CEA as a prognostic index in colorectal cancer

Introduction: The carcinoembryonic antigen, CEA, is the tumor marker most used in colorectal patients, principally during follow up after radical surgery. High serum CEA levei before surgery is often associated with worse prognosis, in some studies. Objective: The purpose of this study was to evaluate the preoperative carcinoembryonic antigen leveis (CEA) and the frequency of recurrence. Material and Methods: Eighty-three patients with colorectal cancer at Dukes stages A, B or C were evaluated retrospectively. The patients'follow up was at least two years or to death. CEA was detemined in serum by enzyme immunoassay (Sorin Biomedica), normal value 0-5ng/ml. Results: Disease recurrence was observed in 32 patients (38.5%), 13 Dukes B and 19 Dukes C. Seventy five per cent of the patients with CEA higher than 1Ong/ml relapsed and 80% of the patients without recurrence had normal CEA. Disease recurrence in patients with preoperative elevated CEA occurred during the first year of follow up in 56% of the patients. Conclusion: Although the tumor stage is today the most valuable prognostic variable in colorectal cancer, the preoperative CEA value can provide some additional information in the prognosis of the patient.

Material and Methods: Eighty-three patients with colorectal cancer at Dukes stages A, B or C were evaluated retrospectively.
The patients'follow up was at least two years or to death.CEA was detemined in serum by enzyme immunoassay (Sorin Biomedica), normal value 0-5ng/ml.Results: Disease recurrence was observed in 32 patients (38.5%), 13 Dukes B and 19 Dukes C. Seventy five per cent of the patients with CEA higher than 1Ong/ml relapsed and 80% of the patients without recurrence had normal CEA.Disease recurrence in patients with preoperative elevated CEA occurred during the first year of follow up in 56% of the patients.

Conclusion:
Although the tumor stage is today the most valuable prognostic variable in colorectal cancer, the preoperative CEA value can provide some additional information in the prognosis of the patient.

INTROOUCTION
C olorectal cancer is one of the most common cancers in industrialized cities.In Brazil, it is the 4th largest cause of cancer in women and the 6th in men.In 1995,22000 persons had colorectal cancer and 3000 died from this cancer in this country.
The carcinoembryonic antigen (CEA), first described by Gold and Freedman(') in 1965, is a glycoprotein present in a high number (90%) of colorectal adenocarcinoma.Serum determination of CEA is the the most widely used tumor marker in these patients.The main function of this marker is to detect early recurrence during follow up after radical surgery.The serum dosage of CEA at diagnosis has low sensitivity because patients with colorectal cancer can have normal leveIs of serum CEA.The relationship between high CEA serum leveIs in the preoperative period and worse

Address for correspondence:
Nora Manoukian Forones Disciplina de Gastroenterologia -UNIFESP Rua Botucatu, 720 São Paulo/SP -Brasil -CEP 04023-000 prognosis has been investigated in several studies 2 ,3,4,5, so far without definite conclusion.Patients with high CEA serum leveIs before surgery could have a higher incidence of recurrence than patients with normal leveIs.
The aim ofthis study was to retrospectively evaluate the serum CEA leveIs before surgery in operable colorectal cancer patients as a predictor of recurrence.The possible relationships between serum leveIs of CEA and sex, age, diameters of the tumor, histological diffentiation, tumor site (colon or rectum) and Dukes stage have also been evaluated.Thirty years after its description by Gold and Freedman(l), the CEA remains the tumor marker most studied in colorectal cancer, its main function being the monitoring of patients after the surgical resection of the tumor.The preoperative CEA leveI as a prognostic index is not yet defined, although many authors believe that the CEA leveI offers an additional criterion for evaluating the prognosis of colorectal cancer.

