Elevated tumour marker CA19-9: clinical interpretation and influence of obstructive jaundice

Transcription

1 European Journal of Surgical Oncology 2000; 26: doi:.53/ejso , available online at on Elevated tumour marker CA19-9: clinical interpretation and influence of obstructive jaundice D. V. Mann, R. Edwards, S. Ho, W.Y.Lau and G. Glazer Departments of Surgery, St Mary s Hospital, London, UK and Prince of Wales Hospital, The Chinese University of Hong Kong, Shatin, Hong Kong Aims: The tumour marker CA19-9 has been promoted as a reliable test for the detection of pancreatobiliary malignancy, yet its diagnostic role remains poorly defined. In this study the clinical interpretation of a raised serum CA19-9 level has been evaluated, with particular reference to obstructive jaundice. Methods: One hundred and sixty-four patients with a CA19-9 level above 33 U/ml were studied. Serum CA19-9 was compared with clinical diagnosis and correlated with serum bilirubin level. In a subgroup of jaundiced patients (16 benign and 15 malignant cases), follow-up CA19-9 levels were determined 2 weeks after biliary drainage. Results: The median CA19-9 level was lower in benign cases (2 (IQR ) U/ml) than those with pancreatobiliary tumours (9 (IQR ) U/ml; P<0.01), although the overlap was substantial. In benign jaundiced cases, a positive correlation was observed between bilirubin and CA19-9 elevation (R=0.41, P<0.01). Relief of jaundice was associated with a fall in CA19-9 level in all benign cases and in nine of the 15 with malignancy. Conclusion: Confident discrimination between benign and malignant disease could not be made on the basis of a solitary elevated CA19-9 measurement. Hyperbilirubinaemia was associated with a further deterioration in specificity and caution is warranted when interpreting the results in jaundiced patients. Overall, only one half of patients with an elevated CA19-9 level ultimately proved to harbour a malignancy Harcourt Publishers Ltd Key words: tumour marker; pancreatobiliary malignancy; obstructive jaundice. Introduction makes its diagnostic role questionable in icteric patients. 7 Indeed, this research was stimulated by our own experience The carbohydrate antigen CA19-9 is commonly expressed of patients presenting with jaundice of benign aetiology in the cells of tumours of the upper gastrointestinal tract, 1 who were found to have grossly elevated CA19-9 levels. and measurement of serum glycoprotein CA19-9 has been The purpose of this study was to clarify the clinical suggested as a reliable means of diagnosing malignant interpretation and diagnostic value of an elevated serum pancreatobiliary neoplasms. 2 Despite early promise, this CA19-9 level, with special reference to coexistent obstructive tumour marker has failed to gain an established role in jaundice. clinical practice, partly due to uncertainty of the predictive value of a positive test. It is now appreciated that this marker is not exclusively associated with malignant processes, and Patients and methods elevated circulating levels have been reported in a wide range of benign conditions including liver disease, 3,4 ascending The patients enrolled in this study were admitted to St cholangitis 5 and pancreatitis. 6 A particular difficulty arises Mary s Hospital (London) or the Prince of Wales Hospital in patients when the finding of an elevated CA19-9 level (Hong Kong) during a 4-year period from 1993 to leads to a diagnosis of presumed upper gastrointestinal Although most of the cases were managed primarily, in tract malignancy. many instances the CA19-9 measurement was performed The relationship between jaundice and the biology of by another clinician, with subsequent referral of the patient. CA19-9 is incompletely understood, and further confounds To ensure completeness, we also analysed the hospital the clinical situation. It is known that serum CA19-9 may tumour marker registry to identify patients with an elevated be elevated in the presence of hyperbilirubinaemia, which CA19-9 level during this period. One hundred and sixty- four patients found to have an elevated serum CA19-9 measurement (>33 U/ml) were evaluated. Liver Presented in abstract form at the meeting of the Pancreatic biochemistry, other relevant tumour markers, ultimate Society of Great Britain and Ireland, November Correspondence to: Mr G. Glazer, Consultant Surgeon, 84a St clinical diagnosis and supporting radiological/pathological Johns Wood High Street, London NW8 7SH. Tel: +44 (0) evidence were analysed. The distinction between benign and 3020; Fax: +44 (0) ; g.glazer@ic.ac.uk malignant disease processes was based on the pathological, /00/ $35.00/ Harcourt Publishers Ltd

