Machens A, Ukkat J, Hauptmann S, Dralle H. Abnormal Carcinoembryonic Antigen Levels and Medullary Thyroid Cancer ProgressionA Multivariate Analysis. Arch Surg. 2007;142(3):289-293. doi:10.1001/archsurg.142.3.289
Medullary thyroid cancer cells are capable of secreting carcinoembyronic antigen (CEA). An abnormal preoperative CEA level may have important implications for the management of this condition.
Tertiary referral center at a university hospital.
One hundred fifty patients with a histopathologic diagnosis of medullary thyroid cancer and preoperative CEA measurements using the same assay.
Main Outcome Measure
We used univariate and multivariate analyses to quantify the relationship between preoperative CEA level and tumor progression.
On multivariate analysis, abnormal preoperative CEA levels were significantly associated with the initial operation rather than reoperation, larger primary tumors, positive lymph nodes, and distant metastasis. When analyses were limited to the 54 patients with increased CEA levels before the initial operation, there was a respective significant association between successive CEA levels (4.7-10.0, 10.1-30.0, 30.1-100.0, and >100.0 ng/mL) and lymph node metastases (>10 positive nodes: 0%, 9%, 53%, and 69% [P<.001]), involvement of cervical lymph node compartments (central: 33%, 36%, 73%, and 93% [P=.002]; lateral [ipsilateral]: 20%, 27%, 67%, and 88% [P=.001]; and lateral [contralateral]: 22%, 10%, 36%, and 73% [P=.008]), and distant metastasis (0%, 27%, 13%, and 75% [P<.001]). When CEA levels exceeded 30.0 ng/mL, surgical cure was exceptional.
In medullary thyroid cancer, an abnormal CEA level heralds advanced disease. Carcinoembryonic antigen levels greater than 30.0 ng/mL indicate central and lateral (ipsilateral) lymph node metastases, whereas CEA levels greater than 100.0 ng/mL signify lateral (contralateral) lymph node metastases and distant metastasis.
Early lymphatic and systemic dissemination of tumor cells is the hallmark of medullary thyroid cancer, a sometimes elusive neoplasm of neural crest derivation. Medullary thyroid cancer cells arise from parafollicular cells and have retained the capability of secreting calcitonin as a neuroendocrine peptide and carcinoembryonic antigen (CEA) as a serum protein. Unlike absolute calcitonin or CEA levels, serial measurements of calcitonin and CEA doubling times correlate well with tumor progression and survival.1
On the ultrastructural level, calcitonin is stored in dense-cored secretory granules,2 from which it can be released into the bloodstream by external stimuli such as intravenous administration of calcium or pentagastrin. This responsiveness to stimulation has rendered calcitonin an indispensable tool for early detection of occult medullary thyroid cancer, facilitating early surgical intervention and cure.3 Unlike calcitonin, CEA is integrated into the cell membrane,2 where it can be bound by monoclonal anti-CEA antibodies during immunoscintigraphy.4- 6 As a membrane-bound protein, CEA is less susceptible to stimulation and therefore less suited to identify occult disease.3,7 Based on semiquantitative and clinical data from no more than 41 patients with medullary thyroid cancer, CEA has been connected to increased tumor aggressiveness, tumor recurrence, and poor prognosis.7,8 Because tumor aggressiveness continues to be poorly defined, the surgical implications of an abnormal preoperative CEA level remain unclear.
During the past 2 decades, advances in surgical technology (optical magnification devices and biceps forceps coagulation) and imaging (magnetic resonance imaging, positron emission tomography, and positron emission tomography–computed tomography) have facilitated systemic microdissection of lymph nodes in the neck and mediastinum,9,10 refining the staging of these tumors. Apart from better surgical outcome, these advances now permit histopathologic quantification of tumor aggressiveness. This clinicopathological study was undertaken to clarify the surgical implications of an abnormal preoperative CEA level in a large series of patients with medullary thyroid cancer.
