Prognostic Impact of Serum Calcitonin and Carcinoembryonic Antigen ...

0021-972X/05/$15.00/0 Printed in U.S.A.

The Journal of Clinical Endocrinology & Metabolism 90(11):6077– 6084 Copyright © 2005 by The Endocrine Society doi: 10.1210/jc.2005-0044

Prognostic Impact of Serum Calcitonin and Carcinoembryonic Antigen Doubling-Times in Patients with Medullary Thyroid Carcinoma Jacques Barbet, Loı¨c Campion, Franc¸oise Kraeber-Bode´re´, Jean-Franc¸ois Chatal, and the GTE Study Group Institut National de la Sante´ et de la Recherche Me´dicale, Unite´ 601 (J.B., F.K.-B., J.-F.C.), F-44000 Nantes, France; Universite´ de Nantes (J.B., F.K.-B., J.-F.C.), F-44000 Nantes, France; and Rene´ Gauducheau Cancer Center (L.C.), F-44805 Nantes-St. Herblain, France Context: After unsuccessful surgery, medullary thyroid carcinoma (MTC) may be fatal or remain stable for decades, and precise survival predictors are needed. Objective: This study assesses the prognostic value of calcitonin and carcinoembryonic antigen (CEA) doubling-times (DT). Design: This is a retrospective study on 65 MTC patients from 2.9 – 29.5 yr after surgery. Setting: Data registered in the database of the French Neuroendocrine Tumor Group were analyzed anonymously. Patients: All patients had abnormal calcitonin levels after total thyroidectomy and bilateral lymph node dissection. Intervention: Calcitonin and CEA serum levels were measured during routine disease follow-up. Main Outcome Measure: To assess DT as prognostic factors, a

T

HE PROGNOSIS OF medullary thyroid carcinoma (MTC) is intermediate between differentiated and anaplastic thyroid carcinomas (1–3). In the absence of distant metastasis, the generally accepted first line treatment of MTC is total thyroidectomy and bilateral lymph node dissection. When surgery has not fully normalized calcitonin (4), the natural history of MTC varies from rapid progression and survival for a few years, to very slow progression or stable disease extending over decades (2, 5, 6). Until recently, therapeutic options for locally recurrent tumors were limited to repeated surgery and/or external beam radiotherapy, the impact of which on survival has been controversial (7). Chemotherapy has a transient and limited efficacy in advanced stages of the disease (1, 8). Radioimmunotherapy, using anticarcinoembryonic antigen (anti-CEA) antibodies, and radiolabeled octreotide have recently been studied as new therFirst Published Online August 9, 2005 Abbreviations: CEA, Carcinoembryonic antigen; 95% CI, 95% confidence interval; DT, doubling-time(s); EORTC, European Organization for Research and Treatment of Cancer; HR, hazard ratio; MEN, multiple endocrine neoplasia; MTC, medullary thyroid carcinoma; PVE, proportion of variance explained; TNM, tumor-node-metastasis. JCEM is published monthly by The Endocrine Society (http://www. endo-society.org), the foremost professional society serving the endocrine community.

patient population was extracted from the database. Results: When calcitonin DT was less than 6 months, 5- and 10-yr survivals were three of 12 (25%) and one of 12 (8%), respectively; when between 6 months and 2 yr, 5- and 10-yr survivals were 11 of 12 (92%) and three of eight (37%), whereas all 41 patients with calcitonin DT greater than 2 yr were alive at the end of the study. Tumor-NodeMetastasis (TNM) stage, European Organization for Research and Treatment of Cancer (EORTC) score, and calcitonin DT were significant predictors of survival by univariate analysis, but only calcitonin DT remained an independent predictor of survival by multivariate analysis (P ⫽ 0.002) with a proportion of variance explained (PVE) of 37.4%. Calcitonin DT was a better predictor than CEA (PVE 63.3% and 47.0%, respectively). Calcitonin DT calculated using only the first four measurements was also an independent predictor of survival (P ⬍ 0.000001; PVE 40.4%). Conclusion: Calcitonin DT may be superior to initial clinical staging and among the most powerful prognostic indicators in MTC. (J Clin Endocrinol Metab 90: 6077– 6084, 2005)

