Overall survival curves according to perioperative change in carcinoembryonic antigen levels among 55 patients undergoing staged hepatectomy combined with portal vein embolization. The 1-year, 3-year, and 5-year survival rates, respectively, were 81.8%, 47.7%, and not reached in group A; 96.8%, 66.1%, and 46.2% in group B; and 84.6%, 35.6%, and 0.0% in group C. A significant difference was observed between group B and group C (P = .007). No significant difference was observed between group A and group B (P = .19) or between group A and group C (P = .67).
Overall survival curves according to perioperative change in carcinoembryonic antigen levels among 158 patients seen with initially resectable liver metastases. The 1-year, 3-year, and 5-year survival rates, respectively, were 93.1%, 78.5%, and 54.8% in group A; 93.7%, 73.6%, and 36.6% in group B; and 85.0%, 21.2%, and 0.0% in group C. Significant differences were observed between group A and group C and between group B and group C (P < .001 for both). No significant difference was observed between group A and group B (P = .24).
Overall survival curves according to Fong et al4 clinical risk score. The 1-year, 3-year, and 5-year survival rates, respectively, were 97.7%, 88.6%, and 61.1% in patients with clinical risk scores of 0 to 1; 91.1%, 65.2%, and 35.1% in patients with clinical risk scores of 2 to 3; and 88.2%, 38.7%, and not reached in patients with clinical risk scores of 4 to 5. Significant differences were observed between patients with clinical risk scores of 0 to 1 vs 2 to 3 (P = .01), between patients with clinical risk scores of 0 to 1 vs 4 to 5 (P < .001), and between patients with clinical risk scores of 2 to 3 vs 4 to 5 (P = .001).
Disease-free survival curves according to Fong et al4 clinical risk score. The 1-year, 3-year, and 5-year disease-free survival rates, respectively, were 86.4%, 55.4%, and 24.5% in patients with clinical risk scores of 0 to 1; 74.0%, 29.3%, and 16.9% in patients with clinical risk scores of 2 to 3; and 58.8%, 7.1%, and not reached in patients with clinical risk scores of 4 to 5. Significant differences were observed between patients with clinical risk scores of 0 to 1 vs 2 to 3 (P = .02), between patients with clinical risk scores of 0 to 1 vs 4 to 5 (P < .001), and between patients with clinical risk scores of 2 to 3 vs 4 to 5 (P = .02).
Overall survival curves according to perioperative trend of carcinoembryonic antigen levels in 135 patients with Fong et al4 clinical risk scores of 2 to 3. Significant differences were observed between group A and group C and between group B and group C (P < .001 for both). No significant difference was observed between group A and group B (P = .98).
Elie Oussoultzoglou, Edoardo Rosso, Pascal Fuchshuber, Victor Stefanescu, Bala Diop, Giorgio Giraudo, Patrick Pessaux, Philippe Bachellier, Daniel Jaeck. Perioperative Carcinoembryonic Antigen Measurements to Predict Curability After Liver Resection for Colorectal MetastasesA Prospective Study. Arch Surg. 2008;143(12):1150–1158. doi:10.1001/archsurg.143.12.1150
Perioperative carcinoembryonic antigen (CEA) blood level is a predictor of outcome after resection of colorectal liver metastases (CLMs).
Prospective clinical study.
Department of digestive surgery and transplantation.
Between January 1, 2000, and December 31, 2004, CEA levels were routinely measured 1 week before and 6 weeks after CLM resection in 213 patients. The patients were divided into the following 3 groups: group A (n = 69) with normal preoperative and postoperative CEA levels, group B (n = 111) with elevated preoperative and normal postoperative CEA levels, and group C (n = 33) with elevated preoperative and postoperative CEA levels.
Main Outcome Measures
The use of perioperative CEA levels to predict outcome after resection.
