Abstract: To investigate the diagnostic accuracy and clinical efficacy of combined CEA-CA19-9 serum values in patients with pancreatic cancer. The serum levels of CEA and CA19-9 in 108 patients with pancreatic cancer and 72 patients with benign pancreatobiliary disease were evaluated during four years. Two positive tests of CEA and CA 19-9 were assumed as positive and at least one negative test as negative. The sensitivity, specificity, accuracy, positive and negative predictive value and positive and negative likelihood ratio of the combined CEA-CA19-9 test were calculated in diagnosing pancreatic cancer. The sensitivity and specificity of CEA were 83.78 and 69.44%, respectively. When serum values of CA19-9 were investigated between two groups, 89.24% of sensitivity and 89.88% of specificity were detected. The combination of CEA and CA19-9 revealed the highest sensitivity (82.46%) and specificity (100%). After construction of Receiver Operating Characteristics (ROC) curve, the Area Under Curve (AUC) for CEA was 0.76 (95% CI; 0.68-0.83), whereas that for CA19-9 was 0.86 (95% CI; 0.78-0.91). Both values were acceptable for clinical investigations but the AUC of CA19-9 was statistically more than the related area for CEA (p = 0.006). The AUC for the model containing combined evaluation of CEA-CA19-9 was 0.91 (95% CI; 0.85-0.96). Present study revealed more diagnostic accuracy with serum CA19-9 than with CEA in detecting pancreatic malignancies, but the accuracy could not be raised statistically by combining these tumor markers.
INTRODUCTION
The American Cancer Society estimated that 31, 860 Americans would be diagnosed with pancreatic cancer in 2004 and that 31, 270 would die from the disease. At the time of diagnosis, more than one half of pancreatic cancers have metastasized and only 8 percent are localized (Freelove and Walling, 2006). A patient history, physical examination and serum bilirubin and alkaline phosphatase levels can point to pancreatic cancer, but they are not diagnostic. Although conventional Computed Tomography (CT) and transabdominal ultrasonography are appropriate for initial imaging, dual-phase helical CT scanning is the best option if available (Freelove and Walling, 2006). Different tumor-associated antigens have been used to diagnose pancreatic cancer; however, they have not been satisfactory enough (Banfi et al., 1996; Futakawa et al., 2000; Kokhanenko et al., 2001; Ni et al., 2005; Ozkan et al., 2003; Pezzilli et al., 1995; Satake et al., 1994). Thus, there have been studies assessing the combination of different tumor-associated antigens. These studies have yielded promising accuracy in diagnosing benign from malignant cases (Lucarotti et al., 1991; Ozkan et al., 2003). Lucarotti et al. (1991) assessed the accuracy of combination of CEA, CA 19-9 and CA 50 which was associated with an overall correct classification of 91%. We aim to determine the efficacy of combined CEA-CA19-9 values in comparison with separate assessment of each tumor marker (CEA or CA19-9) in diagnosing pancreatic malignancies.
MATERIALS AND METHODS
This study was approved by the institution’s human subjects committee of Shahid Beheshti University of Medical Sciences and informed consent was obtained from the patients. We evaluated the serum levels of CEA and CA 19-9 in 108 patients with pancreatic cancer (PC group) referring to Taleghani Hospital of Shahid Beheshti University of Medical Sciences from 2000 to 2004. We also performed the same tests in a control group (Control group) which was selected from patients who were hospitalized for benign problems (e.g., cholecystitis, GI bleeding, choledochal cysts and…) in gastroenterology ward of the same center. We assumed two positive tests (CEA and CA 19-9) as positive and at least one negative test as negative. Then we calculated the accuracy measures (including sensitivity, specificity, accuracy, positive and negative predictive values and positive and negative likelihood ratios) of these tests and their combination in diagnosing pancreatic cancer.
Statistical analysis was performed using SPSS 12 for Windows. Continuous variable of age was analyzed using Student’s t-test. The comparisons of categorical variables (sex, CEA, CA 19-9 and their combination) were carried out utilizing the χ2 test or the Fisher's exact test as appropriate.
The area under a Receiver-Operating Characteristic (ROC) curve (AUC) was used as the main end-point of the study to decide whether which test has the best discriminator power and clinically more useful. The ROC curve can be constructed correlating true- and false-positive rates (sensitivity and 1-specificity, respectively) of the test. The AUC represents the probability that a random pair of test results will be ranked correctly as to their disease state (Hanley et al., 1982). Otherwise it gives the probability of the correct classification using the test in a sample pair of two patients (one with pancreatic cancer, one without pancreatic cancer). Theoretically, a 45-degree bisector would be a score predicting not better than a random guess. Thus, the AUC of this random score would be 0.5.
Usually tests with an AUC value of 0.7 and higher are accepted for clinical evaluations. Differences between the AUC values of both predictive tests and their combination were calculated, using the software MedCalc 8.2.1.0 (MedCalc Software, Mariakerke, Belgium). P-value <0.05 was considered statistically significant.
RESULTS
A total of 180 patients were evaluated during the study period (2000-2004). The mean age of them was 56.4 years. They consisted of 108 patients (male: 77, female: 31) with pancreatic malignancy and 72 (male: 35, female: 37) patients with benign pancreatobiliary disease. All patients with pancreatic cancer had documented pathologic confirmation of malignancy by surgical biopsy, endoscopic bushing or fine needle aspiration. The patients with malignancy were significantly older than patients with benign disease (p<0.001) and there were more men in malignant patients than benign control group (p = 0.002; Table 1).