Based on the availability of a preoperative measurement of the serum CEA level using the same monoclonal electrochemiluminescence immunoassay (Elecsys; Roche Diagnostics, Mannheim, Germany; reference threshold level, <4.6 ng/mL), a total of 150 patients with medullary thyroid cancer (77 with initial operations and 73 with reoperations) were included in this retrospective study. Informed consent was obtained before each surgical procedure that represented standard practice of care in accordance with the practice guidelines of the German Society of Surgery.11 All 150 patients had undergone operations for therapeutic purposes at the Department of General, Visceral, and Vascular Surgery, Martin-Luther-University Halle-Wittenberg, for medullary thyroid cancer between July 1, 1997 (introduction of the electrochemiluminescence immunoassay), and October 31, 2005. All patients had undergone total thyroidectomy (n = 77 [51.3%]) or completion total thyroidectomy (n = 24 [16.0%]) at this institution or elsewhere (n = 49 [32.7%]). Systematic dissection of the central (n = 135 [90.0%]), lateral (ipsilateral [n = 124 (82.7%)] and contralateral [n = 115 (76.7%)]), and mediastinal lymph node compartment (n = 33 [22.0%]) was conducted as described previously.12 All specimens were submitted to pathological examination. Postoperative normalization of the serum calcitonin level, or biochemical cure, was assumed when the upper reference limit of 10.0 pg/mL for the assay (ELSA-hCT; CIS Bio International, Gif-sur-Yvette, France) was not exceeded basally or with pentagastrin stimulation. After obtaining informed consent, calcitonin stimulation was performed through intravenous bolus injection of pentagastrin (0.5 μg/kg of body weight; Pentavlon [Laboratoire SERB, Paris, France]); venous blood samples were drawn before the injection and 2 and 5 minutes thereafter.
After gross evaluation at the Department of Pathology, the entire thyroid gland was divided vertically to separate the left and right lobes. The thyroid halves were then sectioned horizontally from the superior to the inferior pole, as described previously.13 After fixation in formalin, the whole thyroid gland was embedded in paraffin. The soft tissue and lymph nodes were processed separately. Conventional staining (hematoxylin-eosin) and, where appropriate, calcitonin immunohistochemical analysis were performed on every surgical specimen, using the standard avidin-biotin-peroxidase complex approach. The diagnosis of medullary thyroid cancer was based on the World Health Organization histological classification of tumors.14 When more than 1 primary tumor was identified on histopathology, only the largest primary tumor was counted because this method was shown to yield comparable patterns of lymph node metastasis in multifocal and unifocal medullary thyroid cancer.15 Although a diagnosis of lymph node metastasis always required pathological confirmation, this need was waived for pulmonary metastasis when there was such unequivocal evidence on computed tomography, magnetic resonance imaging, positron emission tomography, scintiscan, or any combination thereof.
We tested categorical variables on univariate analysis using the 2-tailed Fisher exact test. We adjusted for multiple testing by using the Bonferroni correction. To study dose effects, continuous variables (primary tumor size and lymph node metastases) were grouped in increments of 10 mm and 5 lymph nodes, respectively. Multivariate conditional logistic regression models were then fitted to identify histopathologic variables associated with an abnormal preoperative CEA test result. The level of significance was set at P<.05.
Because medullary thyroid cancers tend to be more advanced and more difficult to cure at reoperation than at the initial operation, all patients were stratified by operative status and CEA test result. On univariate analysis (Table 1), an abnormal preoperative CEA level was associated with larger primary tumors, extrathyroidal growth, more lymph node metastases, and distant metastasis at both the initial operation and reoperation. Most of these associations were significant on univariate analysis. Sporadic and hereditary tumors were balanced among the groups, despite a certain tendency of hereditary tumors to be associated with a negative CEA test result owing to earlier diagnosis through genetic screening (detection bias). An abnormal preoperative CEA level was obtained more often before initial surgery (54/77 [70%]) than before reoperation (37/73 [51%]). As expected, postoperative normalization of increased calcitonin levels, or biochemical cure, was more often achieved at initial operation than at reoperation (28/70 [40%] vs 11/60 [18%]) (P = .008, 2-tailed Fisher exact test). These rates were highest with a negative CEA test result before the initial operation (12/15 [80%]; Table 1).
