherited forms, it can be associated with other endocrine tumors, in syndromes called multiple endocrine neoplasia (MEN) 2A and MEN 2B (1). The primary curative treatment for MTC is total thyroidectomy, usually accompanied by at least central neck dissection. More extensive surgery and adjuvant therapies such as external-beam radiation therapy and tyrosine kinase inhibitors are generally reserved for more advanced disease.

Unfortunately, 70% of patients with MTC who present with a palpable thyroid nodule have cervical nodal metastases and 10% have distant metastases (DM) (1). Classically, older age and higher stage of disease at the time of diagnosis have been shown to be the strongest predictors for poor prognosis in MTC (2 –4). The risk designation based on the tumor (T), node (N), and metastases (M) staging system has recently been updated by the American Joint Committee on Cancer (AJCC) from the seventh to the eighth edition (5). Our institution previously reported on 65 MTC patients from 1946 to 1970, supporting higher stage as a strong predictor for worse survival (6). However, many of the reported studies did not evaluate the significance of tumor markers as predictors for outcomes.

The Revised American Thyroid Association (ATA) Guidelines for the Management of Medullary Thyroid Carcinoma recommend measuring serum calcitonin (Ctn) level 3 months after surgery to determine the need for further workup of persistent disease (1). A recent study (7) reported on Ctn levels postoperatively within 1 week of surgery, but long-term follow-up was not provided to determine whether early postoperative Ctn normalization correlates with long-term biochemical and structural cure. The prognostic value of tumor marker doubling time has also been reported (1), with doubling time of <6 months demonstrating the highest and >2 years showing the lowest rate of persistent or recurrent disease. However, in most cases, the disease has already progressed or recurred concomitant to the elevation of tumor markers. Hence, it is essential to analyze factors at the time of MTC diagnosis and initial surgery that could predict the disease course and outcomes. A few studies have evaluated these predictors (7 –11), while other studies have evaluated factors that influence postoperative biochemical cure and found that higher stage, higher preoperative Ctn level, larger primary tumor size, gross extrathyroidal extension (ETE), and cervical LN metastases were associated with a lower rate of biochemical cure (12 –16). However, most of these studies did not stratify sporadic and familial MTC patients, and some included patients with C cell hyperplasia who underwent prophylactic thyroidectomy. These observations underlie the importance of investigating a large MTC cohort managed and followed closely to further identify perioperative factors that affect long-term outcomes.

Materials and Methods

Results

Predictors of OS and DSS

The 5- and 10-year OS were 89.8% [confidence interval, CI, 84.8–95.0] and 81.2% [CI 74.0–89.2], respectively. On univariate analysis (Table 2), age >55 years (HR 3.72), male sex (HR 3.45), M1 (HR 9.22), gross ETE (HR 5.51), AJCC tumor stages 3 and 4 versus 1 and 2 (HR 5.39) (Fig. 1a), T category 3 and 4 versus 1 and 2 (HR 5.01), N1b (HR 3.74), tumor size (HR 1.36 for every centimeter increase in tumor size), 5 or more LN involvement (HR 2.50), ratio of involved/resected LN (HR 3.84), and Ctn (HR 8.63 for value >500 pg/mL) (Fig. 2) and CEA levels measured 3–6 months postoperatively were significantly associated with worse OS. On multivariable analysis (Table 2), age >55 years, gross ETE, M1 status, and every 100 unit increase in postoperative nadir Ctn level were associated with worse OS. Male sex and N1b status were not predictors of OS in multivariable analysis.

OPEN IN VIEWER

FIG. 1.

(a) Overall survival comparing MTC patients with AJCC TNM stages 3 and 4 vs. 1 and 2 (p = 0.0007). (b) Disease-specific survival comparing MTC patients with AJCC TNM stages 3 and 4 vs. 1 and 2 (p = 0.005). AJCC, American Joint Committee on Cancer; MTC, medullary thyroid cancer.

OPEN IN VIEWER

FIG. 2.

Overall survival comparing MTC patients with postoperative Ctn >1000 pg/mL vs. 10–5000 pg/mL vs. 0–10 pg/mL (p = 0.008). Ctn, calcitonin.

OPEN IN VIEWER

Table 2.

