Structural Doubling Time Predicts Overall Survival in Patients with Medullary Thyroid Cancer in Patients with Rapidly Progressive Metastatic Medullary Thyroid Cancer Treated with Molecular Targeted Therapies

Abstract

Purpose: To evaluate the impact of structural disease progression of metastatic lesions after initial surgery on overall survival (OS) of patients presenting with metastatic medullary thyroid cancer (MTC). We used tumor volume doubling time (TVDT) as a marker of structural disease progression and aimed to correlate the average structural tumor volume doubling time (midDT) with OS in MTC patients after initial surgery.

Methods: In this retrospective study, we examined the clinical characteristics; average tumor volume doubling times of neck, lung, and liver metastasis; and disease-specific survival of patients with metastatic MTC.

Results: Tumor growth is constant in MTC metastasis, irrespective of location of the metastasis. The median correlation coefficient (r) and the coefficient of determination (r²) were similar in lung metastasis (r = 0.91, r² = 0.95) and liver metastasis (r = 0.88, r² = 0.94), and comparable in neck metastasis (r = 0.73, r² = 0.85). Patients with metastatic MTC with a midDT ≤1 year have a worse prognosis than those with higher midDT (p = 0.002). Those with midDT ≤1 year had a median OS of 11.1 years [confidence interval (CI) 7.4–14.8 years]. In contrast, patients with midDT 1–3 years had a median OS of 16.5 years [CI 10.3–22.6 years]. All patients with midDT ≥3 survived by the end of the follow-up period. Preliminary results suggest that measurement of midDT can predict response to molecular targeted therapies.

Conclusions: In conclusion, TVDT is a strong predictor of OS in patients with recurrent or metastatic MTC, can be used as a marker of progression, and potentially can help select patients who may benefit from molecular targeted therapy.

Introduction

Medullary thyroid cancers (MTCs) are rare tumors of the thyroid gland with a prevalence of 1–2% in the United States (1). Only 25% of these cases are hereditary due to a germline mutation in the RET proto-oncogene (2). Their clinical course can vary from an extremely indolent tumor to a very aggressive cancer. Prognosis has been measured traditionally using the American Joint Committee on Cancer (AJCC) staging system that is based on the size of primary tumor, presence of lymph node metastases, distant metastases, and/or extrathyroidal extension. Using the AJCC 7th and 8th edition staging systems, the 5-year overall survival (OS) rates are 95% in stage I, 91% in stage II, 89% in stage III, and 68% in stage IV (3). Not surprisingly, patients with distant metastasis have a worse OS (10-year OS of 40%) than those with locally metastatic tumors (5-year OS 75%) (4).

Similar to nonmedullary differentiated thyroid cancer (DTC), treatment of MTC varies depending on the tumor size and location, tumor growth rate, estimated overall prognosis, potential benefit from proposed treatment options, potential adverse effects from the chosen therapies, and patient comorbidities and patient wishes. Treatment options include active surveillance for the very indolent tumors; localized therapy such as surgery, radiofrequency ablation, and radiation therapy for recurrent thyroid cancer that is amenable to these approaches; and systemic therapy with molecular targeted therapy (MTT) for rapidly progressive or symptomatic metastatic thyroid cancer. While a static staging system like the AJCC provides valuable initial disease-specific prognostic information, it does not adequately predict mortality in the setting of recurrent or progressive disease, cannot predict tumor growth rate over time, and cannot guide disease management decisions (1).

After initial surgery, dynamic risk stratification using calcitonin doubling time (Ct-DT) and carcinoembryonic antigen doubling times (CEA-DTs) has been shown to be important predictors of mortality in MTC. Both doubling times (DTs) can adequately predict tumor progression over time, development of metastatic disease, and have been used to forecast progression-free survival with chemotherapy in patients with progressive metastatic MTC (5–10). For this reason, the American Thyroid Association and National Comprehensive Cancer Network guidelines recommend both initial anatomic staging and postoperative calcitonin and CEA levels for ongoing dynamic risk stratification (1,11). However, treatment decision-making cannot be solely based on the change in tumor marker over time, but relies also on the presence, size, location, and change in volume of structural disease overtime.