The median survival was 45.4 months. The 5-year overall and disease-free survival rates were 50.2% and 21.9%, respectively, in group A, 38.5% and 18.3% in group B, and 0.0% and 0.0% in group C (P < .001). Univariate analysis showed that patients with elevated preoperative and postoperative CEA levels, multiple CLMs, large CLMs (≥5 cm), advanced Fong clinical risk score, bilobar distribution, and hepatic pedicle lymph node involvement had significantly poorer overall and disease-free survival. By multivariate analysis, only perioperative CEA level, hepatic pedicle lymph node involvement, and number and size of CLMs were independent prognostic factors. The 5-year survival rates showed good correlation with perioperative CEA levels in all 3 patient groups.
The predictive value of perioperative CEA levels is demonstrated. Carcinoembryonic antigen levels as early as 6 weeks after surgery may be helpful in assigning patients to adjuvant chemotherapy after resection of CLMs.
Liver resection has been recognized as the only potentially curative treatment for patients with colorectal liver metastases (CLMs).1,2 Despite curative intent, most patients who undergo surgical resection do not have long-term survival.1- 4 Prognostic factors and clinical scoring systems for long-term survival and recurrence have been described for use after potentially curative CLM resection.1,3- 24
However, the available prognostic factors and the current definition of curative resection consider only the resection margins and the absence of extrahepatic disease.4,23,25 They fall short of predicting overall curability after CLM resection because they evaluate only a static aspect of a progressive disease. Therefore, prognostic factors reflecting the biologic dynamics of the disease after surgical resection might be more helpful as predictors of outcome.
Carcinoembryonic antigen (CEA) blood level is the most widely used tumor marker to detect recurrences after resection of primary colorectal cancer and CLMs.26- 29 Considering the short half-life of CEA,30 this tumor marker may be of clinical value for a dynamic evaluation of this disease. Few series have studied the prognostic value of preoperative and postoperative changes in CEA levels among patients undergoing CLM resection.31,32 The aim of the present study was to prospectively evaluate the use of preoperative and postoperative CEA measurements to predict cure after CLM resection in a large single-institutional series.
During the study period, preoperative and postoperative CEA levels were measured according to an established protocol and were collected prospectively. The preoperative CEA measurement was performed within 1 week before liver resection. Blood samples for postoperative CEA measurement were obtained 6 weeks after liver resection during a short hospitalization scheduled for removal of a transcystic biliary drain routinely placed at the time of liver resection. Both CEA measurements were performed in our hospital laboratory using an electrochemiluminescence immunoassay (Elecsys 2010; Roche Diagnostics GmbH, Mannheim, Germany). A normal CEA level defined by this test is less than 5.0 ng/mL (to convert CEA level to micrograms per liter, multiply by 1.0). Patients who underwent noncurative liver resection (ie, grossly incomplete liver resection or complete resection with persistent extrahepatic disease) were excluded from this study. Patients in whom complete resection of all known extrahepatic disease was achieved at the time of liver resection were included. In patients who underwent planned 2-stage hepatectomy, only preoperative and postoperative CEA levels corresponding to the second liver resection were analyzed. The criteria for performance of preoperative portal vein embolization or staged hepatectomy (1-stage or 2-stage) were previously described.33,34 Briefly, they concern patients with initially nonresectable liver metastases (ie, multiple bilobar or small future remnant liver). Major liver resection was considered when at least 3 contiguous liver segments (as defined by Couinaud35) or 2 segments of the hepatic dome (IVb, VII, and VIII) were resected. The decision to perform hepatic lymph node dissection was based on prognostic factors described in a previous prospective study.14
Between January 1, 2000, and December 31, 2004, 233 patients were operated on for CLMs. Of these, 215 patients underwent curative R0 liver resection, defined as complete and microscopic margin free. Two patients were excluded because of incomplete CEA measurements. The remaining 213 patients constitute the study population. These patients were divided into 3 groups according to preoperative and postoperative CEA levels. Group A comprised 69 patients with normal preoperative and postoperative CEA levels. Group B comprised 111 patients with elevated preoperative and normal postoperative CEA levels. Group C comprised 33 patients with elevated preoperative and postoperative CEA levels.
