- Email: [email protected]
Comparison of three immunoassays for CA 19-9
Clinica Chimica Acta 400 (2009) 123–127
Contents lists available at ScienceDirect
Clinica Chimica Acta j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / c l i n c h i m
Comparison of three immunoassays for CA 19-9 Kristina Hotakainen a,c,⁎, Pirjo Tanner a, Henrik Alfthan a, Caj Haglund b, Ulf-Håkan Stenman a,c a b c
Department of Clinical Chemistry, Helsinki University Central Hospital, Haartmaninkatu 2, FIN-00290, Helsinki, Finland Department of Surgery, Helsinki University Central Hospital, Haartmaninkatu 2, FIN-00290, Helsinki, Finland Department of Clinical Chemistry, Helsinki University, Haartmaninkatu 8, FIN-00290, Helsinki, Finland
a r t i c l e
i n f o
Article history: Received 17 September 2008 Received in revised form 15 October 2008 Accepted 27 October 2008 Available online 13 November 2008 Keywords: Tumor markers Carbohydrate antigen Pancreas ROC curve
a b s t r a c t Background: Carbohydrate antigen (CA) 19-9 is a valuable tumor marker for gastrointestinal (GI) cancers, but results obtained by different assays vary and a change of method during follow-up may cause problems. Methods: We determined CA 19-9 using Abbott i2000 Architect (CA 19-9XR assay), Roche Elecsys 2010 and Bayer Immuno 1 analyzers. Between-method correlation was studied in 610 serum samples from patients with GI disease. Clinical performance was compared using samples from 68 patients with benign GI diseases and 106 cancer patients. Reference intervals for the Abbott and Roche methods were determined (n = 315). Results: The Architect assay measured significantly lower concentrations than the other assays in controls and benign disease (p b 0.0001), and the upper reference limit was lower (26 kU/l) than for Elecsys (36 kU/l). The area under curve (AUC) for discrimination between benign disease and pancreatic cancer was significantly larger (0.90; 95% CI 0.83–0.98) than for Elecsys (0.78) and Immuno 1 (0.76). However, for all assays a cut-off of 37 kU/l provided optimal separation between benign and malignant disease. Conclusions: The Architect CA 19-9XR assay measures lower concentrations than the other two assays in benign conditions. Thus it provides the best discrimination between benign and malignant GI disease. © 2008 Elsevier B.V. All rights reserved.
1. Introduction The CA 19-9 tumor marker test is based on an antibody (1116-NS19-9) [1] recognizing an epitope occurring in several copies on highmolecular weight mucins expressed by several cancers [2]. CA 19-9 is most valuable as a serum marker for pancreatic and biliary cancer, but increased concentrations occur in several other GI malignancies, e.g., gastric, colorectal, liver cancer and also in breast, lung, and gynaecological cancers [3–9]. However, elevated levels may also occur in benign diseases [10–14]. Serum CA 19-9 concentrations are elevated in 70–90% of patients with pancreatic cancer, the concentrations reflect tumor burden and high concentrations are associated with adverse outcome [15–20]. Post-therapeutic monitoring of marker levels provides information on treatment response and recurrence [21–23]. A cut-off of 37 kU/l was established in the first clinical study on CA 19-9, which was performed with an immunoradiometric assay [24]. Presently, automatic methods are mainly used, but most manufacturers still recommend the same cut-off even if the calibration of different methods varies. Because of this, monitoring of CA 19-9 levels
Abbreviations: CA, carbohydrate antigen; GI, gastrointestinal; AUC, area under curve; ROC curve, receiver operating characteristics curve. ⁎ Corresponding author. Helsinki University Central Hospital/Biomedicum Helsinki, Room A424b, Haartmaninkatu 8, P. O. B. 700, FIN-00029. Tel.: +358 9 4717 1734; fax: +358 9 4717 1731. E-mail address: kristina.hotakainen@hus.fi (K. Hotakainen). 0009-8981/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.cca.2008.10.033
is complicated if different methods are used, and a change of methods often causes changes in marker levels that are not related to the disease [25]. When the Immuno 1 assay was discontinued we determined reference values for CA 19-9 using two alternative methods, Abbott i2000 Architect (CA 19-9XR assay) and Roche Elecsys 2010. In order to compare the clinical validity of the three assays we also studied the CA 19-9 values obtained with each of the methods in various clinical conditions. 2. Materials and methods 2.1. Patients and samples The correlation between the methods was studied by analyzing 610 serum samples requested by clinicians from patients under examination or follow-up for GI diseases at Helsinki University Central Hospital during a three-month period. Clinical and laboratory data, including the concentrations of alkaline phosphatase (ALP), alanine aminotransferase (ALT), aspartate aminotransferase (AST) and bilirubin were retrieved by chart review. Among 185 patients diagnosed with cancer, 106 with active disease were included in the study on clinical validity. Of these, 30 had pancreatic cancer, 43 colorectal and 33 patients other cancers (liver, thyroid, ovarian and breast cancers, one melanoma, one lymphoma and a neuroendocrine tumor). The 68 benign cases included various conditions in which CA 19-9 was requested to test for an occult malignancy but examinations revealed benign diseases; biliary and alcoholic cirrhoses, sclerosing cholangitis, biliary stones, rheumatic and respiratory diseases, inflammations and infections. For determination of reference values, blood samples from apparently healthy individuals (290 males and 222 females) aged 18–70 years were collected in Venoject EDTA-tubes (Terumo), plasma was separated immediately and stored at −20 °C for up to 1 month before assay. The study was performed in accordance with the Helsinki declaration and approved by the institutional review board.