apeutic modalities, with encouraging results in early clinical trials (9 –11). Tyrosine kinase inhibitors could also be effective in these tumors that express mutated forms of the RET oncogene (12) Synergistic effects have been demonstrated between radioimmunotherapy and chemotherapy using taxanes (13) or doxorubicin (14), and between unlabeled CEA antibodies and dacarbazine (15). These new therapeutic options confirm the need for an early distinction between highrisk patients who need to be treated and low-risk patients who warrant a “watch and wait” behavior. Among the various prognostic parameters that could identify high-, moderate-, and low-risk groups with the aim of defining optimal therapeutic strategies (1–3, 5, 6, 16, 17) advanced age, advanced stage of the disease, and associated multiple endocrine neoplasia (MEN) 2B appear to be the best commonly accepted factors of poor prognosis. Individual parameters have been associated within different staging systems, and Kebebew et al. (2) concluded that the European Organization for Research and Treatment of Cancer (EORTC) prognostic scoring system (18), which takes into account age, gender, nature, and stage of the disease for all thyroid cancers, had the highest predictive value. Serum kinetics of MTC markers, such as calcitonin and CEA, could be alternative predictors of survival (19), and Miyauchi et al. (20) have proposed calcitonin doubling-

6077

6078

J Clin Endocrinol Metab, November 2005, 90(11):6077– 6084

times (DT) as a prognostic factor. The purpose of the present study was to evaluate the predictive value of calcitonin and CEA serum kinetics, and especially DT of both markers in a cohort of 65 patients who have been followed for 6 months to 29.5 yr after surgery. Patients and Methods

Barbet et al. • Calcitonin and CEA Doubling-Times in MTC

The inverse values of DT counted positive for progression (1/DT) and of t1/2 counted negative for regression (⫺1/t1/2) were used in all statistical analyses to avoid the discontinuity problem. For stable cases, 1/DT was set to zero. However, for clarity and easier clinical interpretation, changes in serum marker levels are reported in the Results and Discussion sections as DT (progression) or t1/2 (regression), expressed in years. Survival analysis was performed using the SAS 8.2 software package (SAS, Inc., Cary, NC). For all tests, a value of P ⱕ 0.05 was considered as significant. All P values given are the results of two-sided tests.

Patient selection MTC patients treated in participating centers in France can be registered, after formal agreement, in the database of the French Neuroendocrine Tumor Group (GTE). The history of their disease and routine calcitonin and CEA blood level measurements are then recorded in the database, which is managed according to the regulations of the French “Commission Nationale Informatique et Liberte´.” Patients who 1) were not operated on by total thyroidectomy and lymph node resection, or 2) were cured by surgery (negative postsurgery calcitonin testing), or 3) had not been measured for serum calcitonin at least four times after surgery, or 4) whose status could not be confirmed at the time of the study were excluded. All other patients entered in the database before February 2003 were included, for a total of 65 patients. Data extracted from the database by the GTE were made anonymous before any analysis.

Calcitonin and CEA serum determination Calcitonin and CEA measurements were obtained using a single technique for a given patient but, for calcitonin, three assays were used (CT US IRMA, Brahms Diagnostica, Berlin, Germany; IRMA hCT, CIS Bio International, Gif-sur-Yvette, France; Advantage, Nichols Institute Diagnostics, San Juan Capistrano, CA) and, for CEA, a total of 15 different assays from nine different manufacturers were used (Axsym and Architect, Abbott Laboratories, Abbott Park, IL; ACS and Advia Centaur, Bayer Diagnostics, Tarrytown, NY; Vidas, bioMe´rieux, Marcy l’E´toile, France; Kryptor, Brahms Diagnostica; Elsa 2 and RIAgnost, CIS Bio International; Immunlite and Immunlite 2000, Diagnostic Products Corporation, Los Angeles, CA; IRMA, Immunotech, Marseille, France; Vitros ECI, Ortho-Clinical Diagnostics, Inc., Raritan, NJ; Cobas core, Elecsys 2010 and Modular, Roche Molecular Diagnostics, Pleasanton, CA). This could affect comparisons between patients, but had no influence on the estimation of DT.