In nonmedullary DTC patients, tumor volume doubling time (TVDT) has been used to predict structural disease progression and guide decision-making in patients with small intrathyroidal papillary thyroid cancer on active surveillance (12,13). In addition, it was shown to be an important clinical predictor of OS in DTC patients presenting with lung metastasis (14). Furthermore, in patients with lung metastasis from DTC, measurement of the average tumor volume doubling time (midDT) was shown to be an important risk stratification model that properly selects patients with progressive metastatic disease who are most likely to benefit from MTT and can also predict response and disease-specific survival after MTT. Using that model, patients with midDT ≤1 year were at the highest risk for mortality from metastatic DTC at baseline and most likely to respond and demonstrate disease-specific survival benefit from MTT, as opposed to patients with midDT ≥3 years who had better prognosis at baseline and were more likely to endure unnecessary adverse effects with systemic therapy without added survival benefit (15).

In this study, we aimed to evaluate the impact of structural disease progression of metastatic lesions after initial surgery on OS of patients presenting with metastatic MTC. Assuming that tumor growth in MTC follows a pattern of exponential growth similar to nonmedullary DTCs (16), and assuming the structural DT reflects the known exponential DT in calcitonin and CEA, we used TVDT as a marker of structural disease progression and aimed to correlate the midDT with OS in MTC patients after initial surgery. We hypothesized that patients with shortest midDT have worse prognosis that those with longer midDT.

Methods

Patients

After obtaining Institutional Review Board approval, we retrospectively reviewed the charts of patients with MTC with structural metastases. These patients were diagnosed with MTC between 1954 and 2017 and were followed at Memorial Sloan Kettering (MSK) during the course of their disease. All patients were treated with total thyroidectomy and had documented structural metastases in neck nodes, liver, and/or lungs, either immediately after initial surgery or during the follow-up period. To accurately measure tumor volume DTs, we included patients with ≥1 measurable metastatic lesion on at least 3 consecutive imaging studies performed at least 3 months apart, during a time period without systemic therapy or organ-targeted treatment. Liver lesions were evaluated on magnetic resonance imaging (MRI) of the liver, lung lesions on computed tomography (CT) scans of the chest, and lymph node lesions on ultrasonography of the neck. For patients who initiated systemic therapy (MTT or chemotherapy), DTs were calculated before the initiation of therapy. Similarly, for patients who underwent repeat surgery or organ-specific treatment (i.e., hepatic embolization), DTs were calculated before treatment. We excluded patients with the following: (a) insufficient imaging as per the mentioned criteria; (b) age ≤18 years at the time of diagnosis of metastatic MTC; (c) metastasis in the brain or bone without metastasis in neck nodes, lung, or liver; and (d) concomitant second primary malignancies.

Evaluation of lung lesions

Standard-dose noncontrast CT images of the chest across three or more different time points were reviewed for each patient. The 2 largest measurable lung nodules >5 mm were measured in 2 dimensions using 1 PACS workstation by 1 of 2 investigators (M.Y. or T.Y.). Longitudinal and transverse measurements were obtained by manually positioning electronic calipers on the CT image that demonstrated the largest cut of the selected nodule. Similar to prior studies looking at TVDT (14), each nodule was assumed to be ellipsoid in shape. Tumor volume (in mm³) for each nodule was calculated using the following formula: π/6 longest diameter × smallest diameter × smallest diameter.

Evaluation of liver lesions

MR images with contrast of the abdomen across three or more different time points were reviewed for each patient. The two most prominent and consistent liver lesions were measured in transverse cuts using a PACS workstation. Longitudinal and transverse measurements using electronic calipers were obtained of the largest cut of the selected lesion. Each lesion was assumed to be ellipsoid in shape, and tumor volume in milliliters was calculated using the following formula: π/6 longest diameter × smallest diameter × smallest diameter.

