THE RELIABILITY OF HIGHLY ELEVATED CA 19-9 LEVELS

CA 19-9 is used as a tumour marker of the upper gastrointestinal tract. However, extremely elevated CA 19-9 levels are found also in patients with benign diseases. Cholestasis was present in 97.1 % of patients with high elevated CA 19-9, independent of their primary disease. 50% of patients with non-malignant diseases and increased CA 19-9 levels showed liver cirrhosis, cholecystitis, pancreatitis and/or hepatitis. In 8.8% no explanation was found for the extremely high CA 19-9 level. The results provide evidence of different factors influencing the CA 19-9 level.


INTRODUCTION
Highly elevated CA 19-9 levels are seen predominantly in patients with gastrointestinal malignancies.The higher the CA 19-9 level, the more likely is a diagnosis of malignancy.Some quite contrary clinical observations (Klee et aI. , 1993) provoked further investigation of the sensitivity and specificity of this marker.

MA TERIALS AND METHODS
832 CA 19-9 measurements have been analysed, in relation to clinical findings, from patients of the Departments of Surgery and the Department of Internal Medicine.Indication for measurement were suspicion of malignancy or monitoring of proven tumours.All measurements were performed using the CA 19-9 IMx® assay [enzyme immunoassay; normal range <45 U/ml] (Abbott).

RESULTS
We found highly elevated CA 19-9 levels (more than 500 U/ml) in 4.6% (n=38) ofthe total measurements.Levels between 45 U/ml and 499 U/ml occurred in 26.8% (n=223) , normal values were obtained in 68.6% (n=571).Cholestasis, accompanied by bilirubin levels from 3.8 mg/dl to 25.7 mg/dl was found in 97.1 % of the group with highly elevated CA 19-9, independent of the underlying diagnosis .
50% ofthe group with non-malignant diseases and highly elevated CA 19-9 levels had liver cirrhosis, cholecystitis, pancreatitis and/or hepatitis.41.2% of this group has bilirubin levels of more than 2.9 mgldl.In 8.8% an explanation forthe extremely high CA 19-9 level was not established.Cholestasis is well known as an important factor causing false positive CA 19-9 results.However, in this study we also found bilirubin levels of 13.8 mgldl and 11.2 mgl dl accompanied by CA 19-9 levels of 0.67 Vim I, and 44 Ulml respectively (Fig. 2).
The high rate of false negative CA 19-9 values seen in colorectal cancer (67%), stomach malignancies (49%), liver malignancies (75%) and in other tumours such as kidney, breast and ovary (72%) reflects the relatively poor sensitivity of 41.2%.This was even true for carcinomas of the biliary system, pancreas and small bowel (Fig. 3).The specificity was 71.8%.The positive predictive value of51 .8% and the negative predictive value of 62.4% emphasize the risk of false interpretation of CA 19-9 levels.

DISCUSSION
In evaluating CA 19-9 levels, a number of different factors possibly influencing the result must be considered.Iwase et a!. (1992) found CA 19-91evels in the cystic fluid of benign biliary liver cysts to be increased more than one hundred-fold compared with corresponding serum levels.In addition , they observed positive immunohistochemical staining for CA 19-9 in the cytoplasm of the epithelial cells ofthe cyst wall.Involvement of the bile duct epithelium in benign as well as malignant disorders may therefore account for elevated peripheral CA 19-9 levels.
In malignant tumours, the sensitivity and specificity of CA 19-9 are strikingly enhanced with advanced tumour stage (Kouri et a!., 1992) and are positively correlated with the grade of tumour differentiation (Iwamura et a!., 1992;Ohshio et a!., 1990).
False negative CA 19-9 values are common in patients who are negative for the Lewis blood group phenotype (Tempero et al., 1987).Kawa et al. (1991) examined different tumour markers in healthy subjects and patients with pancreatic carcinoma.Within the tumour group, 30% of the patients with false negative CA 19-9 levels were Lewis negative.However, the reason for negative CA 19-9 levels in the majority of cases of clinically proven pancreatic cancer is still unknown.
Cholestasis is the most common cause for false positive CA 19-9 levels, but even in the absence of gross cholestasis, hepatitis, liver cirrhosis, pancreastitis and cystic fibrosis can be associated with elevated serum CA 19-9 levels (lalanko et al., 1984;Duffy et al., 1985).
Increased CA 19-9 levels must therefore be interpreted with caution in the investigation of possible malignancy.Nevertheless it should be stressed that where there have been high preoperative CA 19-9 levels and a substantial drop after surgical treatment, CA 19-9 measurement represents one of the most important parameters in the monitoring of tumour growth.

Figure 3 .
Figure 3. Rate of false negative CA 19-9 levels in different tumor localizations.