PATIENTS ANO METHOOS
As in the majority of previous studies(2,4),we also did not observed any differences of CEA leveIs when comparing sex, age, tumor site, tumor diameter or histological differentiation.Carpellan et.al. (6), observed higher CEA sensitivity leveIs for rectal cancer than for colon cancer,  4).The Chi-square test between the patients with or without recurrence according to the serum CEA leveIs either normal or elevated, was statistically different (p= 0.008).
The sensitivity, specificity, the positive predictive value and the negative predictive value are described in Table 5.
The frequency of recurrence according to Dukes stage was 28% for the B stage and 56% for the C stage.In Dukes stage B, the CEA leveI was elev~ted in 30% of the patients who developed recurrence, and normal in 82% of patients without recurrence.In Dukes stage C, 19 patients developed recurrence, of which 12 (63%) had a CEA leveI greater than 5ng/mI.Among the other 15 patients without relapse, 10 (67%) had normal serum leveIs.These differences were not significant by the Chi square test (p>0.05). in 34).All the patients were folIowed for at least two years or up to death.
Serum samples were taken preoperatively, within 30 days of surgery and stored at -20°C.Serum leveIs of CEA were measured by Elisa, using the Sorin Biomedica kit.Normal values are from 0-5ng/mI. .We calculated the sensitivity, specificity and positive and negative predictive value for CEA as an index oftumor recurrence.Statistical analysis consisted of Kruskal-Wallys analysis of variance or Chi square test.Significance was defined as p < 0.05.

RESULTS
The mean and standard mean error of CEA values in Dukes stages A, B and C were 2 :t 1.96; 5.8 :t 5.82; 23.6 :t 23.59 (Tablel).There was no significant difference with respect to sex (p=0.94),age before or after 45 years (p=0.94),cancer site (p=0.77),diameter of the tumor (p=0.96), or histological differentiation (p=0.87).
The CEA values were differentiated in alI three stages by the Kruskal-Wallys analysis (p= 0.043), in which patients with.Dukes stage' A were the same as those with stage B, but those with stage C were different from both A and B. Among the 46 patients with Dukes stage B, 9 (34%) had elevated serum CEA leveis and among the 34 patients with stage C, 17 (50%) had elevated CEA serum leveI.The 3 patients with Dukes'stage A had normal serum CEA leveIs (Table2).
Twenty six patients (31 %) had CEA leveIs greater than 5ng/ml (Table 2).Disease recurrence was found in 32 patients, 13 Dukes B and 19 Dukes C, 22 had metastases, most ferquently in the liver (57%), and 10 had local recurrence.CEA leveIs were elevated in 7 of the 12 patients who developed liver metastases.Before the completion of 1 year of follow up, 12 patients had relapsed.Among the 32 patients with recurrence of the disease during the follow up, 16 had preoperative elevated CEA serum leveIs and 16 had normal CEA leveIs.In 9 of the 16 patients with an elevated preoperative CEA, the recurrence appeared during the first year of follow up, while in the 16 with normal serum CEA levei the diagnosis of recurrence was done after 1 year in 81 % of the patients (Table 3).This differenc~was significant by the Chi square test (p= 0.04).
CEA preoperative leveIs were normal in 50% of the patients who developed recurrence and elevated in the others, and therefore patients with normal serum CEA leveI can develop recurrence during follow up.The sensitivity was lower than in the studies of Wang(9) and Tate(lO)but similar to McCall(\ I).Seventy five per cent of pittients with CEA leveI higher than 10ng/ml developed recurrence, but on the other hand, among the 50 patients without recurrence, the CEA leveI were normal in 80%.Tsushiya et al(4)observed that patients with preoperative CEA leveI greater than 10ng/ml in Dukes stage B or C had decreased survival rates.In the group of patients with elevated preoperative CEA and recurrence, this occurred in the first year of follow up.Chu (7) and Groslin(3) aIso reported that patients with eIevated CEA leveI had recurrence after the first 6 months.Carpelan(12) found a significant difference in survi vaI in patients with preoperative serum CEA above 5ng/mI and those with a lower CEA leveI.Northover(\3) reported that the risk of recurrent disease within two years of primary surgery was more than double in those whose serum CEA was raised preoperati vel y.
In this study we can conclude that preoperative CEA higher than 10ng/ml was associated with poorer prognosis.In these patients the recurrence wouId be more frequently in the first year of follow up.The results from this series are consistent with several published works.Although the information provided by the tumor stage is still the most valuable prognostic variable today, the preoperative elevated CEA leveI provides additional information on the prospects for survival, suggesting a more aggressive tumor and probabIy identifying a subgroup of patients for adjuvant therapy.

Table 3 Number of patients with CEA > or < 5ng/ml and the time of recurrence Table 5 Number of patients with CEA > or < 5ng/ml and with or without recurrence
The sensitivity ofCEA at the time of diagnosis ofthe tumor at Dukes stages A or B was 18%.and 50% at stage C, which was a result similar to Chu et alO).The mean serum leveIs were higher for tumors in Dukes stage C than for stages B or A. The differences in serum CEA were statistically significant.The CEA leveI differentiated Dukes stage A from C, and C from B, but not A from B.