Data analysis Single exponentials were fitted to serum calcitonin and CEA concentrations by nonlinear least square regression. Data were weighted by the inverse of the measured concentrations, and best-fit values were reported as DT for progressive disease or half-life (t1/2) for tumor regression. sd were calculated as asymptotic se. DT or t1/2 were evaluated after all major therapeutic treatments, such as surgery or external beam radiation therapy, and were not calculated when serum concentrations remained below the normal cut-off value. DT were considered as “stable” when too long to be estimated (one case for calcitonin, two cases for CEA). Overall survival was calculated from the day of surgery until death of the patient. Cause-specific survival curves for each scoring system were calculated using the method of Kaplan-Meier, and compared using the log-rank test for trend. Deaths from a cause other than MTC, or survival after the end of the observation period, were considered as censoring events. Age, gender, Tumor-Node-Metastasis (TNM) staging, EORTC score (18), and inverse values of calcitonin and CEA DT were considered as independent variables. Patient age at time of diagnosis and gender and other variables with significant influence in the univariate analysis were further considered in multivariate analysis through forward Cox’s proportional hazard model. The proportion of variance explained (PVE) of the different staging systems was calculated using the Cox proportional hazards model according to the method of Schemper and Stare (21, 22). This measure of the amount of variability in survival times that can be assigned to known prognostic factors and staging has a similar interpretation as R-squared in linear regression.

Results Clinical and biological data

A total of 65 MTC patients with abnormal calcitonin after surgery for whom at least four data points were available to calculate DT were selected in the database by the GTE. Descriptive statistics by age, gender, TNM and EORTC scoring, and calcitonin and CEA DT of the 65-patient population and of the subgroup of patients for whom CEA DT data were available are listed in Table 1. Six patients had familial MTC, four patients had MEN 2A, and two had MEN 2B. There was no TNM stage I, which is consistent with the fact that all patients considered in this study had abnormal calcitonin levels postsurgery. Of these 65 patients, 53 had calcitonin DT ranging from 0.04 –25.1 yr (median, 2.60 yr), one patient was stable, and calcitonin serum levels decreased for 11 patients with t1/2 ranging from 1.69 –18.4 yr (median, 6.16 yr). The DT were calculated (two examples are shown in Fig. 1) and their distributions are reported in Table 1. For only five of the 41 patients (12%) with DT or t1/2 less than 5 yr, the estimated sd (asymptotic se) was larger than one half the DT. Precision on the estimation of long DT or t1/2 (⬎5 yr) was lower; for 10 of the 26 patients (42%), sd was more than 0.5 DT or t1/2. These patients were scored as DT more than 2 yr. Of the entire population of 65 patients, CEA measurements were missing for seven patients or fell within the normal range for 12 other patients. Normal CEA did not correlate with slow progression; two patients with normal CEA levels showed fast (DT ⬍ 0.5 yr) and two intermediate TABLE 1. Clinical characteristics of the patient population Whole population

Subgroup with CEA DT available

No. of patients 65 46 Age [median (range)] 51 (6 to 75) 48 (6 to 75) Age ⱕ45 yr 24 19 ⬎45 yr 41 27 Gender (male/female) 31/34 23/23 Stage (TNM) [n (%)] I 0 0 II 10 (15.4) 8 (17.4) III 49 (75.4) 33 (71.7) IV 6 (9.2) 5 (10.9) EORTC group [n (%)] 1 10 (15.6) 9 (19.6) 2 18 (28.1) 13 (28.3) 3 24 (35.8) 16 (34.8) 4 12 (18.8) 7 (15.2) 5 1 (1.6) 1 (2.2) EORTC score (range) 68 (28 to 110) 68 (28 to 110) 1/DT calcitonin [median 0.29 (– 0.59 to 25) 0.26 (– 0.29 to 25) (range)] 1/DT CEA [median (range)] 0.29 (– 0.62 to 14.3)


J Clin Endocrinol Metab, November 2005, 90(11):6077– 6084 6079

FIG. 1. Tumor marker kinetics. Calcitonin and CEA plasma levels (f) were monitored periodically after diagnosis and initial surgery in MTC patients. Data were then fitted to single exponentials by nonlinear least-square analysis (solid lines). Two illustrative examples are shown in the figure. Patient A (calcitonin, A; CEA, B) showed fast progression of both markers; estimated calcitonin and CEA DT (⫾SD) were 0.29 ⫾ 0.03 and 0.34 ⫾ 0.05 yr, respectively. Patient B (calcitonin, C; CEA, D) showed slower progression (note the change in time scale) of both markers; estimated calcitonin and CEA DT (⫾SD) were 2.5 ⫾ 0.2 and 3.5 ⫾ 0.3 yr, respectively.