124
K. Hotakainen et al. / Clinica Chimica Acta 400 (2009) 123–127
Table 1 Distribution of CA 19-9 concentrations (kU/l) determined with Architect CA 19-9XR and Elecsys assays in 315 apparently healthy individuals Architect
Elecsys
Mean Median Range
6.2 3.7 2–76.4
11.0 8.8 0.6–70.4
Percentiles 95th 97.5th 99th
17.5 25.9 51.0
29.5 36.4 53.8
2.2. Assays CA 19-9 was determined according to the manufacturers' instructions on Abbott Architect i2000 (CA 19-9XR assay), Elecsys 2010, and Immuno 1 analyzers, with functional detection limits of 2 kU/l, 3 kU/l and 5 kU/l, respectively. Samples were analyzed within 6 h with the Architect and Immuno 1 assays, and after one freeze–thaw cycle with the Elecsys assay. Reagents for the Architect were kindly provided by Abbott Diagnostics (Espoo, Finland). Reagents for the Immuno 1 method were no more available when reference values were determined. The recommended cut-off levels are 37 kU/l for the Architect and Immuno 1 assays, and 34 kU/l for Elecsys. 2.3. Effect of sample type, collection and storage The effect of sample type, collection vessel and freeze–thaw cycles was analyzed by determining CA 19-9 by the Architect method in plasma samples collected in Venoject EDTA-tubes (n = 20), plain serum tubes (Becton Dickinson VT-100SPZ; n = 30) and in serum gel separator tubes (Becton Dickinson Vacutainer SST™ II Advance; n = 30). Serum samples were analyzed within 6 h, after storage for 3 days at +4 °C (n = 30, range 5–715 kU/l), and at −20 °C for 3 weeks, 3 months (n = 22, range 7–928 kU/l) and 1 year (n = 21, range 5–3393 kU/l). Stability in plasma samples was not studied, since serum samples were used for routine analyses. 2.4. Statistical analyses The upper reference limit was calculated on the basis of the 97.5th percentile for healthy controls. Correlation between methods was analyzed using Passing–Bablok regression [26]. Sensitivity and specificity of the assays were compared using receiver operating characteristics (ROC) curve analysis using patients with benign GI diseases as the reference group. CA 19-9 levels before and after storage were compared using the Wilcoxon signed ranks test, and the correlation between the CA 19-9 concentration and ALP, ALT, AST and bilirubin was analyzed using Kendall's tau b test. All analyses were performed using SPSS statistical software (version 12.0.1; Chicago, IL).
3. Results 3.1. Effect of sample type, collection and storage The results for serum samples collected in plain tubes were not significantly different from those in gel tubes or plasma. Storage at +4 °C for 3 days, or at −20 °C for 3 weeks and 3 months had no significant effect on the result from serum samples (−3.4% after 72 h; +9.3% after 3 weeks, and +2.1% after 3 months). However, after 1 year of storage at −20 °C the mean concentration decreased by 20.6% (p b 0.0001; Wilcoxon signed ranks test).