Evaluation of lymph nodes

Ultrasound images of the thyroid done at MSK across three or more different time points were reviewed for each patient. The two largest lymph nodes were identified and measured using a PACS workstation. Length, width, and depth measurements using electronic calipers were obtained of the largest cut of the selected lesion. Each lesion was assumed to be ellipsoid in shape, and tumor volume in milli liters was calculated using the following formula: π/6 length × width × depth.

Laboratory studies

Calcitonin (pg/mL) and CEA (μg/L) measurements that coincided within three months of the measured images were collected.

DT measurements

Using the same concept previously used by Miyauchi et al. to define thyroglobulin DT (5) and assuming the changes in tumor volume are exponential (Fig. 3), we computed a regression line, log y = log a + bx, using a nonlinear square regression with “x” defined as time (in years) after the initial imaging scan and “y” defined as tumor volume. TVDT was calculated as (log 2)/b. We then calculated the average tumor volume doubling time for each given patient (referred to herein as midDT) using the average of the calculated DTs of the two selected nodules.

FIG. 3.

Tumor growth rate over time in MTC.

Ct-DT and CEA-DT were calculated for each patient using at least 3 consecutive calcitonin and CEA levels obtained within a 3-month window of the nodules' measurement. To address the issues of discontinuation in DT among patients with increasing or decreasing serum DT overtime, we calculated 1/Ct-DT and 1/CEA-DT values for all patients.

Statistical analysis

Continuous data are presented as means and standard deviations or median and ranges, as appropriate for each variable.

To use TVDT as a marker of tumor progression in MTC, we had to document that the tumor growth is constant over time, after an exponential growth irrespective of the tumor growth rate and location of the metastasis. For this purpose, we analyzed the measured tumor volumes for each given nodule over time. We then calculated the correlation coefficient (r) and the coefficient of determination (r²) for each given nodule and calculated the average correlation coefficient r and the coefficient of determination (r²) for all nodules in a given organ (neck nodes, lung, and liver).

The main objective of the study was to determine whether the midDT could predict OS from metastatic DTC. Based on previously defined thyroglobulin and midDT prognostic breakpoints in DTC and calcitonin and CEA prognostic breakpoints in MTC, we grouped the patients into 3 clinically relevant groups: those with midDT ≤1 years, those with midDT between 1 and 3 years, and those with midDT ≥3 years (5–7,14,15,17). Using Kaplan–Meier survival analysis and log rank testing, we calculated OS from MTC diagnosis based on midDT groups (Fig. 1A).

FIG. 1.

Overall survival curves based on midDT (A), Ct-DT (B), and CEA-DT (C). CEA-DT, carcinoembryonic antigen doubling time; Ct-DT, calcitonin doubling time; midDT, average structural tumor volume doubling time; MTC, medullary thyroid cancer.

We then grouped patients according to their serum Ct-DT and CEA-DT using similar prognostic breakpoints: those with Ct-DT and/or CEA-DT ≤1 year, those with Ct-DT and/or CEA-DT between 1 and 3 years, and those with Ct-DT and CEA-DT ≥3 years. We repeated the OS analysis from MTC diagnosis based on Ct-DT and CEA-DT groups, respectively (Fig. 1B, C). Given that Ct-DT often varies over time in patients with recurrent or rapidly progressive disease and with interim therapies, we calculated 1/Ct-DT for all patients.

Thirteen patients with distant metastatic structurally progressive MTCs were treated with MTT during the course of their follow-up at MSK. To assess MTT effect on TVDT in metastatic MTC, we selected two patients with liver metastasis who had sufficient imaging before and after MTT therapy. Both had midDT ≤1 year and were treated with MTT (one with sorafenib and vandetanib sequentially, the other with vandetanib alone). We measured tumor volume and midDT of a dominant liver nodule before and after MTT on at least three consecutive CTs (Fig. 2).

FIG. 2.

midDT of two selected patients before and after MTT therapy. MTT, molecular targeted therapy; TVDT, tumor volume doubling time.

All analyses were performed using SPSS software (version 24; SPSS, Inc., Chicago, IL). A p-value of ≤0.05 was considered statistically significant.