(0.5 ⬍ DT ⬍ 2 yr) calcitonin DT. In the subgroup of 46 patients with quantifiable CEA DT (CEA-subgroup), the median age was 48 yr (range, 6 –75 yr) and the male-to-female ratio was 1.0:1.0 (23/23). The mean follow-up time was 10.5 yr (median, 8.8 yr). According to the TNM staging system, there was no stage I, 17.4% stage II, 71.7% stage III, and 10.9% stage IV disease. According to the EORTC scoring system, this subgroup consisted of 48% groups 1 and 2, 35% group 3, and 17% groups 4 and 5. The median EORTC score was the same, 68, for this subgroup and the whole population. Thus, overall, clinical parameters in this subgroup of patients were not different from those of the whole population (Table 1). In this subgroup, CEA serum levels increased for 40 patients with DT ranging from 0.07–24.2 yr (median 2.4 yr), were stable for two patients, and decreased for four patients with t1/2 ranging from 1.60 –36.2 yr (median, 15.1 yr). Thus, CEA and calcitonin observations were very similar in most patients (Fig. 1). As shown in Fig. 2, the progression rates, measured as the inverses of calcitonin and CEA DT, correlated well (R2 ⫽ 0.9611).

Survival statistics

The mean follow-up time was 9.6 yr (median, 8.3 yr). The overall cause specific mortality was 17% at 5 yr and 29% at 10 yr. All patients with calcitonin DT more than 2 yr (or with stable or regressing disease) survived until the end of the follow-up, whereas all those with DT less than 0.5 yr died within 0.5–13.3 yr, and only four of the 12 patients with intermediate DT (between 0.5 and 2 yr) survived until the end of the study. The patient population was thus stratified into three groups: group 1 (low risk), 41 patients with calcitonin DT more than 2 yr (or with stable or regressing disease); group 2 (intermediate risk), 12 patients with calcitonin DT between 0.5 and 2 yr; and group 3 (high risk), 12 patients with calcitonin DT less than 0.5 yr. No correlation was found between postsurgery calcitonin or CEA levels and survival. Indeed. if median values were slightly higher in the high-risk group, the differences were very small compared with the range of observed values in all three risk groups: calcitonin, high risk, 438 (18 – 6629); inter-

6080

J Clin Endocrinol Metab, November 2005, 90(11):6077– 6084


FIG. 2. Correlation between calcitonin and CEA progression rates. To solve the discontinuity problem between progressive disease and tumor regression and to make possible the correlation analysis, results are shown as the inverse values of DT (progression rates). For all available individual pairs of data (except one patient with very fast progression, calcitonin DT: 0.04 ⫾ 0.01 yr and CEA DT: 0.05 ⫾ 0.01 yr, purposely set off-scale), CEA progression rate (1/DT) is plotted against calcitonin progression rate (1/DT) in the figure. Linear regression gives a slope of 0.67 and a correlation coefficient of 0.9611.

mediate risk, 269 (12–3487), low-risk group, 371 (30 –9000) pg/ml; CEA, high risk, 35 (2–368); intermediate risk, 11 (2– 124), low-risk group, 11 (1–184) ng/ml. Along the same line, absolute levels of calcitonin measured shortly before death varied from 371– 488,000 pg/ml, showing that these absolute levels have no predictive value. Five- and 10-yr survival rates are given in Table 2 for the different TNM and EORTC prognosis groups, as well as for the groups defined on the basis of calcitonin DT. Survival TABLE 2. Survival statistics at 5 and 10 yr

Staging variable

Whole population (n ⫽ 65)

Subgroup with CEA DT available (n ⫽ 46)

5-yr 10-yr 5-yr 10-yr survival survival survival survival

TNM I–II III IV–V EORTC 1–2 3 4 –5 Calcitonin DT (all points) ⬎2 0.5–2 ⬍0.5 Calcitonin DT (first four points) ⬎2 0.5–2 ⬍0.5 CEA DT ⬎2 0.5–2 ⬍0.5