Table 2 Age- and gender specific distribution of CA 19-9 concentrations (kU/l) in apparently healthy individuals determined with the Architect CA 19-9XR assay Females
Males
All
Age (years)
18–50
51–70
All females
18–50
51–70
All males
18–70
n Mean Median Range 97.5th percentile
102 7.9 5.1a 2–76.4 61.2
120 6.6 3.4 2–35 27.5
222 7.2 3.8 2–76.4 31.8
98 4.4 3.0b 2–24 21.4
192 6.7 4.3 2–39 24.1
190 5.9 3.6 2–39 23
512 6.5 3.7 2–76.4 25.8
The difference between all men and women and women of different age groups was not significant. a Mann–Whitney p = 0.002 against males aged 18–50 years. b p b 0.0001 against males aged 51–70 years.
Fig. 1. Correlation between (A), Architect CA 19-9XR and Elecsys; (B), Architect and Immuno 1 and (C), Immuno 1 and Elecsys assays in patients under examinations or followup for GI diseases (n = 610) analyzed by Passing–Bablok regression (solid line); dotted line, x =y identity line. (A), Elecsys= 0.83 Architect + 7.0 kU/l (95% CI 0.75–0.92 for slope and 6.51– 7.98 for y-intercept, respectively); (B), Immuno 1 =0.77 Architect +3.46 kU/l (95% CI 0.67– 0.84 for slope and 3.31–3.67 for y-intercept, respectively); (C), Elecsys= 0.76 Immuno 1 + 6.19 kU/l (95% CI 0.0.74–0.79 for slope and 5.47–6.32 for y-intercept, respectively).
K. Hotakainen et al. / Clinica Chimica Acta 400 (2009) 123–127 Table 3 Diagnostic concordance of Architect (Arc), Immuno 1 (Imm) and Elecsys assays using various cut-offs in benign diseases, pancreatic cancer and colorectal cancer
125
did not. The AUC values were also higher for the Architect (AUC = 0.67) than for Immuno-1 (0.63) and Elecsys (0.65) in the group of other cancers (Fig. 2 and Table 4).
Concordance Benign GI-diseases Arc 37 kU/l vs. Imm 37 kU/l Arc 37 kU/l vs. Elecsys 34 kU/l Imm 37 kU/l vs. Elecsys 34 kU/l Arc 26 kU/la vs. Imm 22 kU/lb Arc 26 kU/la vs. Elecsys 36 kU/la Imm 22 kU/lb vs. Elecsys 36 kU/la Pancreatic cancer Arc 37 kU/l vs. Imm 37 kU/l Arc 37 kU/l vs. Elecsys 34 kU/l Imm 37 kU/l vs. Elecsys 34 kU/l Arc 26 kU/la vs. Imm 22 kU/lb Arc 26 kU/la vs. Elecsys 36 kU/la Imm 22 kU/lb vs. Elecsys 36 kU/la Colorectal cancer Arc 37 kU/l vs. Imm 37 kU/l Arc 37 kU/l vs. Elecsys 34 kU/l Imm 37 kU/l vs. Elecsys 34 kU/l Arc 26 kU/la vs. Imm 22 kU/lb Arc 26 kU/la vs. Elecsys 36 kU/la Imm 22 kU/lb vs. Elecsys 36 kU/la
n
%
35/68 39/68 56/68 38/68 47/68 55/68
51 57 82 56 69 81
27/30 25/30 28/30 36/30 26/30 27/30
90 83 93 87 87 90
30/43 30/43 34/43 31/43 31/43 37/43
70 70 79 72 72 86
Cut-off recommended by the manufacturers of Architect and Immuno 1 assays is 37 kU/l, and that for Elecsys is 34 kU/l. a Upper reference limits (97.5th percentile) determined in the present study. b 97.5th percentile given by the manufacturer of Immuno 1.