Patients who had undergone neoadjuvant chemotherapy were referred to our department for further treatment planning. Among them, 23 patients (13.2%) had normal CEA levels and were assigned to group A. The decision to give additional adjuvant therapy was made after receiving final pathologic examination results in a multidisciplinary conference that included oncologists, gastroenterologists, radiologists, and surgeons. A change in CEA level from before to 6 weeks after liver resection was not considered in the decision to give adjuvant therapy in these patients. For patients treated with neoadjuvant chemotherapy, liver resection was delayed until at least 3 weeks after the last cycle. Postoperative adjuvant chemotherapy was postponed until measurement of the 6-week postoperative CEA level. This was facilitated by the fact that patients were admitted to the hospital 6 weeks after surgery for routine removal of the external transcystic biliary drain placed at the time of liver resection.
Before hepatectomy, all patients underwent abdominal and thoracic computed tomography, as well as liver magnetic resonance imaging. To rule out primary tumor recurrence before resection of metachronous metastases, pelvic magnetic resonance imaging or rectal endoscopic ultrasonography was used in patients with rectal primary tumors, and colonoscopy was used in patients with colon primary tumors. Positron emission tomography was not routinely used for preoperative staging at the time of the study.
All patients were followed up every 3 to 6 months with a physical examination, CEA measurement, abdominal ultrasonography, or computed tomography. No patient was lost to follow-up. The diagnosis of recurrence was based on postoperative imaging or biopsy results. Patients with persistent elevated CEA levels 6 weeks after liver resection were not considered as having recurrences. In patients whose CEA levels remained elevated beyond 6 weeks on routine follow-up examinations, further imaging studies and biopsies as indicated were performed to detect measurable recurrent disease.
Fisher exact, χ2, and Mann-Whitney tests were used. Survival was calculated from the date of surgery. Kaplan-Meier survival was calculated, and significant differences were examined using log-rank test. Multivariate analysis using Cox proportional hazards models was used to identify independent risk factors for overall and disease-free survival. A difference was considered statistically significant if P < .05. Only variables with P < .20 on univariate analysis were included in the multivariate analysis. All statistical calculations were performed using commercially available software (SPSS 10.0; SPSS Inc, Chicago, Illinois).
They were 136 male and 77 female patients. The mean (SD) age was 63.7 (9.6) years (age range, 39.7-84.2 years). Sixty-one patients were 70 years or older. The primary tumor was located in the colon in 164 patients and in the rectum in 49 patients. According to the pTNM classification, the primary tumor was staged pT1 or pT2 in 27 patients and pT3 or pT4 in 186 patients. The primary tumor lymph nodes were invaded in 135 patients, not invaded in 77 patients, and unknown in 1 patient. In group A patients, the carbohydrate antigen (CA) 19-9 level before hepatectomy was normal (<34 U/mL) in 56 patients, mildly elevated (34-44 U/mL) in 10 patients, and highly elevated (298-402 U/mL) in 3 patients (to convert CA 19-9 level to kilounits per liter, multiply by 1.0). The mean (SD) CA 19-9 level was 31.4 (69.5) U/mL (range, 1-402 U/mL).
Colorectal liver metastases were metachronous in 107 patients (50.2%) and bilobar in 90 patients (42.3%) (Table 1). The mean (SD) size of the largest liver metastases was 46.8 (27.4) mm (range, 15-160 mm). The mean (SD) interval between primary tumor resection and liver resection was 21.6 (17.8) months (range, 3-96 months). Eighty-two patients had solitary liver metastases, 36 had 2 metastases, 36 had 3 metastases, and 59 had more than 3 metastases.