2 operative (including endoscopic) findings and clinical course (including post-mortem where available). Only those patients in whom a confident clinical diagnosis could be established were included in the analysis. To investigate the effect of jaundice on CA19-9 level, a subseries of 31 patients with obstructive jaundice were studied before and after biliary drainage. These patients were selected when relief of jaundice was achieved by stone extraction, stent placement or operative (palliative) bypass. Patients with malignant disease undergoing resection or chemo/radiotherapy were specifically excluded as the aim was to determine any (reversible) influence of jaundice alone on CA19-9 level. Serum CA19-9 was measured by a standardized laboratory assay (CA19-9 microparticle enzyme immunoassay, Abbott Laboratories, Chicago, USA) which was identical at the two hospital sites. The data were analysed by distribution-free (nonparametric) methods. Unpaired data were compared using Kruskal Wallis Analysis of Variance by Ranks (K-W ANOVA) and the Mann Whitney U-test with Bonferroni correction for multiple comparisons. Repeated measurements (paired data) were analysed by the Wilcoxon Matched Pairs test. The data are presented as median with interquartile range (IQR) in parentheses. Proportions were evaluated by Fisher s Exact Probability. Correlation studies were conducted in the standard non-parametric manner for the Spearman Rank Order correlation coefficient (R). Statistical significance was accepted at the P<0.05 level. Statistical analyses were performed using STATISTICA version 4.0 (Statsoft, Inc., OK, USA) for the personal computer. Elevated tumour marker CA Jaundiced Non-jaundiced Benign Pancreatobiliary primary Liver metastases Clinical diagnosis Other primary Fig. 1. CA19-9 elevation and clinical diagnosis in jaundiced and non-jaundiced cases. biliary group there was no overall difference between the individual tumour types (Table 2). Co-existing jaundice (serum bilirubin >17 μmol/l) and elevated CA19-9 was present in 46 of the benign and 52 of the malignant group of patients (Table 1). Where the jaundice was of benign aetiology a positive correlation was observed between serum CA19-9 level and bilirubin concentration (R=0.41, P<0.01) (Fig. 2). This relationship Results did not apply to malignant jaundiced cases, in whom no correlation was found (R= 0.04, P=0.8) (Fig. 3). One hundred and sixty-four patients with elevated CA19-9 Thirty-one patients with obstructive jaundice were studied levels were studied. Of these, 90 (55%) cases had malignant before and after biliary drainage. There were 16 with benign pathology and 74 (45%) were ultimately found to have aetiology (gallstones, n=13; chronic pancreatitis, n=3) and benign disease (Table 1). For the purposes of evaluation, 15 with a malignancy (pancreatic carcinoma, n=5; the patients with malignant disease were allocated to one ampullary carcinoma, n=3; cholangiocarcinoma, n=5; and of three subgroups: a) primary pancreatobiliary tumours, metastatic colonic carcinoma, n=2). Relief of obstruction including hepatoma (n=44); b) metastatic liver disease at was achieved in this subset of patients by means of either presentation (n=28); and c) other primary neoplasms (n= stone removal, biliary stenting (endoscopic/percutaneous) 18). There was extensive overlap of individual CA19-9 values or operative (palliative) bypass. Resolution of jaundice was for all diagnostic groups, irrespective of the presence of confirmed biochemically by a decline in bilirubin in both jaundice (Fig. 1). The positive predictive value for groups (Figs 4 and 5). At this time point (which was at malignancy of an elevated CA19-9 level was 58% in non- least 2 weeks after intervention), a repeat measurement of jaundiced and 