Predictors of Overall and Disease-Specific Survival in Medullary Thyroid Cancer

Predictors of worse survival	OS (n events/total = 28/163), univariate		OS (n events/total = 28/163), multivariable		DSS (n events/total = 18/163), univariate		DSS (n events/total = 18/163), multivariable
	HR [CI]	p	HR [CI]	p	HR [CI]	p	HR [CI]	p
Age >55 years	3.72 [1.64–8.44]	0.002	3.21 [1.25–8.14]	0.014	3.89 [1.39–10.91]	0.009	—
Male sex	3.45 [1.46–8.12]	0.005	2.42 [0.91–6.41]	0.07	3.99 [1.31–12.14]	0.01	—
White race	1.08 [0.15–8.06]	0.94	—		0.75 [0.09–5.73]	0.78	—
Sporadic (vs. hereditary)	2.51 [0.95–6.62]	0.06	—		4.11 [0.94–17.90]	0.06	—
Distant metastases	9.22 [3.96–21.48]	<0.0001	5.11 [1.93–11.06]	0.0006	13.22 [4.98–35.07]	<0.0001	10.45 [3.74–29.19]	<0.0001
Lateral neck involvement	3.74 [1.72–8.12]	0.0009	—		5.12 [1.82–14.41]	0.002	—
Gross extrathyroidal extension	5.51 [2.57–11.83]	<0.0001	4.62 [1.83–14.29]	0.002	8.15 [3.22–20.66]	<0.0001	6.58 [2.50–17.31]	0.0001
Stage 3 + 4 vs. 1 + 2 a	5.39 [2.04–14.23]	0.0007	—		18.42 [2.45–138.57]	0.005	—
AJCC TNM stage a		0.0013	—		—		—
1	—
2	0.64 [0.07–5.76]
3	1.65 [0.30–9.09]
4	5.89 [2.01–17.24]
T category 3 + 4 vs. 1 + 2	5.01 [2.37–10.57]	<0.0001	—		9.20 [3.44–24.60]	<0.0001	—
N1b vs. N1a	1.81 [0.61–5.37]	0.28	—		1.77 [0.50–6.22]	0.37	—
Largest tumor size	1.36 [1.11–1.66]	0.002	—		1.49 [1.21–1.83]	0.0002	—
No. of LN involved (5+ vs. <5)	2.50 [1.18–5.27]	0.02	—		2.91 [1.4–7.39]	0.02	—
Ratio of involved/resected LN	3.84 [1.37–10.83] b	0.01	—		4.14 [1.19–15.35] b	0.03	—
Ctn at diagnosis (pg/mL)	1.004 [1.002–1.007] c	0.0004	—		1.004 [1.001–1.007] c	0.01	—
Ctn at diagnosis (pg/mL)		0.19	—			0.04	—
0–500	—				—
500–1000	2.58 [0.64–10.32]				6.24 [0.65–59.97]
>1000	3.23 [0.94–11.11]				11.77 [1.54–89.97]
Postoperative nadir Ctn (pg/mL)		0.008 c	1.002 [1.001–1.004] c	0.005		0.005	—
0–10	—				—
>10–500	2.81 [0.99–7.80]				5.35 [1.15–24.78]
500–1000	—				—
>1000	8.63 [2.79–26.69]				13.89 [2.79–69.08]
CEA at diagnosis (ng/mL)	1.002 [1.000–1.004]	0.04	—		1.002 [1.000–1.004]	0.04	—
Postoperative CEA (ng/mL)	1.006 [1.003–1.009]	0.0002	—		1.004 [1.000–1.008]	0.07	—
Multifocal tumor	0.850 [0.36–2.01]	0.71	—		1.04 [0.37–2.91]	0.95	—
Bilateral tumor	1.69 [0.78–3.66]	0.19	—		2.45 [0.96–6.22]	0.06	—

The effect of male sex was not statistically significant when this was added to the multivariable model. When we added M1, N1b, and Ctn/CEA into the multivariate model, M1 but not N1b or Ctn/CEA was a significant predictor for worse overall survival.

a

Categorized by both AJCC seventh and eighth editions.

b

For every 1 unit increase in ratio.

c

For every 100 pg/mL increase in Ctn.

CI, confidence interval; DSS, disease-specific survival; HR, hazard ratio; OS, overall survival.