Twenty-seven patients underwent associated procedures at the time of hepatectomy. These included resection of colorectal primary tumors (n = 12), resection of solitary peritoneal implants (n = 12), partial diaphragmatic resection (n = 2), and right adrenal gland resection (n = 1). Hepatic pedicle and retroduodenopancreatic lymph node dissection was performed in 132 patients. Among them, 12 patients (9.1%) had microscopic hepatic pedicle lymph node metastases discovered at pathologic examination. No patient had macroscopic evidence of hepatic pedicle lymph node involvement.
The 3 study groups were similar in age, sex, primary tumor site, pTNM stage, type and topography of liver metastases, disease-free interval between primary tumor resection and diagnosis of liver metastases, number of liver metastases, size of largest liver metastases, presence of resectable solitary peritoneal implant, hepatic pedicle lymph node involvement, and neoadjuvant and adjuvant chemotherapy regimens. However, preoperative portal vein embolization was significantly more frequent in group C (39.4% [13 of 33 patients]) than in group A (15.9% [11 of 69 patients]) or group B (27.9% [31 of 111 patients]) (P =.03).
Major hepatectomy was performed in 98 patients. Preoperative right portal vein embolization was performed in 55 patients (28 before planned 1-stage hepatectomy and 27 before planned 2-stage hepatectomy). Liver resections included 47 bisegmentectomies (10 requiring additional subsegmental metastasectomies), 32 right hepatectomies (3 requiring additional metastasectomies in the left lobe), 23 extended right hepatectomies (7 requiring additional metastasectomies in the left lateral lobe), 5 left hepatectomies (1 requiring an additional metastasectomy), 2 extended left hepatectomies, 16 left lobectomies (13 requiring additional metastasectomies in the right lobe), 53 unisegmentectomies (18 requiring additional metastasectomies), 7 trisegmentectomies (1 requiring left lobectomy and 1 requiring an additional metastasectomy), and 28 subsegmental metastasectomies.
Adjuvant chemotherapy regimens included the following: fluorouracil plus folinic acid in 73 patients, fluorouracil plus folinic acid and oxaliplatin in 67 patients, and fluorouracil plus folinic acid and irinotecan in 38 patients.
The 1-year, 3-year, and 5-year overall survival rates in 55 patients who underwent staged hepatectomy (90.9%, 56.5%, and 29.4%, respectively) were similar to those in 158 patients who did not undergo staged hepatectomy (92.4%, 69.8%, and 38.4%, respectively) (P =.09). Moreover, the perioperative change in CEA level significantly predicted survival in patients undergoing staged hepatectomy combined with portal vein embolization (Figure 1 and Figure 2). Finally, the 5-year survival rate was significantly lower in patients undergoing 2-stage hepatectomy for initially nonresectable disease (18.9%) compared with that in patients with initially resectable liver metastases (38.4%) (P =.02).
The mean (SD) follow-up after hepatectomy was 34.7 (17.2) months (range, 3.1-78.5 months). Among 114 surviving patients (53.5%), 64 patients were recurrence free after a mean (SD) follow-up of 37.0 (16.5) months (range, 10.3-78.5 months). Fifty patients had recurrences after a mean (SD) follow-up of 44.6 (17.1) months (range, 20.0-77.4 months). The remaining 99 patients (46.5%) died after a mean (SD) follow-up of 28.1 (14.9) months (range, 3.1-66.3 months) after hepatectomy. Death was directly related to cancer relapse except in 2 patients, one with cardiac infarction 34.2 months after hepatectomy and the other with head and neck cancer 22.7 months after hepatectomy. Recurrences included the following: liver (n = 72), lung (n = 57), peritoneum (n = 14), bone (n = 10), primary tumor site (n = 9), disseminated disease (n = 9), retroperitoneal lymph nodes (n = 6), brain (n = 3), adrenal glands (n = 1), and ovary (n = 1). Perioperative change in CEA level did not correlate with site of recurrence. The overall median survival was 45.4 months. The 1-year, 3-year, and 5-year overall survival rates were 92.0%, 66.3%, and 35.8%, respectively. Disease-free survival rates were 74.1%, 32.3%, and 17.7% at 1 year, 3 years, and 5 years, respectively.