53% in jaundiced cases. Overall differences CA19-9 revealed a fall in the serum level for all benign in serum CA19-9 between the (sub)groups were confirmed cases (median CA (IQR 1333) U/ml at by K-W ANOVA (H=40.6, 3 d.f., P<0.01), and individual presentation vs 96 (IQR ) U/ml after relief of jaundice, subgroups were then compared with the Mann Whitney U- T=0, Z=3.5, P<0.01 by Wilcoxon Marched Pairs test) test using the Bonferroni correction. The median CA19-9 (Fig. 6). Within the malignant group a variable response was lower in the group with benign disease (2 (IQR was seen (Fig. 7). Three cases of ampullary carcinoma and ) U/ml) than those with primary pancreatobiliary two each of pancreatic, cholangio and metastatic colonic tumours (9 (HQR ) U/ml; U=646, P<0.01) and carcinoma responded with a decline in serum CA19-9. patients presenting with metastatic liver disease (1793 (IQR Conversely, three patients each with pancreatic ) U/ml; U=570, P<0.01). In contrast, patients with adenocarcinoma and cholangiocarcinoma were found to primary malignancies at other sites had a median CA19-9 have a further elevation of the tumour marker at the time comparable to those with benign disease (55 (IQR of re-measurement, despite a similar fall in bilirubin. This ) U/ml; U=420, P=0.14). Within the pancreato- differential response following relief of jaundice (elevation

3 476 D. V. Mann et al. Table 1. Patient characteristics Benign (n=74) Malignant (n=90) Diagnosis Jaundiced Non-jaundiced Diagnosis Jaundiced Non-jaundiced Gallstones 25 4 Pancreatic carcinoma 17 8 Liver disease 12 3 Liver metastases Chronic pancreatitis 5 3 Ampullary carcinoma 3 1 Diverticular disease 0 4 Cholangiocarcinoma 8 1 Heart failure 2 0 Hepatoma 3 1 Urinary tract infection 0 2 Gallbladder carcinoma 2 0 Other (peptic ulcer, liver 2 12 Other primary (colorectal, 5 13 abscess etc) gastric, breast etc) Total Table 2. Pancreatobiliary tumour site and CA19-9 elevation E+08 Tumour Number CA19-9 Median (IQR) U/ml Pancreas ( ) Ampullary ( ) Cholangio/gallbladder ( ) Hepatoma ( ) No overall difference between subgroups by K-W ANOVA, P= Fig. 3. Correlation between elevated CA19-9 and bilirubin for malignant obstructive jaundice (exponential line fit, R= 0.04, P=0.8). it is appreciated that in many centres this tumour marker is not used routinely, in our experience patients are often referred with the finding of an elevated level and a presumptive diagnosis of an upper gastrointestinal tract malignancy. This report therefore represents a selected group of patients, but nevertheless addresses an important clinical Fig. 2. Correlation between elevated CA19-9 and bilirubin for problem. It should be emphasized that we have not benign obstructive jaundice (exponential line fit, R=0.41, P<0.01). attempted to redefine the overall test characteristics of CA19-9, which requires an analysis of those individuals with a normal CA19-9 level, since these parameters have in 0/16 benign and 6/15 malignant cases) was confirmed on been extensively documented. 8,9 Nor have we set out to Fisher s Exact Probability (P=0.007). exhaustively compare CA19-9 with alternative tumour With respect to other tumour markers, serum markers in every instance, since relative discriminative carcinoembryonic antigen (CEA, normal level < μg/l) was powers for the commonly used antigens have been elevated in all 13 patients with confirmed colorectal liver previously reported. metastases (median CEA 153 (IQR ) μg/l), and Clinical interpretation of CA19-9 measurement requires positively correlated with CA19-9 level (R=0.8, P<0.05). an appreciation of the normal biology of this tumour There were four cases of hepatocellular carcinoma, all of whom had diagnostic elevation of serum alpha fetoprotein in excess of 0 μg/l (normal range < μg/l). Discussion This study has focused on the clinical interpretation and diagnostic value of an elevated serum CA19-9 level. While marker. The CA19-9 antigen is defined by a murine monoclonal antibody (1116-NS-19-9), originally generated against a human colorectal carcinoma cell line. 11 The epitope is an oligosaccharide corresponding to sialylated Lewis blood group Le a12 (and is therefore not expressed in the 5% of the population lacking the Lewis gene). In tissues the antigen occurs as a glycolipid (monosialoganglioside), but in the serum as a mucin (carbohydrate-rich glycoprotein). 13 Although initially derived from a large bowel cancer,