To quantify the effect of all histopathologic variables on the preoperative CEA test result, a multivariate logistic regression model was fitted (Table 2). Herein, the magnitude of the odds ratio reflects the strength of the association between the respective histopathologic variable and the preoperative CEA test result. On multivariate analysis, an abnormal preoperative CEA level was significantly associated with the initial operation rather than reoperation, larger primary tumors, lymph node metastases, and distant metastasis at the time of operation. Primary tumor size (≤30 mm) and lymph node metastases were connected to an abnormal preoperative CEA level in a dose-dependent fashion (Table 2). Beyond the 30-mm threshold, primary tumor size as a variable lost statistical significance in the multivariate model. As opposed to primary tumor size, extrathyroidal growth was not linked to an abnormal preoperative CEA level (Table 2).
The significant association of an abnormal preoperative CEA test result with advanced medullary thyroid cancer on multivariate analysis prompted additional investigations of dose-effect relationships in the 54 patients with increased preoperative CEA levels before the initial operation. These patients were categorized into 4 incremental CEA groups, with a minimum occupancy of 20% of eligible patients per group to ensure a more balanced distribution (Table 3). We found a respective significant association between successive CEA groups (4.7-10.0, 10.1-30.0, 30.1-100.0, and >100.0 ng/mL) and the number of lymph node metastases (>10 positive nodes: 0%, 9%, 53%, and 69% [P<.001]), involvement of various cervical lymph node compartments (central: 33%, 36%, 73%, and 93% [P = .002]; lateral [ipsilateral]: 20%, 27%, 67%, and 88% [P = .001]; and lateral [contralateral]: 22%, 10%, 36%, and 73% [P=.008]), the frequency of distant metastasis (0%, 27%, 13%, and 75% [P<.001]), and postoperative normalization of increased calcitonin levels, or biochemical cure (58%, 33%, 7%, and 0% [P = .001]). Only nominally significant were primary tumor size (grouped in 10-mm increments) and extrathyroidal growth. Neither variable retained statistical significance after correction for multiple testing. Sporadic and hereditary tumors were evenly balanced among the groups.
In this study, an abnormal preoperative CEA level was associated with the initial operation rather than reoperation, distant metastasis at the time of operation, and, in a dose-dependent fashion, the primary tumor size in 10-mm increments (up to 30 mm) and the number of lymph node metastases in increments of 5 positive nodes. The removal of tumor tissue elsewhere obviously reduced CEA serum concentrations before reoperation at our institution, sometimes beyond the cutoff level required for an abnormal preoperative CEA level. A similar phenomenon has been observed for calcitonin levels before reoperation.16 The almost linear, dose-effect relationship of an abnormal preoperative CEA level with incremental tumor size and number of lymph node metastases is indicative of a causal relationship, as is the magnitude of the odds ratio, a measure of the strength of the association.17 Beyond the 30-mm threshold, primary tumor size was no longer associated with an abnormal CEA test result, presumably because of the increasingly dominant contribution of extrathyroidal tumor mass. Most of this extrathyroidal tumor burden was controlled for in the multivariate model by the use of lymph node metastases and distant metastasis as independent variables, which may have dwarfed the relative contribution of very large primary tumors to the CEA test result. Extrathyroidal growth was unrelated to the preoperative CEA test result on multivariate analysis (Table 2). This finding implied that it is not the locally invasive growth pattern but rather the total mass of medullary thyroid cancer cells that accounts for overall CEA secretion, regardless of the location of those cells within or outside the thyroid gland.