Univariate analysis showed that the status of hepatic pedicle lymph nodes, CEA level 6 weeks after hepatectomy (P < .001), perioperative change in CEA level, and number, size, and distribution of liver metastases were significant prognostic factors of overall survival and disease-free survival (Table 1). Analyzed as a continuous variable, the CEA level 6 weeks after hepatectomy also affects survival (P < .001). The primary tumor lymph node status, disease-free interval between primary tumor resection and diagnosis of liver metastases, CEA level before hepatectomy (<200 vs ≥200 ng/mL), and CEA level before neoadjuvant chemotherapy (normal vs abnormal) did not reach statistical significance (P = .06). Finally, the CEA level before hepatectomy (normal vs abnormal, P =.19) and the CEA level before neoadjuvant chemotherapy (<200 vs ≥200 ng/mL, P =.94) did not affect survival.
Analysis of the change in CEA level during neoadjuvant chemotherapy was performed by excluding patients with normal CEA levels before and after chemotherapy. The remaining 156 patients who received neoadjuvant chemotherapy were divided into the following 2 groups: group 1 comprising 49 patients with an increase in CEA level after completion of chemotherapy and group 2 comprising 107 patients with a stable CEA level or a decrease in CEA level after completion of chemotherapy. Univariate analysis showed that change in CEA level after neoadjuvant chemotherapy was also a predictor of overall survival (1-year, 3-year, and 5-year overall survival rates, respectively, were 91.8%, 54.7%, and 19.3% in group 1 vs 90.7%, 67.3%, and 34.9% in group 2; P =.048). However, disease-free survival was unaffected by change in CEA level after neoadjuvant chemotherapy (1-year, 3-year, and 5-year disease-free survival rates, respectively, were 69.3%, 12.7%, and 8.4% in group 1 vs 69.2%, 30.6%, and 16.4% in group 2; P =.12).
In group A patients, the preoperative CA 19-9 level did not affect survival (4-year survival, 56.6% for patients having normal levels vs 46.9% for patients having abnormal levels; P =.17). Patients having recurrence with an increased CEA level had similar survival vs those having recurrence with a normal CEA level (5-year survival, 28.8% vs 36.5%; P =.27).
In multivariate analysis, perioperative change in CEA level, hepatic pedicle lymph node metastases, and number and size of liver metastases were significant and independent prognostic factors for overall and disease-free survival (Table 2). In addition, sex was a significant prognostic indicator for disease-free survival.
The clinical risk score proposed by Fong et al4 for prediction of long-term outcomes after liver resection of CLMs was validated in the present series. Our study population was divided into the following 3 groups: 44 patients with a clinical risk score of 0 to 1, 135 patients with a clinical risk score of 2 to 3, and 34 patients with a clinical risk score of 4 to 5. As expected, patients with a clinical risk score representing more advanced disease (0-1 vs 2-3 vs 4-5) had significantly poorer overall and disease-free survival (P < .001 for both) (Figure 3 and Figure 4). For patients with a clinical risk score of 2 to 3, change in perioperative CEA level significantly correlated with survival. For these patients, 5-year survival rates were 45.0%, 40.2%, and 0.0% in groups A, B, and C, respectively, and the median survival times were 53.8, 45.5, and 28.3 months, respectively (Figure 5). For patients with clinical risk scores of 0 to 1 or 4 to 5, no such correlation was found, most likely because of small sample sizes.
The present prospective study demonstrates that postoperative CEA measurement, particularly preoperative and postoperative CEA levels after CLM resection, is a significant prognostic factor for overall and disease-free survival. Furthermore, normalization of CEA levels 6 weeks after CLM resection may indicate improved long-term outcomes in these patients.