3.2. Reference values The upper reference limit in 315 apparently healthy individuals based on the 97.5th percentile was 25.9 kU/l with the Architect assay and 36.4 kU/l with Elecsys (Table 1). The concentrations in females tended to be higher than in males but the difference was not significant. To further study the effect of age and gender, 197 additional samples were analyzed with the Architect assay. Inclusion of the additional samples did not change the age- and gender specific distribution or reference values for the Architect (Table 2). In males, the CA 19-9 concentration tended to increase with age, whereas it decreased in females: the concentrations in females below 50 years of age were significantly higher than in males of the same age group (p = 0.002) (Table 2). 3.3. Between-method correlation and concordance Overall correlation between the methods was satisfactory but at low and moderately elevated concentrations typical of controls and patients with benign disease, the correlation was poor (Fig. 1). The diagnostic concordance between the Architect CA 19-9XR and the other methods was low (51–57%) in benign controls and higher (83– 90%) in pancreatic cancer using the cut-offs given by the manufacturers. In colorectal cancer, the concordance ranged from 70 to 86%. In all conditions, the concordance was similar when different cut-offs in the range of 22–37 kU/l were used and better between Immuno 1 and Elecsys than between these and Architect (Table 3). 3.4. Clinical validity With the Architect assay, the AUC value for differentiation between pancreatic cancer and benign GI disease (AUC = 0.90; 95% CI 0.83–0.98) was significantly higher than for Immuno-1 (AUC = 0.76, 95% CI 0.65– 0.87, p b 0.0002) and Elecsys (AUC = 0.78; 95% CI 0.68–0.89, p = 0.0012). In colorectal cancer, the AUC values were lower, but the AUC for Architect (0.71) was significantly larger than those of Immuno 1 and Elecsys (0.54 and 0.56, respectively), and provided significant discrimination from benign disease (p b 0.0001), while the other methods
Fig. 2. ROC analysis of CA 19-9 methods for differentiation between benign GI disease and (A), pancreatic cancer (B), colorectal cancer and (C), other cancers.
126
K. Hotakainen et al. / Clinica Chimica Acta 400 (2009) 123–127
Table 4 Distribution of CA 19-9 values in benign GI diseases and various cancers using the Architect CA 19-9XR (Arc), Immuno 1 (Imm) and Elecsys (E) assays Diagnosis
n
Method
Median (range) kU/l
IQRa (kU/l)
AUCb
95% CIc
P
Sensitivity at 95% specificity (%)
Benign GI disease
68
Pancreatic cancer
30
Colorectal cancer
43
Other cancers
33
Arc Imm E Arc Imm E Arc Imm E Arc Imm E
16 (2–210) 59 (5–506) 44 (3–1000) 194 (5.3–306,000) 236 (19–60,500) 134 (6–12,700) 49 (2–6060) 57 (5–2350) 52 (3–2480) 26 (2–1170) 83 (5–7500) 67 (5–2690)
6–31 18–123 20–85 49–3685 43–1220 53–834 16–140 25–187 28–115 15–150 47–198 38–134
0.90d 0.76 0.78 0.71e 0.54 0.56 0.67 0.63 0.65
0.83–0.98 0.65–0.87 0.68–0.89 0.61–0.82 0.43–0.65 0.45–0.68 0.56–0.79 0.51–0.74 0.53–0.76
b0.0001 b0.0001 b0.0001 b0.0001 0.47 0.26 0.005 0.04 0.02
60 43 47 35 9 23 27 15 21
The AUC values denote differentiation between malignant and benign GI diseases. a IQR, interquartile range 25th to 75th percentile. b AUC for differentiation between cancer and benign GI-disease. c CI, confidence interval of the AUC. d p = 0.0002 against Immuno 1 and p = 0.0012 against Elecsys. e p = 0.002 against Immuno 1.