4 Elevated tumour marker CA = Gallstones = Chronic pancreatitis 500 = Pancreatic carcinoma = Ampullary carcinoma = Cholangiocarcinoma = Metastatic tumour After relief jaundice 0 After relief of jaundice Fig. 4. Changes in bilirubin before and after relief of jaundice for benign cases. Fig. 5. Changes in bilirubin before and after relief of jaundice for malignant cases (dashed lines correspond to cases with continued elevation of CA19-9 shown in Fig. 7). 0 1 = Gallstones = Chronic pancreatitis After relief of jaundice Fig. 6. Changes in serum CA19-9 before and after relief of jaundice for benign cases. measurement of the antigen in the blood is more sensitive for upper gastrointestinal malignancies, particularly pancreatic carcinoma. 8 Using cut-off points defined from serum levels in healthy populations (usually U/ml) the marker is elevated in patients with tumours of the pancreatobiliary system, although this is not an exclusive association and high circulating levels are seen with other gastrointestinal and non-gastrointestinal cancers. 2 Overall, the assay has a reported sensitivity of 80% and specificity of 90% for the diagnosis of pancreatic carcinoma. 14,15,16,17 Initial enthusiasm for the CA19-9 marker has waned recently with the appreciation that serum levels may also be elevated in a wide variety of benign conditions (gastrointestinal and extra-intestinal) 9, and the diagnostic accuracy is diminished in these circumstances. The extensive overlap in CA19-9 values from benign and malignant cases shown in the current study indicates that confident discrimination can rarely be made on the basis of a single measurement. It is recognized that jaundice can cause an elevation of serum CA19-9, 17,18 although the nature of the interaction is not fully understood. In this study we have demonstrated that in jaundice due to benign disease there is a positive correlation between serum bilirubin and CA19-9 elevation, whereas no such relationship exists for malignant jaundiced cases. The explanation probably relates to the synthesis of CA19-9 by normal biliary tract epithelium, 1 and evidence that the liver actively clears glycoproteins from the circulation. 19 The healthy biliary tract may be thought of

5 478 D. V. Mann et al = Pancreatic carcinoma = Ampullary carcinoma = Cholangiocarcinoma = Metastatic tumour 1 After relief of jaundice Fig. 7. Changes in serum CA19-9 before and after relief of jaundice for malignant cases (dashed lines indicate cases with continued elevation of CA19-9; the corresponding changes in bilirubin are indicated in Fig. 5). 48 h 21 and several weeks 18,22 have been reported. Comparisons are difficult to make because of variations in case mix and therapy. In this study we have focused on the relief of mechanical biliary obstruction, but surgical resection of pancreatobiliary tumours may similarly result in a decline in CA (in the absence of residual disease) and the response of CA19-9 to chemo/radiotherapy has been used as a prognostic indicator. 23 How can the interpretation of an elevated serum CA19-9 level be rationalized? We have shown that it is not possible to reliably distinguish benign from malignant disease processes on the basis of this tumour marker. This is particularly relevant in the presence of jaundice, and such patients should not necessarily be assumed to harbour a malignancy. Overall, the positive predictive value of an elevated CA19-9 in this study was only of the order of 50%. Thus, we would caution the use of a single (static) measurement of this marker in the routine work-up of patients. The role of planned serial (dynamic) measurement will need to be established by ongoing work, although it is possible to draw some tentative conclusions from the findings of this study. For the patient with obstructive jaundice (and a high serum CA19-9), the radiological and pathological investigation will naturally focus on the liver, bile ducts and pancreas. If diagnostic uncertainty persists, one possibility may be to relieve the jaundice and then repeat the CA19-9 assay over a period of 2 3 weeks. A rising level in the absence of further jaundice is suggestive of an underlying malignancy, although a falling value does not exclude the diagnosis. The finding of a high serum level in a non-icteric patient should prompt investigation for an occult gastrointestinal or other neoplasm, and thereafter for a plausible benign cause. It remains to be seen whether newly developed markers of malignancy can improve on the performance of CA19-9 in these clinical contexts. 