Numerous tumor markers have been described to assess patients with colorectal cancer, including levels of p53, CEA, cytokeratin, vascular endothelial growth factor, and CAs such as CA 19-9, CA 242, and CA 72-4.36- 40 Among these markers, CEA level remains the most widely used. For colorectal cancer, the prognostic value of preoperative CEA measurement was highlighted by several studies.41- 45 Other studies32,46- 51 demonstrated the benefit of systematic postoperative CEA monitoring to diagnose early recurrence after resection or to assess response to chemotherapy in patients with CLMs. However, few studies32,51 have studied the usefulness of perioperative changes in CEA level for evaluating the success of surgical resection (ie, whether a presumed complete resection as defined by intraoperative and pathologic findings is effectively curative). The present study shows that patients with elevation of both preoperative and postoperative CEA levels have significantly lower median survival (28.3 months for group C) compared with patients with normal postoperative CEA levels (53.8 months for group A and 45.5 months for group B), indicating that postoperative CEA levels can be used as a marker for complete vs incomplete (cytoreductive) surgery. Therefore, patients with elevated postoperative CEA levels are particularly vulnerable to recurrence and should be considered for additional treatment. Finally, in the present series, the finding of the worst prognosis in patients requiring 2-stage hepatectomy was related to the predominance of these patients in group C. In fact, because of their higher disease burden, they are at higher risk of noncurative resection, as reflected by higher elevated postoperative CEA levels and worse outcomes. However, in selected patients, planned 2-stage hepatectomy remains the only surgical option that offers patients a chance of long-term survival.
The potential clinical value of preoperative and postoperative CEA levels was underscored several years ago by Hohenberger et al.32 They demonstrated that postoperative CEA level was the most important predictive factor for overall and disease-free survival among 141 patients who underwent R0 resection for CLMs. They suggested that CEA level was the best criterion to judge whether curative resection has been performed. However, their study included only patients with unilobar resectable disease and no more than 3 liver metastases. Moreover, data about chemotherapy were not reported. It is notable that preoperative vs postoperative change in CEA level remained the most powerful predictor of long-term outcomes in the present series, which included patients with more advanced metastatic disease such as multiple and bilobar metastases, patients with initially nonresectable disease, and patients requiring staged hepatectomy combined with portal vein embolization.
The clinical risk score developed by Fong et al4 and validated by Mann et al22 was also predictive of outcome and survival in the present series. In our study, persistent elevated postoperative CEA levels after CLM resection identified a subgroup of patients with the worst prognosis. This was particularly true for patients with a clinical risk score of 2 to 3. The clinical risk score is based on 5 static preoperative variables and evaluates disease status before CLM resection. The inclusion of postoperative CEA measurement allows assessment of tumor behavior in sequence, particularly in patients with elevated preoperative CEA levels. This measurement better identifies those patients with the best chance for cure after metastasectomy. For patients with normal preoperative CEA levels, the clinical risk score is the only good prognostic tool available because CA 19-9 is of no prognostic value in this situation, as shown by the present study and in a previous study by Weber et al.13
The reported rates of normal preoperative CEA levels in patients with CLM range from 16% to 35.1%.22,52 In a previous study13 of patients with CLMs operated on between 1988 and 1998, the percentage of patients with normal preoperative CEA levels was 19.5% (43 of 221 patients). At that time, patients received neither oxalipatin- nor irinotecan-based chemotherapy regimens. Therefore, in the present series, the wide range in the rates of patients with normal preoperative CEA levels could be explained by the effect of neoadjuvant chemotherapy on preoperative CEA levels. Indeed, 69 patients (32.4%) (group A) had normal preoperative CEA levels. This rate is an overestimation of patients not expressing CEA because 174 study patients (81.7%) received neoadjuvant chemotherapy, which led to a normalization of CEA levels in 23 patients (13.2%).