Based on the ROC analysis, a cut-off value of 37 kU/l for Architect gave the best combination of sensitivity (87%) and specificity (80%) for pancreatic cancer. With the same cut-off for Immuno 1 the corresponding values were 83% and 37%, and 90% and 43% for Elecsys, respectively. With cut-off values based on the reference limit (26 kU/l) for Architect the sensitivity increased to 90% but specificity decreased to 65%. To investigate how liver diseases affect the results for CA 19-9 obtained with the various assays, we studied the correlation with markers of liver function. In benign diseases, CA 19-9 determined with Immuno 1 (n = 58–73) was significantly associated with ALP (p = 0.008, r = 0.21), bilirubin (p b 0.0001, r = 0.32), AST (p = 0.002, r = 0.29) and ALT (p b 0.0001; r = 0.31), and those of Elecsys with ALT (n = 43, p = 0.039, r = 0.22), while the correlations with the Architect assay were not significant (p N 0.05, r b 0.1). 4. Discussion This study shows that the three CA 19-9 assays give quite variable results especially at low and moderately elevated concentrations. Results obtained with the Architect CA 19-9XR assay are significantly lower than those with the other assays in patients with benign conditions (Table 4) and lower than with the Elecsys in apparently healthy controls (Table 1), while the concentrations in cancer patients are more similar with all assays studied. Thus, the upper reference limit for the Architect assay based on the 97.5th percentile was lower, 26 kU/l, than the traditional cut-off of 37 kU/l, while the corresponding limit for the Elecsys assay was 36.4 kU/l. The upper reference limit could not be determined for Immuno 1; the manufacturer reports an upper reference limit of 22 kU/l but recommends a cut-off of 37 kU/l. However, based on the correlation (Fig. 1), the calibration of the Immuno 1 assay appears to be more similar to that of the Elecsys than the Architect assay. Our upper reference limit for the Architect assay is virtually identical to that determined by La'Ulu and Roberts [27], 26.4 kU/l. The corresponding value for Elecsys determined in our study (36.4 kU/l) was close to that determined by the manufacturer (34 kU/l). Vestergaard et al. determined the CA 19-9 concentrations in 497 healthy blood donors using the Immulite Automated Analyzer (Diagnostic Products), and reported a median CA 19-9 concentration of 4.1 kU/l and a 97.5th percentile of 28.7 kU/l [28]. Interestingly, Del Villano et al. [24] determined the CA 19-9 values by the original immunoradiometric assay in 1020 healthy blood donors, and found levels above 37 kU/l in only 0.6% of these. They also measured nearly 2-fold higher concentrations in females younger than 30 years of age as compared to men of the same age. This is in accordance with our
results; the median CA 19-9 concentration in apparently healthy females between 18 and 30 years was 4.9 kU/l vs. 2.2 kU/l in males (p = 0.003). However, we found no significant overall difference between the genders. This justifies the use of common reference limits for males and females. In ROC analysis, the Architect gave higher AUC values than the other assays and the difference was significant for pancreatic and colon cancer. The best discrimination between malignant and benign GI disease was obtained with a cut-off of 37 kU/l. The same decision limit could also be used for the two other methods. The AUC values for discrimination between pancreatic cancer and healthy controls were very high for the Architect and Elecsys assays (0.98 and 0.97, respectively, not shown), and for this purpose the optimal cut-off corresponded to the upper reference limit, 26 kU/l, for the Architect and 37 kU/l for Elecsys assay. Since we studied a cross-sectional sample of patients under examination and follow-up for GI diseases rather than a screening population the AUC values reflect the utility of the assays for this purpose, but they do not indicate the validity for detecting cancer in a screening setting. Various benign GI diseases causing obstructive jaundice [13] and impaired hepatic function tend to cause falsely elevated tumor marker levels. Elevated CA 19-9 concentrations were more often observed in benign hepatobiliary diseases (with or without obstructive jaundice) with the Elecsys and Immuno 1 than with the Architect assay. With the Architect assay the results were not significantly