A 51-year-old woman presented with progressive severe icterus associated with advanced inoperable pancreatic carcinoma. Biopsy of the liver revealed metastasis of a moderately differentiated adenocarcinoma, possibly of pancreatic origin. Imaging studies did not show any primary tumor in the pancreas but did reveal a deep vein thrombosis of the left lower extremity and a pulmonary embolism. Bone scans revealed metastases. The patient began treatment with chemotherapy (gemcitabine), radiotherapy to alleviate pain, and placement of a percutaneous transhepatic stent in the right biliary system because of progressive icterus.
Routine biochemical investigation of the patient’s serum revealed increased total bilirubin (275 μmol/L, reference value <22 μmol/L), which consisted mainly of direct bilirubin (215 μmol/L, reference value <4 μmol/L). The cytosolic liver enzymes alanine aminotransferase (115 U/L) and aspartate aminotransferase (143 U/L) were moderately increased. The serum protein concentration was 68 g/L (reference interval 60–85 g/L), but capillary electrophoresis of serum proteins (CAPILLARYS 2, Sebia) demonstrated a low albumin fraction together with a marked additional peak observed between the β and the γ region (Fig. 1⇓ ). Results of immunofixation with antibodies against G, M, and A heavy chains and κ and λ light chains were negative, indicating that this additional electrophoretic fraction did not indicate the presence of a paraprotein. Agarose gel electrophoresis confirmed the abnormal pattern, ruling out analytical interference caused by atypical ultraviolet absorbance.
Paraproteins are the most common cause of abnormal patterns in electrophoretic analysis results for plasma proteins. However, abnormal serum protein electrophoresis patterns may also indicate the presence of transitory plasma proteins originating from tissues as well as nonprotein interfering substances, most commonly iodinated contrast agents and antibiotics showing ultraviolet absorption at 200 nm, the wavelength used to quantify proteins in capillary electrophoresis (1). The presence of paraproteins can be excluded by use of immunofixation or immunoelectrophoresis. The presence of drug interferences can be excluded by reanalysis of the sample by use of dye-stained gel electrophoresis.
Electrochemiluminescence immunoassay (Modular E, Roche) of the patient’s serum revealed an exceptionally high concentration of the mucin tumor marker CA 19–9 (2.2 × 106 kU/L, reference value <37 kU/L). The serum concentration of carcinoembryonic antigen was also increased (423 μg/L, reference value <3 μg/L).
To further analyze the mucins we used high-pressure gel permeation chromatography performed by use of a Waters 650E advanced protein-purification system. We used the Wisp 712 automatic sampler to inject 25 μL of serum, and we performed chromatographic separations on an 80 × 300 mm Protein PAK glass 300 SW column (Waters Nihon Millipore). The obtained fractions were analyzed for CA 19–9, and this analysis showed that the elution of the CA 19–9 fraction mirrored the void volume of the column, a result corresponding to a molecular mass of >1000 kDa. Electrophoresis of the isolated CA 19–9 fraction confirmed that the mobility of the glycoprotein was in the β-γ region.
The patient’s serum showed pronounced increases in the membrane-bound liver enzymes alkaline phosphatase (1002 U/L, reference interval 30–120 U/L) and γ-glutamyltransferase (1153 U/L, reference interval 9–36 U/L). The molecular mass distribution of γ-glutamyltransferase and alkaline phosphatase activity levels (determined by use of high-pressure gel permeation chromatography) also showed a macromolecular character: 75.5% of serum γ-glutamyltransferase and 20.5% of alkaline phosphatase activity were attributable to a macromolecular complex, suggesting that membrane vesicles were present in the serum (2). After the serum sample had been extracted with 1-butanol, the macromolecular fraction of the membrane-bound enzymes vanished. In contrast, the molecular mass distribution of the CA 19–9 mucin was unchanged after 1-butanol extraction, a finding that argues against the hypothesis that the high apparent molecular mass of the mucin was attributable to lipid binding.