Results

Patient characteristics

The characteristics of all 43 patients who met our selection criteria are summarized in Table 1. The patients had a median age of 48 years at the time of MTC diagnosis. The median size of the primary tumor was 2.2 cm (range 0.7–13 cm), with 47% having extrathyroidal extension, and 42% with vascular invasion. Out of the 38 patients wherein germline genetic analysis was done, 11 (29%) patients had an RET germline mutation. A third of patients (33%) had metastatic disease only in the lymph nodes; the remainder of patients had either distant metastases in addition to local lymph node disease (44%), or distant metastases alone (23%). Thirteen had lung metastases, 13 had liver metastases, and 17 had lymph node metastases either at the time of diagnosis or during the follow-up period. Thirteen of the 43 patients received MTT in addition to surgical and nonsurgical interventions.

Table 1.

Clinical Characteristics of the Patient Cohort

Characteristics	% (n)
Age at diagnosis, years	(43)
Mean ± SD	48 ± 15
Median	48
Range	13–77
Sex
Female	51 (22)
Size of primary tumor, cm
Mean ± SD	3.2 ± 2.6
Median	2.2
Range	0.7–13
Presence of germline RET mutation
Yes	26 (11)
No	63 (27)
Unknown	11 (5)
Extrathyroidal extension
Yes	47 (20)
No	42 (18)
Unknown	11 (5)
Vascular invasion
Yes	42 (18)
No	35 (15)
Unknown	23 (10)
8th edition AJCC staging at diagnosis
I	11 (5)
II	16 (7)
III	28 (12)
IVa	33 (14)
IVc	12 (5)
Metastatic disease location
Neck nodes	33 (14)
Lung only	9 (4)
Liver only	12 (5)
Lung + liver	2 (1)
Neck nodes + liver	16 (7)
Neck nodes + lung	16 (7)
Neck nodes + lung + liver	12 (5)
Nadir calcitonin (after initial treatment)
Median	471
Nadir CEA (after initial treatment)
Median	14
MTT
Yes	30 (13)
Follow-up since diagnosis, years
Mean ± SD	11.8 ± 5.4
Median	11.0
Range	2.2–24.0
Status at end of follow-up
Alive	74 (32)
Dead	26 (11)
Cause of death
MTC	82 (9)
Other	9 (1)
Unknown	9 (1)

AJCC, American Joint Committee on Cancer; CEA, carcinoembryonic antigen; MTT, molecular targeted therapy; SD, standard deviation.

The patients were followed for a median of 11 years (range 2.2–24.0 years). At the end of follow-up, all patients had persistent disease: 56% had structural incomplete disease, 14% had biochemical incomplete disease, 4% had recurrence of disease, and 26% were deceased. Of the 11 patients who died, the primary cause of death was MTC (82%). One patient died of sepsis and the cause of death was not known for one patient.

Overall survival

For the cohort of 43 patients, the median OS from the time of MTC diagnosis was 20.7 years [95% confidence interval (CI) 13.7–27.6 years]. OS differed in patients with metastases limited to the neck from the OS of patients with distant metastases to the liver or lung. Those with metastases limited to the neck had 100% survival, with a median follow-up time of 8.5 years (range 4.5–42 years). In the group with distant metastases, 11 of 26 patients died (42%), with a median follow-up of 11 years (range 2.3–22.4 years). The OS for patients with distant metastases was 100% at 5 years, 77% at 10 years, 61% at 15 years, and 45% at 20 years.

Distribution of tumor growth in MTC

In the 80 MTC metastatic lesions measured in the 43 patients, the tumor growth rate was remarkably constant over time with a median follow-up period of 11 years (median r = 0.83, r² = 0.91; Fig. 3). When we evaluated the lesions based on location, the median correlation coefficient (r) and the coefficient of determination (r²) were similar in lung metastasis (r = 0.91, r² = 0.95) and liver metastasis (r = 0.88, r² = 0.94), and comparable in neck metastasis (r = 0.73, r² = 0.85), demonstrating that in MTC, tumor growth is constant over time irrespective of metastatic location. Similar to DTC (study under review), the correlation coefficient in nodal metastasis is smaller than in distant metastasis. This may be due to variability in the measurement of nodal metastasis with time due to operator-dependent technique (sonogram), or intervening infectious and inflammatory changes that allow for size fluctuation with time.