correlated with liver function tests, which they were with the other two assays. Elevated CA 19-9 concentrations have also been reported in benign respiratory diseases [11,14,29] and we observed increased concentrations (up to 210 kU/l) with all assays in patients with severe chronic obstructive pulmonary disease (n = 3) and three pneumonia leading to multi-organ failure. Because all the assays studied use the same 1116-NS-19-9 antibody, the between-assay variation is obviously caused by differences in assay design and calibration. However, the differences in clinical specificity are probably not explained by calibration alone. The Architect CA 19-9XR assay uses two-step incubation, while Elecsys and Immuno 1 assays are single-step assays. Furthermore the labels and solid phases are different. Mucins of variable size expressing the CA 19-9 epitope occur in different diseases [30,31] and it is possible that detection of the various forms of CA 19-9 is dependent on assay design. The sialylated lacto-N-fucopentaose II-epitope defined by the 1116-NS-19-9 antibody is related to the Lewis A blood group determinant, and Lewis-negative persons (5–9% of a Caucasian population) cannot express the antigen. The upper reference limit of CA 19-9 varies six-fold (7.9–50.6 kU/l) according to the difference in blood group
K. Hotakainen et al. / Clinica Chimica Acta 400 (2009) 123–127
genotype [28]. It is possible, that CA 19-9-negative pancreatic cancer patients represent Lewis negative cases. However, Lewis blood group type is not routinely considered in the interpretation of CA 19-9 levels in our hospital, and we did not have access to this information. In conclusion, the assays studied show large variation in results for individual patients and thus new baseline concentrations need to be defined when changing assay method during patient follow-up. The Architect assay gave the lowest results in benign conditions and similar results as the other two assays in cancer patients, and thus it provides the best differentiation between benign and malignant disease. Acknowledgments The study was funded, in part, by the Stockmann foundation, the Finska Läkaresällskapet and the Helsinki University Research Funds. We thank Abbott Laboratories, Diagnostics Division, for providing reagents. References [1] Koprowski H, Steplewski Z, Mitchell K, Herlyn M, Herlyn D, Fuhrer P. Colorectal carcinoma antigens detected by hybridoma antibodies. Somatic Cell Genet 1979;5: 957–71. [2] Magnani JL, Steplewski Z, Koprowski H, Ginsburg V. Identification of the gastrointestinal and pancreatic cancer-associated antigen detected by monoclonal antibody 19-9 in the sera of patients as a mucin. Cancer Res 1983;43:5489–92. [3] Victorzon M, Haglund C, Lundin J, Roberts P. A prognostic value of CA 19-9 but not of CEA in patients with gastric cancer. Eur J Surg Oncol 1995;21:379–84. [4] Berthiot G, Marechal F, Cattan A, Deltour G. Serum levels of CA-50, CA-19.9, CA125, neuron specific enolase and carcinoembryonic antigen in lung cancer and benign diseases of the lung. Biomed Pharmacother 1989;43:613–20. [5] Yasumoto K, Takahashi Y, Mai M, Kawashima A. Meningeal carcinomatosis preceded by a rapid increase in serum CA19-9 levels in a patient with breast cancer. Int J Clin Oncol 2005;10:276–80. [6] Papantoniou V, Tsiouris S, Koutsikos J, Ptohis N, Lazaris D, Zerva C. Increased serum carbohydrate antigen 19-9 in relapsed ductal breast carcinoma. Hell J Nucl Med 2006;9:36–8. [7] Molina R, Ojeda B, Filella X, et al. A prospective study of tumor markers CA 125 and CA 19.9 in patients with epithelial ovarian carcinomas. Tumour Biol 1992;13:278–86. [8] Borras G, Molina R, Xercavins J, Ballesta A, Iglesias J. Tumor antigens CA 19.9, CA 125, and CEA in carcinoma of the uterine cervix. Gynecol Oncol 1995;57:205–11. [9] Dietel M, Arps H, Klapdor R, Muller-Hagen S, Sieck M, Hoffmann L. Antigen detection by the monoclonal antibodies CA 19-9 and CA 125 in normal and tumor tissue and patients' sera. J Cancer Res Clin Oncol 1986;111:257–65. [10] Aybek H, Aybek Z, Sinik Z, Demir S, Sancak B, Tuncay L. Elevation of serum and urinary carbohydrate antigen 19-9 in benign hydronephrosis. Int J Urol 2006;13: 1380–4. [11] Marechal F, Berthiot G, Deltour G. Serum levels of CA-50, CA-19.9, CA-125, CA-15.3, enolase and carcino-embryonic antigen in non neoplastic diseases of the lung. Anticancer Res 1988;8:677–80.