Serum concentration of CEA and CA19-9 were assessed in both groups before any surgical interventions. Of the 108 patients with malignancy, 62 (83.8%) and 83 (89.2%) had positive results in laboratory analysis of CEA and CA19-9, respectively. Forty seven (79.7%) of the malignant patients were detected to have positive serum levels of CEA and CA19-9. The true negative rates of CEA and CA19-9 for malignancy were 69.4 and 88.9%, respectively in patients with benign pancreatobiliary disease. The results of biochemical evaluation in both groups are presented in Table 2.
| Table 1: | Demographic characteristics of patients in both groups |
| Table 2: | Laboratory analysis of patients in both groups |
| Table 3: | Statistical comparison of CEA, CA19-9 and CEA-CA19-9 for predicting pancreatic cancers |
| PPV: Positive Predictive Value, NPV: Negative Predictive Value, PLR: Positive Likelihood Ratio, NLR: Negative Likelihood Ratio | |
The sensitivity and specificity of CEA were 83.78 and 69.44%, respectively. When serum values of CA19-9 were investigated between two groups, 89.24% of sensitivity and 89.88% of specificity were detected. The combination of CEA and CA19-9 revealed the highest sensitivity (82.46%) and specificity (100%). We also realized that positive serum values of CEA and CA19-9 could be slightly more accurate for pancreatic cancers diagnosis compared with laboratory assessment of each marker, separately (92.25% vs. 76.71 and 89.09, respectively). As a result, serum values of CEA or CA19-9 in patients with malignant disease were significantly higher than the subjects in control group (Both, p<0.001). It is also necessary to add, this finding was shown to be same in combined evaluation of CEA-CA19-9 (p<0.001). Table 3 shows diagnostic values of these tumor markers for predicting pancreatic cancers.
To determine the efficacy of combined CEA-CA19-9 evaluation against separate assessment of them to predict malignancy, a receiver operating characteristics (ROC) curve was constructed. As it is presented in Fig. 1, the Area Under Curve (AUC) for CEA was 0.76 (95% CI; 0.68-0.83), whereas that for CA19-9 was 0.86 (95% CI; 0.78-0.91). Both values were acceptable for clinical investigations but the AUC of CA19-9 was statistically more than the related area for CEA (p = 0.006). The AUC for the model containing combined evaluation of CEA-CA19-9 was 0.91 (95% CI; 0.85-0.96). Although this value was slightly more than AUC of CA19-9 but the statistical difference could not be revealed (p = 0.083).
| Fig. 1: | Receiver Operating Characteristic (ROC) curves of using CEA, CA 19-9 and their combination in the diagnosis of pancreatic carcinoma. For CEA, AUC = 0.76 (95% CI; 0.68-0.83); for CA19-9, AUC = 0.86 (95% CI; 0.78-0.91) and for CEA-CA19-9, AUC = 0.91 (95% CI; 0.85-0.96) |
DISCUSSION
Although pancreatic cancer accounts for only 2% of all new cancers in the United States, it is the 4th leading fatal cancer. The overall five-year survival rate is 4%; however, localized cancers have a 17% survival rate (Freelove and Walling, 2006). This fact warrants a great effort to diagnose this fatal malignancy as soon as possible. The US Preventive Services Task Force (USPSTF) does not recommend screening of average-risk asymptomatic patients with abdominal palpation, ultrasonography, or serologic tumor markers (Freelove and Walling, 2006), But a non-invasive accurate diagnostic modality can have a handsome potential of turning into a cost-effective screening test. Serum x levels of tumor-associated antigens have been studied as non-invasive methods to diagnose pancreatic cancer (Banfi et al., 1996; Futakawa et al., 2000; Kokhanenko et al., 2001; Ni et al., 2005; Ozkan et al., 2003; Pezzilli et al., 1995; Satake and Takeuchi, 1994). However, they were not satisfactory enough as a sole diagnostic method. Thus, combinations of these antigens have been used as more accurate diagnostic modalities for pancreatic cancer. Ni et al. (2005) have reported a sensitivity of 92% for combination of CEA and CA 242 while Lucarotti et al. (1991) stated an accuracy of 91% for combination of CEA, CA 19-9 and CA 50.
Qin et al. (2004), prospectively compared serum CA19-9 concentrations with CEA in three categories of patients with cholangiocarcinoma, benign biliary diseases and healthy individuals. He reported the sensitivity of 77.14 and 68.57% for CA19-9 and CEA values in cholangiocarcinoma, respectively and explained that serum CA19-9 and CEA concentrations were significantly elevated in patients with cholangiocarcinoma compared with the other two groups. He also suggested that serum CA19-9 determination is a useful addition to the available tests for differential diagnosis of cholangiocarcinoma; furthermore sensitivity and specificity could be raised by combining of these two markers.
Our study revealed more diagnostic accuracy with serum CA19-9 than with CEA in detecting pancreatic malignancies, but the accuracy could not be raised statistically by combining these tumor markers. We also found that the application of CEA and CA19-9 could be effective to differentiate cancer from benign pancreatobiliary disease because of high sensitivity and specificity in our study.
Since the serum levels of CA19-9 could increase in some benign diseases such as pancreatitis, cholangitis, hepatitis and cirrhosis as well as malignant conditions (Akdogan et al., 2001; Giannini et al., 2000; Mann et al., 2000), we believe that we can even sustain better results by choosing our control group as a real sample of normal population instead of patients in gastroenterology ward, although none of the patients in our control group had signs or symptoms of mentioned diseases. However we suggest that CA19-9 values determination could be postponed in patients suspicious to have pancreatic malignancies after recovery from benign conditions which interfere with the results.
The next step of this agenda could be designing more extensive studies to evaluate the efficacy of CEA-CA19-9 combinations in screening pancreatic cancer in early, localized stages which has a 4 times greater five-year survival (17 vs. 4% (Freelove and Walling, 2006) than more advanced stages of this malignancy.