100 81 63

89 70 42

100 79 75

88 66 50

93 83 60

83 70 40

91 81 58

80 64 44

100 92 25

100 37 8

100 90 14

100 17 0

100 94 23

96 64 15

100 92 13

95 56 13

100 75 0

96 23 0

statistics were not different in the subgroup of 46 patients with available CEA data and the whole population. Because cut-off values for the groups based on calcitonin DT were selected on the basis of overall survival, it is not surprising that these groups also give a very clear delineation in terms of 5- and 10-yr survival. Patients still alive at 5 and 10 yr were 37 of 37 and 22 of 22, respectively, in group 1; 11 of 12 and three of eight, in group 2; and only three of 12 and one of 12, respectively, in group 3. In the CEA-subgroup, the overall cause-specific mortality was 18% at 5 yr and 31% at 10 yr, and the survival statistics broken down by TNM, EORTC, or calcitonin DT were quite similar to those of the whole population. Groups defined on the basis of CEA DT, using the same cut-off values of 6 months and 2 yr, also gave a clear delineation of 5- and 10-yr survival: patients still alive at 5 and 10 yr were 27 of 27 and 17 of 18 if CEA DT was more than 2 yr, nine of 12 and two of 10 if CEA DT was between 6 months and 2 yr, and zero of four and zero of four if CEA DT was less than 6 months (Table 2). Prognostic factors and staging system

The results of the univariate and multivariate analyses of survival data vs. clinical and biological variables are summarized in Table 3. EORTC score and calcitonin DT (analyzed in terms of their inverse values, 1/DT, as explained in Patients and Methods) were significant predictors of survival by univariate analysis. Although TNM has some prognostic value as shown in Table 3 and in Fig. 3, the correlation with calcitonin DT was not easy to analyze; there were more TNM stage 4 in the high-risk group (25%) than in the other groups (0 and 10%), and the low-risk group did not look different from the intermediate-risk group. EORTC scoring seems better correlated to calcitonin DT; there were 60% group 4 and 5 in the high-risk group, compared with 8% in the intermediate-risk group and 15% in the low-risk group. Accordingly, the number of patients in EORTC groups 1 and 2 was much lower in the high-risk group, but still there was one EORTC group 1 patient and one EORTC group 2 patient among the 10 highrisk patients. In addition, although MEN 2B is considered to be a bad prognostic factor (23), all patients with familial MTC, and particularly two patients with MEN 2B, were found in the best prognostic group and were alive at the end of the study, 12.3 and 21.7 yr after initial surgery. Thus, calcitonin DT appears to be a good prognostic factor independently of the sporadic or familial type of MTC. However, calcitonin DT remained an independent predictor of survival by multivariate analysis [hazard ratio (HR) ⫽ 1.76; 95% confidence interval (95% CI) ⫽ 1.27–2.43; P ⫽ 0.002] demonstrating that this parameter carries additional information. The PVE is a measure of the correlation between the dependent variable (here survival) and the independent variable in the context of a multivariate analysis. In this study, the calcitonin DT had a much higher PVE (37.4%) than EORTC (12.1%) or TNM (5.4%) staging. Here, cut-off values play no role and the calcitonin DT was measured independently of survival. For the CEA-subgroup, EORTC score, calcitonin and CEA


J Clin Endocrinol Metab, November 2005, 90(11):6077– 6084 6081

TABLE 3. Clinical and biological prognostic variables Variables

Univariate

P value

Whole population (calcitonin DT calculated on all available data) Age ⱕ45 vs. ⬎45 yr NS 0.069 Gender NS 0.122 TNM stage NS 0.122 II vs. III NS 0.102 II vs. IV S 0.028 EORTC score S 0.001 EORTC group S 0.003 1–2 vs. 3 NS 0.299 1–2 vs. 4 –5 S 0.001 1/DT calcitonin S 0.0007 Subgroup with CEA DT available (n ⫽ 46) Age ⱕ45 vs. ⬎45 yr NS Gender NS TNM stage NS II vs. III NS II vs. IV NS EORTC score S EORTC group NS 1–2 vs. 3 NS 1–2 vs. 4 –5 NS 1/DT calcitonin S 1/DT CEA S

0.229 0.399 0.337 0.176 0.117 0.028 0.177 0.313 0.067