24,25 In the meantime, while the diagnosis of malignancy in the biliary tree or pancreas depends ultimately on histological or cytological proof, on many occasions this may be difficult to obtain, and decisions will continue to be made on clinical and radiological grounds. as an excretion pathway for CA19-9, and high levels have been measured in normal bile. 5,20 Obstruction to bile flow from any cause might therefore be expected to produce an increase in the serum level of CA19-9. In malignant disease, however, synthesis of CA19-9 antigen by the proliferating cells would be expected to contribute to the total level to a varying degree (depending on tumour size, stage and grade) 14 in a manner independent of any associated jaundice. Because of the reduced diagnostic accuracy of serum CA19-9 in the presence of jaundice, some authors have adopted a higher cut-off point in this context, thereby References improving the test specificity. 15,21 In this study, we have evaluated the effect of hyperbilirubinaemia by examining 1. Atkinson B, Ernst C, Herlyn M, Steplewski Z, Sears H, the pattern of change of CA19-9 after the relief of jaundice. Koprowski H. Gastrointestinal cancer-associated antigen in In benign cases, resolution of cholestasis was uniformly immunoperoxidase assay. Cancer Res 1982; 42: associated with a fall in serum CA19-9, consistent with the 2. Steinberg W. The clinical utility of the CA19-9 tumour- observed correlation between the two variables. In the associated antigen. Am J Gastroenterol 1990; 85: Collazos J, Genolla J, Ruibal A. CA19-9 in non-neoplastic malignant group a differential response was found. A decline liver diseases: a clinical and laboratory study. Clin Chim Acta in CA19-9 level was noted for the majority of tumours, 1992; 2: however in other instances an elevation of the marker 4. Deugnier Y, Rabot A, Guyader D, et al. Serum increase and occurred (despite a similar fall in bilirubin). These data are liver overexpression of carbohydrate 19-9 antigen in patients consonant with the theory of dual contribution to the serum with genetic haemochromatosis. Gut 1994; 35: Albert M, Steinberg W, Henry J. Elevated serum levels of CA19-9 level from biliary obstruction and tumour cell tumour marker CA19-9 in acute cholangitis. Dig Dis Sci 1988; synthesis. Continued malignant cellular proliferation may 33: therefore result in a static or rising CA19-9 level after relief 6. Yoshida E, Scudamore C, Erb S, Owen D, Silver H. Markedly of jaundice. elevated CA19-9 levels in a case of chronic pancreatitis. Can J Surg 1995; 38: The time scale of alterations in serum CA19-9 following 7. Ritts R, Nagorney D, Jacobsen D, Talbot R, Zurawski V. endoscopic or surgical treatment is incompletely Comparison of preoperative serum CA19-9 levels with results understood, and changes at varying time points between of diagnostic imaging modalities in patients undergoing

6 Elevated tumour marker CA laparotomy for suspected pancreatic or gallbladder disease. 17. DelFavero G, Fabris C, Plebani M, et al. CA19-9 and Pancreas 1994; 9: carcinoembryonic antigen in pancreatic cancer diagnosis. 