Tumor volume doubling time

All but 8 patients had at least 2 measurable metastatic lesions. The average structural tumor volume doubling time (midDT) was 1.6 years. When sorted into 3 categories, 10 of 43 patients had midDT <1 year, 15 had midDT between 1 to 3 years, and 18 had midDT >3 years. The midDT >3 years group included patients with very slow tumor growth, those with stable disease, and those with tumor shrinkage.

Patients with metastatic MTC with a midDT ≤1 year were found to have a worse prognosis than those with higher midDT (p = 0.002) (Fig. 1; Table 2). Those with midDT ≤1 year had a median OS of 11.1 years (CI 7.4–14.8 years). In contrast, patients with midDT 1–3 years had a median OS of 16.5 years (CI 10.3–22.6 years). All patients with midDT ≥3 survived by the end of the follow-up period. The 5-year OS was 100% for all midDT groups. At 10 years, the OS rate was 60% for midDT <1 year, 82% for midDT of 1–3 years, and 100% for midDT >3 years. At 15 years, the OS rates were 45%, 61%, and 100%, respectively.

Table 2.

Overall Survival Based on midDT Groups

midDT group	Median OS, years [CI]	5 years, %	10 years, %	15 years, %
<1 year	11.1 [7.4–14.8]	100	60	45
1–3 years	16.5 [10.3–22.6]	100	82	61
>3 years	—	100	100	100

CI, confidence interval; midDT, average structural tumor volume doubling time; OS, overall survival.

Biochemical doubling time

A total of 38 patients had serial calcitonin levels measured and a total of 40 patients had serial CEA measured. The median nadir calcitonin within 6 months of initial surgery was 471 ng/mL. The median Ct-DT was 2.4 years. The median nadir CEA within 6 months of initial surgery was 13.5 pg/mL, and the median CEA-DT was 2.3 years.

Figure 1B and C depicts the OS of MTC patients based on Ct-DT and CEA-DT groups. Similar to midDT, patients with Ct-DT and/or CEA-DTs ≤1 year were at highest risk for mortality from MTC than those with DTs of >1 year. Patients with structural or biochemical DTs >3 years had the best prognosis. At 5 years, the survival rate was 100% for all groups. At 10 years, the survival rate was 67% for Ct-DT and 66.6% for CEA-DT <1 year, 79% for Ct-DT and 86.3% for CEA-DT of 1–3 years, and 100% for Ct-DT and CEA-DT >3 years. At 15 years, the survival rates were 66% and 35%, 68% and 59%, and 100% and 91%, respectively.

TVDT pre/postmolecular targeted therapy

Thirteen of 43 patients received MTT, 8 of whom had lung metastases, 4 who had liver metastases, and 1 who had neck lymph node metastases. From these patients, we selected two patients who had sufficient imaging studies to accurately measure the structural TVDT before and after initiation of MTT (Fig. 2). Both patients had midDT ≤1 year before initiation of MTT. After MTT, the tumor volume decreased, and TVDT was significantly prolonged after therapy initiation, consistent with a positive clinical response. In the first patient who had metastatic MTC with liver metastases, pre-MTT midDT was 0.43 years. He received sorafenib and vandetanib sequentially over the course of 7 months. His tumor showed significant shrinkage, and post-MTT, midDT was prolonged to >3 years. The second patient had metastatic MTC with liver metastases as well. Pre-MTT, midDT was 0.34 years. After 9 months of vandetanib therapy, the tumor shrunk and his post-MTT midDT was again >3 years.