The CA 19–9 monoclonal antibody 1116 NS 19–9 (used in the assay) reacts with sialylated Lea-active pentasaccharide (sialylated lacto-N-fucopentaose II), which is enzymatically synthesized by sialylation of type 1 carbohydrate chains. CA 19–9 contains oligosaccharide structures present on heavily glycosylated high molecular mass mucins (3). The structure of the apomucin backbone typically reveals the presence of Ser/Thr/Prorich regions containing tandemly repeated stretches of amino acids that constitute potential O-glycosylation sites (4). Mucins are synthesized either as membrane-bound or as secreted glycoproteins. These molecules are widely synthesized and expressed by epithelial cells of the gastrointestinal, respiratory, and genitourinary tracts (4). In cells the increase in cAMP concentrations increases the synthesis and release of the carbohydrate antigen 19–9. cAMP is involved in the expression of glycoprotein-associated sialyl Lewis(a) antigen in LS174T cells (5).
The increased exposure of peptide epitopes of mucin glycoproteins in biliopancreatic cancer is due to abnormal glycosylation and/or altered transcription levels of mucin genes (3). The dosage of the Lewis gene [fucosyltransferase 3 (galactoside 3(4)-L-fucosyltransferase, Lewis blood group)] increases the amount of CA 19–9, whereas the dosage of secretor genes decreases it (6). CA 19–9 is the standard tumor marker for pancreatic cancer (7). Increased CA 19–9 concentrations are also found in other cancers, in chronic pancreatitis, and in benign gastrointestinal conditions (7). The majority of tumor cells in gastrointestinal carcinomas, including adenocarcinomas of the stomach, intestine, and pancreas, are strongly positive for CA 19–9 (8). In the patient we describe the cancer was accompanied by cholestasis. Controversy exists regarding the role of cholestasis as a physiopathological mechanism that affects serum CA 19–9 concentrations (7).
Little is known about the molecular heterogeneity of CA 19–9. In colon adenocarcinoma and cell plasma membranes this antigen is expressed on various glycoproteins with molecular masses ranging in size from ±100 kDa to >200 kDa. In cytosol and culture medium the epitope is carried by a single complex glycoprotein that has a very high molecular weight and resembles mucin (5). High molecular mass forms of CA 19–9 have been reported in serum from patients with pancreatic tumors (9).
After we excluded the possibility that the unusual serum electrophoresis pattern was due to the presence of paraproteins or to analytical effects attributable to ultraviolet absorption, we found that this case illustrates that excessive release of a mucin marker protein may affect the serum electrophoresis pattern. Because 1 U of CA19.9 antigen corresponds to approximately 0.8 ng of glycoprotein (10), we estimated that the patient’s serum contained approximately 1.6 g/L of CA 19–9 mucins, corresponding to 2.5% of the patient’s total serum protein, which can be detected by use of standard electrophoresis.
The patient showed a partial radiological and biochemical response to treatment (decrease of serum CA 19–9 from 2.2 × 106 to 5.5 × 104 kU/L). The decrease in CA 19–9 correlated with a decrease in the electrophoretic fraction migrating between the β and γ region. The patient died 13 days after admission.
POINTS TO REMEMBER
Drug interferences, paraproteins, and proteins originating from tissues can cause additional serum protein electrophoresis fractions.
In extreme cases, tumor markers such as CA 19–9 may be so abundant in serum or plasma that they become visible as an additional fraction on serum protein electrophoresis.
In such extreme cases, the laboratory should be proactive in further investigating unexpected results and should communicate appropriately with the clinical team. First, the presence of paraproteins should be ruled out by immunoelectrophoresis or immunofixation. If capillary electrophoresis has been used, then interfering substances that effect ultraviolet absorbance should be investigated (by carrying out standard gel electrophoresis in parallel). Further immunochemical investigations and biochemical characterizations can also be helpful.
Grant/Funding Support: None declared.
Financial Disclosures: None declared.
- © 2008 The American Association for Clinical Chemistry