The 5- and 10-year DSS were 91.9% [CI 87.5–96.6] and 85.2% [CI 78.4–92.7], respectively. On univariate analysis (Table 2), age >55 years, male sex, M1 (Fig. 3a), N1b, gross ETE (Fig. 3b), tumor stages 3 and 4 (Fig. 1b), T category 3 and 4, tumor size, 5 or more LN involvement, ratio of involved/resected LN, Ctn at diagnosis, and Ctn and CEA levels measured 3–6 months postoperatively were associated with worse DSS. On multivariable analysis (Table 2), only gross ETE and M1 status were associated with worse DSS. To determine whether outcomes have changed over time, we compared 10-year survival rates in MTC patients who underwent thyroidectomy between 1995 and 2005 (OS 80.6% [CI 70.8–91.7] and DSS 84.0% [CI 74.7–94.3], respectively) with those who had surgery in the second decade of our study, from 2006 to 2015 (OS 82.2% [CI 71.8–94.2] and DSS 86.7% [CI 76.9–97.8], respectively), and found that there was no statistically significant difference in 10-year OS (p = 0.71) or DSS (p = 0.60).

OPEN IN VIEWER

FIG. 3.

(a) Disease-specific survival comparing MTC patients with distant metastases (M1) vs. those without distant metastases (M0) at initial surgery (p < 0.0001). (b) Disease-specific survival comparing MTC patients with gross ETE vs. those without gross ETE at initial surgery (p < 0.0001). ETE, extrathyroidal extension.

Discussion

In this cohort study of 163 MTC patients treated at our institution with a median follow-up of 5.5 years (range 0.002–20.6), we evaluated demographic, clinicopathologic, and laboratory factors in the perioperative period to identify predictors of long-term outcomes. Our cohort was mostly similar to previously reported multicenter cohorts (4,10,17) and large institutional series (18) in terms of patient demographics. However, 38% of our cohort had a hereditary form of MTC, which is higher than 18–24% reported in some studies (10,17). This is likely because many MTC families are followed at our institution, leading to a higher frequency of screening and diagnosis of hereditary MTC. MEN 2A was the most frequently diagnosed hereditary MTC syndrome. Compared with patients with sporadic MTC, those with hereditary forms of MTC had more bilateral and multifocal disease on presentation and had moderate risk RET mutations similar to another study (17). Patients with hereditary MTC also underwent surgery at a younger age (36–38 years) compared with those with sporadic MTC (56 years). Patients with hereditary MTC diagnosed by screening presented with stage 1 or 2 disease in 85% of cases compared with 36% of patients with sporadic MTC. In the entire cohort, 8% of patients presented with M1 and about one-third with N1b, findings that were more frequent in sporadic than in hereditary MTC. Also, more patients with sporadic MTC had postoperative nadir Ctn levels >150 pg/mL and higher CEA levels compared with patients with the hereditary forms of MTC. These differences could account for favorable long-term outcomes in most patients with hereditary disease in our cohort, similar to those in other studies (3,10,19). Overall, most patients had stage 1 (38%) or stage 4 (36.2%) disease, which is similar to the distribution observed in other large cohort studies (10,17,18).

The favorable survival rates in our study are in line with other reports of favorable long-term outcomes in most MTC patients reported recently (10,11,16 –18). Jung et al. reported improved 5-year recurrence rate but no difference in 5 year OS in 2006–2012 compared with 1982–2005 (11). In our study, comparison between 1995–2005 and 2006–2015 did not show any statistically significant differences in OS or DSS; however, even though statistical analysis was not possible, our cohort did demonstrate an improvement in survival rates compared with the 10-year OS (81.2% vs. 63%) and DSS (85.2% vs. 69%) observed in the historical Mayo cohort from 1946 to 1970 (6). Some possible reasons for this improved trend in survival rates may be the performance of more extensive surgery with routine central neck LN dissection, availability of tumor markers, and higher resolution imaging modalities in recent years. In another study, 10-year OS rates for patients with stages 1, 2, 3, and 4 MTC have been reported to be ∼100%, 93%, 71%, and 21%, respectively (3). In our study, demographic characteristics such as age >55 years and male sex predicted reduced survival, but the effect of sex on survival was not present when adjusted for other variables. There have been conflicting results in the literature regarding the effect of sex on MTC outcomes, with only two studies reporting male sex as a significant predictor of worse DSS at a population level (16,19). Younger age has been associated with better MTC survival, which may be conceived as a surrogate of early diagnosis especially in hereditary MTC; however, in our study, even after accounting for MTC type, age >55 years was a predictor of worse OS. Similar to other large series (3,11,16 –18), including the postoperative nomogram suggested by Ho et al. (8), we found classically described MTC characteristics, including higher Ctn and CEA at diagnosis, N1b, M1, gross ETE, higher AJCC stages (3 and 4), and larger primary tumor size to be associated with worse OS and DSS. Additionally, we found that higher Ctn levels evaluated 3–6 months post-thyroidectomy were associated with worse survival, a finding similar to some other studies (3,18). Older age and higher stage of disease have been reported in other studies to be significant predictors for worse survival (2 –4). Twito et al. recently reported that presence of DM and gross ETE are associated with worse DSS (17). On multivariable analysis in our study, the effect of N1b status on OS and DSS was not statistically significant after accounting for M1 status, likely because all patients with M1 also had N1b disease. We did find that age >55 years, male sex, gross ETE, M1, and high postoperative Ctn predicted worse OS, while gross ETE and M1 were the only factors that significantly predicted worse DSS. Hence, these factors would be most important when prognosticating MTC outcomes, of which M1 status portends the highest risk for worse survival. The effect of sex on survival lost significance after adjusting for postoperative Ctn, suggesting that a part of the influence of sex on MTC survival may be related to the higher burden of disease in men who might present at a more advanced stage.