Several previous studies reported the usefulness of measuring changes in CEA level after complete tumor resection. The reported median half-life of CEA is 4 days (range, 1-88 days).30,53 This suggests that premature measurement of postoperative CEA levels may overestimate the number of patients with persistent CEA level elevation. Conversely, it has been demonstrated that prolonged CEA half-lives are often associated with the existence of overlooked synchronous metastases.53 We chose a 6-week interval for postoperative CEA measurements because we found that it accurately identified patients with persistent CEA level elevation without delaying the initiation of adjuvant chemotherapy. These arguments were sufficient for us to stop routine early postoperative CEA measurements after liver resection.
Carcinoembryonic antigen levels are an established prognostic indicator for monitoring disease progression and for early detection of recurrence.32,46- 51 The present series demonstrated better prognosis for patients having normal preoperative CEA levels compared with patients having elevated preoperative and postoperative CEA levels. However, the survival rates were similar in both groups of patients with normal postoperative CEA levels (groups A and B). There is no evidence that CEA-producing CLMs have more aggressive behavior than non–CEA-producing CLMs. Indeed, the primary tumor and CLM characteristics were similarly distributed in the 3 groups of patients. This study demonstrates that persistently elevated postoperative CEA levels indicate the presence of unidentified residual disease, with subsequent poor prognosis and increased risk of relapse. In this series, postoperative CEA measurements in patients with elevated preoperative CEA levels were useful in predicting ultimate surgical failure after intent-to-cure CLM resection. Elevated CEA levels in the immediate postoperative period can identify the subgroup of patients with clinically undetectable residual disease who will benefit from postoperative chemotherapy. In these patients, postoperative chemotherapy should be regarded as therapeutic rather than as adjuvant. Elevated CEA levels in the early postoperative period are not used as an indication for surgical intervention.
Most series about CLMs have reported various prognostic scoring systems based on different prognostic factors.1,3,4,10,17,22,24,54 All of them are useful in selecting patients for surgery and for predicting their outcomes. However, the major concern today is to better define the radicality of potentially curative resection rather than to identify patients at risk of recurrence. More efficient chemotherapy, multimodality treatment approaches, and innovation in surgical strategies (2-stage hepatectomy resection, portal vein embolization, and radiofrequency ablation) have increased the numbers of patients considered for liver resection and more frequently include patients initially considered as having nonresectable disease. The present study demonstrates the use of combined preoperative and postoperative CEA measurements to identify patients who had incomplete resection in the setting of intent-to cure metastasectomy for colorectal cancer. This study also shows that postoperative CEA measurements are of limited prognostic value in patients with normal preoperative CEA levels.
In conclusion, preoperative CEA measurement is a well-recognized prognostic factor. This study demonstrates that it is not a good predictor of postoperative outcome or of biologic aggressiveness of the disease. Preoperative CEA measurement did not predict performance of complete resection, nor did it accurately predict postoperative disease progression. However, our study showed that postoperative CEA level 6 weeks after CLM resection and its normalization represent the most reliable predictive criteria for complete and successful hepatic metastasectomy.
Correspondence: Daniel Jaeck, MD, PhD, Centre de Chirurgie Viscérale et de Transplantation, Hôpital de Hautepierre, Hôpitaux Universitaires de Strasbourg–Université Louis Pasteur, Avenue Molière, 67200 Strasbourg, France (Daniel.Jaeck@chru-strasbourg.fr).
Accepted for Publication: September 17, 2007.
Author Contributions:Study concept and design: Oussoultzoglou, Rosso, Fuchshuber, Giraudo, and Jaeck. Acquisition of data: Oussoultzoglou, Rosso, Stefanescu, Diop, and Giraudo. Analysis and interpretation of data: Oussoultzoglou, Rosso, Pessaux, and Bachellier. Drafting of the manuscript: Oussoultzoglou, Rosso, Fuchshuber, Stefanescu, Diop, Giraudo, and Bachellier. Critical revision of the manuscript for important intellectual content: Oussoultzoglou, Rosso, Fuchshuber, Pessaux, and Jaeck. Study supervision: Oussoultzoglou, Fuchshuber, Pessaux, Bachellier, and Jaeck.
Financial Disclosure: None reported.