127
[12] Tsavaris N, Mavragani CP, Pikazios D. Rheumatoid arthritis: correlation between rheumatoid factor levels and CA-125 tumour marker elevation. Ann Rheum Dis 2007;66:980. [13] Duraker N, Hot S, Polat Y, Hobek A, Gencler N, Urhan N. CEA, CA 19-9, and CA 125 in the differential diagnosis of benign and malignant pancreatic diseases with or without jaundice. J Surg Oncol 2007;95:142–7. [14] Kodama T, Satoh H, Ishikawa H, Ohtsuka M. Serum levels of CA19-9 in patients with nonmalignant respiratory diseases. J Clin Lab Anal 2007;21:103–6. [15] Steinberg W. The clinical utility of the CA 19-9 tumor-associated antigen. Am J Gastroenterol 1990;85:350–5. [16] Glenn J, Steinberg W, Kurtzman S, Steinberg S, Sindelar W. Evaluation of the utility of a radioimmunoassay for serum CA 19-9 levels in patients before and after treatment of carcinoma of the pancreas. J Clin Oncol 1988;6:462–8. [17] Lundin J, Roberts P, Kuusela P, Haglund C. The prognostic value of preoperative serum levels of CA 19-9 and CEA in patients with pancreatic cancer. Br J Cancer 1994;69:515–9. [18] Boeck S, Stieber P, Holdenrieder S, Wilkowski R, Heinemann V. Prognostic and therapeutic significance of carbohydrate antigen 19-9 as tumor marker in patients with pancreatic cancer. Oncology 2006;70:255–64. [19] Safi F, Schlosser W, Kolb G, Beger H. Diagnostic value of CA 19-9 in patients with pancreatic cancer and nonspecific gastrointestinal symptoms. J Gastrointest Surg 1997;1:106–12. [20] Maisey N, Norman A, Hill A, Massey A, Oates J, Cunningham D. CA19-9 as a prognostic factor in inoperable pancreatic cancer: the implication for clinical trials. Br J Cancer 2005;93:740–3. [21] Stemmler J, Stieber P, Szymala A, et al. Are serial CA 19-9 kinetics helpful in predicting survival in patients with advanced or metastatic pancreatic cancer treated with gemcitabine and cisplatin? Onkologie 2003;26:462–7. [22] Micke O, Bruns F, Schafer U, Kurowski R, Horst E, Willich N. CA 19-9 in the therapy monitoring and follow-up of locally advanced cancer of the exocrine pancreas treated with radiochemotherapy. Anticancer Res 2003;23:835–40. [23] Safi F, Schlosser W, Falkenreck S, Beger H. Prognostic value of CA 19-9 serum course in pancreatic cancer. Hepatogastroenterology 1998;45:253–9. [24] Del Villano B, Brennan S, Brock P, et al. Radioimmunometric assay for a monoclonal antibody-defined tumor marker, CA 19-9. Clin Chem 1983;29:549–52. [25] Passerini R, Riggio D, Salvatici M, Zorzino L, Radice D, Sandri M. Interchangeability of measurements of CA 19-9 in serum with four frequently used assays: an update. Clin Chem Lab Med 2007;45:100–4. [26] Passing H, Bablok. A new biometrical procedure for testing the equality of measurements from two different analytical methods. Application of linear regression procedures for method comparison studies in clinical chemistry, Part I. J Clin Chem Clin Biochem 1983;21:709–20. [27] La'ulu S, Roberts W. Performance characteristics of five automated CA 19-9 assays. Am J Clin Pathol 2007;127:436–40. [28] Vestergaard EM, Hein HO, Meyer H, et al. Reference values and biological variation for tumor marker CA 19-9 in serum for different Lewis and secretor genotypes and evaluation of secretor and Lewis genotyping in a Caucasian population. Clin Chem 1999;45:54–61. [29] Fontana D, Della Beffa V, Fusca M, Loddo F. Extralobar pulmonary sequestration showing high CA 19-19 levels. Minerva Chir 2007;62:39–42. [30] Wu J, Olson J, Walker K. Tumor markers CA 19-9 and CA 195 are also useful as markers for cystic fibrosis. J Clin Lab Anal 1992;6:151–61. [31] Fujita J, Dobashi N, Ohtsuki Y, et al. Detection of large molecular weight cytokeratin 8 as carrier protein of CA19-9 in non-small-cell lung cancer cell lines. Br J Cancer 1999;81:769–73.