Discussion

This study convincingly demonstrates that tumor growth in metastatic MTC is constant over time, irrespective of the location of the tumor deposit. Thus, similar to patients with metastatic DTC, TVDT is an appropriate marker for tumor growth in MTC patients. Furthermore, TVDT measuring structural tumor volume change over time is a strong predictor of prognosis in metastatic MTC.

Our patient cohort reflects the subset of patients with MTC who present with persistent or recurrent metastatic disease. The 5-year OS of all groups was better in this cohort than published data. This is likely due to selection bias, since we excluded from our study patients with rapidly progressive MTC after initial diagnosis that went on directly to be treated with MTT or other systemic therapy, or those who died before we had sufficient imaging tests to measure TVDT. In our cohort, not surprisingly and consistent with prior reports, patients with metastases limited to the neck had 100% OS, while patients with distant metastases had worse outcomes.

Both Ct-DT and CEA-DT were also found to be important predictors of prognosis in MTC as previously demonstrated, thus validating the TVDT results. In fact, the association with survival was statistically significant with midDT and CEA-DT, while survival trended toward statistical significance with Ct-DT. This may be due to known variability of Ct-DT in tumors that present with rapid increase or decrease in tumor volumes. To resolve this issue of DT discontinuation, Barbet et al. used 1/Ct-DT instead of Ct-DT (7). Furthermore, in 13 patients with metastatic MTC, Ito et al. showed that patients with 1/Ct <0.11 survived MTC, while all patients with 1/Ct >0.67 died from their disease (6): that study did not comprise patients with Ct-DT between 0.11 and 0.67. In the current study, all patients with 1/Ct >0.47 did not survive MTC by the end of the follow-up, while the surviving patients had 1/Ct ranging between −0.09 and 0.45.

The current study was not aimed to study the relationship between biochemical DT and midDT in MTC. However, it is our observation that midDT was well related to both Ct-DT and CEA-DT in some but not in all patients. When discordant, TVDT tended to follow CEA DT rather than Ct-DT.

Unfortunately, we had only 2 patients who had metastasis in the 2 out of the 3 selected organs and had consecutive TVDT measurement of these metastasis. One patient had neck and lung metastasis, while the second had neck and liver metastasis. In the first patient, midDT was ≥3 years in both the neck and lung metastases, while in patient 2, the neck midDT was ≤1 year, while liver midDT was 1–3 years. Thus, due to lack of sufficient data, we were not able to consistently compare the TVDT of the 2 metastatic foci in separate organs in these 2 patients.

MTT is now offered for patients with rapidly progressive metastatic thyroid cancer, both those with tumors of follicular origin and MTC. We have previously shown that in patients with lung metastasis from DTC, TVDT measurement is not only an important prognostic indicator, but it can also guide selection of the patients who will benefit from MTT, and it can help assessing response to and disease-specific survival after MTT (14,15). In this initial study on patients with recurrent or metastatic MTC, we similarly demonstrate that midDT is a good prognostic indicator that can help select patients at highest risk for mortality. Those patients are more likely to benefit from MTT than lower risk patients. MidDT also helps identifying patients with slowly progressive disease who are unlikely to succumb to their disease and who would be at more risk from MTT-induced adverse effects without added benefit. Furthermore, our preliminary results seem to suggest that MTT can prolong midDT in patients with rapidly progressive MTC. Further studies with larger cohort of patients are needed to confirm these results.

Our study is limited by its design (retrospective analysis) and small sample size. In addition, we did not have enough patients to correlate midDT between different organs, nor did we have enough patients with brain metastasis to assess midDT in brain metastasis. We did not measure structural DT in bone metastasis as it is not possible to accurately evaluate their volume over time.

In conclusion, TVDT is a strong predictor of OS in patients with recurrent or metastatic MTC, can be used as a marker of progression, and potentially help selecting patients who may benefit from MTT.

Author Disclosure Statement

T.Y. and M.Y. have no conflicts of interest to report. M.M.S. and R.M.T. are consultants for EISAI Pharmaceutical. R.M.T. is also a consultant for Bayer, Inc.

Funding Information

This research was funded, in part, through the NIH/NCI Cancer Center Support Grant P30 CA008748.