To our knowledge, no large studies of a similar scope have yet been undertaken. Considering the ultrastructural localization of CEA in the cell membrane, CEA levels seem to correlate with overall tumor cell mass.3 In keeping with this notion, duration of medullary thyroid cancer response correlated with the duration of the blood CEA response in a phase 1 optimization clinical trial using an anti-CEA antibody for pretargeted radioimmunotherapy.18 In our study, increased preoperative CEA levels were significantly associated with the presence of more than 10 lymph node metastases. This constellation was encountered in one third of our patients with a positive CEA test result (Table 1) and in more than half of our patients before initial operation who had a CEA level greater than 30 ng/mL (Table 3). This condition almost always conflicts with surgical curability19- 21 and is frequently seen with distant lymph node metastases, ie, with involvement of the contralateral neck and/or the mediastinal lymph nodes,22 and with gross distant metastases.19 In agreement with these histopathologic data, CEA immunoscintigraphy often visualizes tumor deposits after initial operation in the mediastinum, liver, and bone.4- 6 Although absolute CEA levels may not be useful for prediction of survival,1 high preoperative CEA levels were found in this series to reflect extent of disease before the initial operation (Table 3). Preoperative CEA levels greater than 30 ng/mL were associated with an approximately 70% rate of involvement of the central and lateral lymph node compartment on the side of the primary tumor. This rate of involvement increased to approximately 90% when CEA levels exceeded 100 ng/mL. In this setting, involvement of the lateral lymph node compartment on the side opposite that of the primary tumor and distantmetastasis at the time of operation were seen in approximately 75% of cases. To ensure removal of all visible tumor deposits and to minimize any surgical morbidity, patients with medullary thyroid cancer, at least those with a high CEA level, should be referred only to select centers that have the surgical skills to deal with extensive disease in the neck and mediastinum. Given the longevity of many patients with extensive local disease, clearance of lymph node metastases is required to keep these tumor deposits from encroaching on adjacent structures such as major veins or recurrent laryngeal nerves, let alone the trachea and esophagus.23 These patients require systematic lymph node dissection.12,15,24
Because they herald the presence of larger tumors, regional lymph node metastases, and systemic metastases, preoperative CEA levels may be useful for planning the extent of lymph node dissection in patients with medullary thyroid cancer. Unlike calcitonin levels, which are susceptible to stimulation3,7 and hence tend to fluctuate on serial measurements, CEA levels are more stable. In the event of a negative CEA test result before the initial operation, total thyroidectomy with central lymph node dissection only may be adequate unless lateral cervical lymph node metastases are identified during the operation. An increased CEA level of 30 ng/mL or less is still compatible with local disease, which may be curable surgically. To this end, total thyroidectomy and systematic dissection of the central and both lateral lymph node compartments are advocated. In stark contrast, a CEA level greater than 30 ng/mL almost invariably signifies systemic disease that is intractable. Local palliation, which must be the prime goal in this situation, can be accomplished through total thyroidectomy and dissection of the central and lateral cervical lymph node compartments. When we consider our data (Table 3), dissection of the lateral cervical lymph node compartment on the contralateral side may not always be necessary for palliation, as long as no gross lymph node metastases are seen intraoperatively and the preoperative CEA level does not exceed 100 ng/mL. Before adoption of this preliminary guidance, confirmation of our findings in independent series is needed.
Correspondence: Andreas Machens, MD, Department of General, Visceral, and Vascular Surgery, Martin-Luther-University Halle-Wittenberg, Ernst-Grube-Straße 40, D-06097 Halle/Saale, Germany (AndreasMachens@aol.com).
Accepted for Publication: February 17, 2006.
Author Contributions:Study concept and design: Machens. Acquisition of data: Machens, Ukkat, Hauptmann, and Dralle. Analysis and interpretation of data: Machens, Ukkat, Hauptmann, and Dralle. Drafting of the manuscript: Machens. Critical revision of the manuscript for important intellectual content: Ukkat, Hauptmann, and Dralle. Statistical analysis: Machens. Administrative, technical, and material support: Machens, Ukkat, Hauptmann, and Dralle.
Financial Disclosure: None reported.