8. Ritts R, DelVillano B, Go V, Herberman R, Klug T, Zurawski Cancer 1986; 57: V. Initial clinical evaluation of an immunoradiometric assay 18. Paganuzzi M, Onetto M, Marroni P, et al. CA19-9 and CA50 for CA19-9 using the NCI serum bank. Int J Cancer 1984; 33: in benign and malignant pancreatic biliary diseases. Cancer ; 61: Frebourg T, Bercoff E, Manchon N, et al. The evaluation of 19. McFarlane IG. Hepatic clearance of serum glycoproteins. Clin CA19-9 antigen level in the early detection of pancreatic cancer. Sci 1983; 64: Cancer 1988; 62: Strom B, Iliopoulos D, Atkinson B, et al. Pathophysiology of. Jalanko H, Kuusela P, Roberts P, Sipponen P, Haglund C, tumour progression in human gallbladder: flow cytometry, Makela O. Comparison of a new tumour marker, CA19-9, CEA, and CA19-9 levels in bile and serum in different stages with alphafetoprotein and carcinoembryonic antigen in patients of gallbladder disease. J Natl Cancer Inst 1989; 81: with upper gastrointestinal diseases. J Clin Pathol 1984; 37: 21. Benamouzig R, Buffet C, Fourre C, Ink O, Moati F, Etienne J. Serum levels of carbohydrate antigenic determinant (CA Koprowski H, Steplewski Z, Mitchell K, Herlyn M, Herlyn D, 9) in obstructive jaundice. Dig Dis Sci 1989; 34: Fuhrer P. Colorectal carcinoma antigens detected by hybridoma 22. Basso D, Meggiato T, Fabris C, et al. Extra-hepatic cholestasis antibodies. Somat Cell Mol Genet 1979; 5: determines a reversible increase of glycoproteic tumour markers 12. Itzjowitz S, Kim Y. New carbohydrate tumor markers. in benign and malignant diseases. Eur J Clin Invest 1992; 22: Gastroenterology 1986; 90: Magnani J, Steplewski Z, Koprowski H, Ginsburg V. 23. Willett CG, Daly WJ, Warshaw AL. CA19-9 is an index of Identification of the gastrointestinal and pancreatic cancercancer. response to neoadjunctive chemoradiation therapy in pancreatic associated antigen detected by monoclonal antibody 19-9 in Am J Surg 1996; 172: the sera of patients as a mucin. Cancer Res 1983; 43: Parker N, Makin CA, Ching CK, et al. A new enzyme- 14. Safi F, Roscher R, Beger H. Tumour markers in pancreatic linked lectin/mucin antibody sandwich assay (CAM 17.1/WGA) cancer: sensitivity and specificity of CA19-9. assessed in combination with CA19-9 and peanut lectin binding Hepatogastroenterology 1989; 36: assay for the diagnosis of pancreatic cancer. Cancer 1992; 70: 15. Steinberg W, Gelfand R, Anderson K, et al. Comparison of the sensitivity and specificity of the CA19-9 and carcinoembryonic 25. Yiannakou J, Newland P, Calder F, Kingsnorth A, Rhodes J. antigen assays in detecting cancer of the pancreas. Prospective study of CAM17.1/WGA mucin assay for Gastroenterology 1986; 90: serological diagnosis of pancreatic cancer. Lancet 1997; 349: 16. Haglund C. Tumour marker antigen CA125 in pancreatic cancer: a comparison with CA19-9 and CEA. Br J Cancer 1986; 54: Accepted for publication 24 November 1999 START: State of the Art Oncology in Europe START project is an evidence based knowledge instrument on state-of-the-art treatment of human malignant tumour and other important clinical topics. Its construction and updating involves almost 200 leading European oncologists and is coordinated by a task force of oncologists based at the Istituto Nazionale dei Tumori and at the European Institute of Oncology in Milan. At present, the following chapters are accessible at our Web site Breast Cancer, Cancer of the Anal Region, Clinical Decision Making, Ewing s Sarcoma, Oesophageal Cancer, Plasma Cell Neoplasms, Small Cell Lung Cancer, Non Small Cell Lung Cancer, Soft Tissue Sarcomas, Osteosarcoma, Antiemetic Therapy, Histiocytic Non Malignant Syndromes. Please note that the START database is still under construction, so we invite you to visit our site, as new chapters are constantly being loaded onto the Internet. Continuous feedback opportunities are offered via to allow communication between the START editing team and the European oncologists who consult the database. Agreements with scientific groups and societies etc. will be allowed to keep the database under continuous, systematic evaluation by expert oncologists in all European countries. If you want to participate commenting, with productive criticism and proposals regarding statements included in any chapter of START, please communicate with the START Editorial Board at the address: start@icil64.cilea.it. The START project has been made possible by unrestricted financial support from Schering-Plough Italy and Compagnia di San Paolo Turin (Italy), who have unconditionally supported the launch and development of this initiative, without any involvement in the preparation or review process. The START project has also been supported by the Europe Against Cancer programme of the Commission of the European Communities and by the American-Italian Cancer Foundation (AICF).