The overall frequencies of LR and DM in our study were higher than previously reported at our institution (6). Factors accounting for this difference could be the increasingly sensitive diagnostic tools such as high-resolution ultrasonography and higher quality CT scanners, and the fact that tumor markers such as Ctn and CEA were not routinely available at the time of analysis of the previous study. In our study, there was a trend toward lower rate of LR and DM in hereditary MTC diagnosed by screening compared with hereditary MTC diagnosed by the presence of clinical or radiographic findings and sporadic MTC; however, due to limited number of events, we could not perform statistical comparisons. This trend is likely the result of earlier diagnosis with less disease burden at the time of initial surgery in patients with hereditary forms of MTC, especially in those diagnosed by screening, underpinning the importance of genetic counseling and proactive screening of patients. Age >55 years at the time of initial surgery and male sex were associated with higher risk of DM, while only male sex was associated with risk of LR. We found classically described MTC characteristics including N1b, M1, gross ETE, higher AJCC stages (3 and 4), and larger primary tumor size to be predictive of both LR and DM. In addition to these, ratio of involved/resected neck LN predicted DM and higher postoperative Ctn predicted both LR and DM. The results of biochemical factors are similar to those reported by others where high postoperative Ctn (absence of postoperative biochemical cure) predicted persistent disease and structural recurrence on long-term follow-up (9,11). Multivariable analyses by Twito et al. demonstrated that gross ETE and presence of DM at the time of diagnosis were predictors of disease persistence or recurrence (17). In our study, multivariable analysis was limited by the small number of events, thus limiting the power, but revealed that gross ETE and N1b predicted LR as well as DM, while high postoperative Ctn was additionally predictive of DM. This highlights the importance of a comprehensive preoperative neck ultrasound, routine central neck dissection, and additional lateral neck dissection when suspicious lymphadenopathy is identified in the lateral compartment, preferably after FNA confirmation. The predictive nature of postoperative Ctn level for DM underlies the importance of incorporating 3–6 months postoperative biochemical data to the initial tumor burden assessment when aiming to predict the long-term risk of disease spread.

Our study has several strengths and limitations. The main strength is that the management and close follow-up of patients occurred at a tertiary care center by a highly specialized team of thyroidologists, oncologists, and thyroid surgeons. The completeness of data in terms of MTC characteristics, genetic status, postoperative Ctn testing, and fairly long follow-up are also strengths of this study. The sample size was large enough to evaluate predictors for survival and disease recurrence/persistence; however, a limitation is that the number of outcome events was not large enough to separately evaluate these predictors within different MTC types or to perform more extensive multivariable analyses. In our study, multivariable analysis after adjusting for factors significant on univariate analysis was limited by the small number of events. Hence, the results of our multivariable analyses need to be validated in a larger cohort with more outcome events. Moreover, the collection of data from a referral center could lead to selection bias toward more aggressive cancer. Another limitation is the retrospective nature of the study based on chart review.

In summary, gross ETE of the tumor and presence of DM at the time of thyroidectomy were the strongest predictors of worse DSS. Demographic factors such as age and sex were not predictive of DSS, LR, and DM when adjusted for other variables. Lateral neck LN involvement, gross ETE, and high postoperative Ctn were the strongest predictors of both LR and DM. Hence, disease burden at the time of initial surgery as well as the biochemical response to surgery appear to be more important than demographic factors for MTC prognosis. These findings highlight the importance of rigorous perioperative assessment so as to better predict MTC outcomes.