Contents lists available at ScienceDirect
Clinica Chimica Acta j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / c l i n c h i m
Comparison of three immunoassays for CA 19-9 Kristina Hotakainen a,c,⁎, Pirjo Tanner a, Henrik Alfthan a, Caj Haglund b, Ulf-Håkan Stenman a,c a b c
Department of Clinical Chemistry, Helsinki University Central Hospital, Haartmaninkatu 2, FIN-00290, Helsinki, Finland Department of Surgery, Helsinki University Central Hospital, Haartmaninkatu 2, FIN-00290, Helsinki, Finland Department of Clinical Chemistry, Helsinki University, Haartmaninkatu 8, FIN-00290, Helsinki, Finland
a r t i c l e
i n f o
Article history: Received 17 September 2008 Received in revised form 15 October 2008 Accepted 27 October 2008 Available online 13 November 2008 Keywords: Tumor markers Carbohydrate antigen Pancreas ROC curve
a b s t r a c t Background: Carbohydrate antigen (CA) 19-9 is a valuable tumor marker for gastrointestinal (GI) cancers, but results obtained by different assays vary and a change of method during follow-up may cause problems. Methods: We determined CA 19-9 using Abbott i2000 Architect (CA 19-9XR assay), Roche Elecsys 2010 and Bayer Immuno 1 analyzers. Between-method correlation was studied in 610 serum samples from patients with GI disease. Clinical performance was compared using samples from 68 patients with benign GI diseases and 106 cancer patients. Reference intervals for the Abbott and Roche methods were determined (n = 315). Results: The Architect assay measured significantly lower concentrations than the other assays in controls and benign disease (p b 0.0001), and the upper reference limit was lower (26 kU/l) than for Elecsys (36 kU/l). The area under curve (AUC) for discrimination between benign disease and pancreatic cancer was significantly larger (0.90; 95% CI 0.83–0.98) than for Elecsys (0.78) and Immuno 1 (0.76). However, for all assays a cut-off of 37 kU/l provided optimal separation between benign and malignant disease. Conclusions: The Architect CA 19-9XR assay measures lower concentrations than the other two assays in benign conditions. Thus it provides the best discrimination between benign and malignant GI disease. © 2008 Elsevier B.V. All rights reserved.
1. Introduction The CA 19-9 tumor marker test is based on an antibody (1116-NS19-9) [1] recognizing an epitope occurring in several copies on highmolecular weight mucins expressed by several cancers [2]. CA 19-9 is most valuable as a serum marker for pancreatic and biliary cancer, but increased concentrations occur in several other GI malignancies, e.g., gastric, colorectal, liver cancer and also in breast, lung, and gynaecological cancers [3–9]. However, elevated levels may also occur in benign diseases [10–14]. Serum CA 19-9 concentrations are elevated in 70–90% of patients with pancreatic cancer, the concentrations reflect tumor burden and high concentrations are associated with adverse outcome [15–20]. Post-therapeutic monitoring of marker levels provides information on treatment response and recurrence [21–23]. A cut-off of 37 kU/l was established in the first clinical study on CA 19-9, which was performed with an immunoradiometric assay [24]. Presently, automatic methods are mainly used, but most manufacturers still recommend the same cut-off even if the calibration of different methods varies. Because of this, monitoring of CA 19-9 levels
Abbreviations: CA, carbohydrate antigen; GI, gastrointestinal; AUC, area under curve; ROC curve, receiver operating characteristics curve. ⁎ Corresponding author. Helsinki University Central Hospital/Biomedicum Helsinki, Room A424b, Haartmaninkatu 8, P. O. B. 700, FIN-00029. Tel.: +358 9 4717 1734; fax: +358 9 4717 1731. E-mail address: kristina.hotakainen@hus.fi (K. Hotakainen). 0009-8981/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.cca.2008.10.033
is complicated if different methods are used, and a change of methods often causes changes in marker levels that are not related to the disease [25]. When the Immuno 1 assay was discontinued we determined reference values for CA 19-9 using two alternative methods, Abbott i2000 Architect (CA 19-9XR assay) and Roche Elecsys 2010. In order to compare the clinical validity of the three assays we also studied the CA 19-9 values obtained with each of the methods in various clinical conditions. 2. Materials and methods 2.1. Patients and samples The correlation between the methods was studied by analyzing 610 serum samples requested by clinicians from patients under examination or follow-up for GI diseases at Helsinki University Central Hospital during a three-month period. Clinical and laboratory data, including the concentrations of alkaline phosphatase (ALP), alanine aminotransferase (ALT), aspartate aminotransferase (AST) and bilirubin were retrieved by chart review. Among 185 patients diagnosed with cancer, 106 with active disease were included in the study on clinical validity. Of these, 30 had pancreatic cancer, 43 colorectal and 33 patients other cancers (liver, thyroid, ovarian and breast cancers, one melanoma, one lymphoma and a neuroendocrine tumor). The 68 benign cases included various conditions in which CA 19-9 was requested to test for an occult malignancy but examinations revealed benign diseases; biliary and alcoholic cirrhoses, sclerosing cholangitis, biliary stones, rheumatic and respiratory diseases, inflammations and infections. For determination of reference values, blood samples from apparently healthy individuals (290 males and 222 females) aged 18–70 years were collected in Venoject EDTA-tubes (Terumo), plasma was separated immediately and stored at −20 °C for up to 1 month before assay. The study was performed in accordance with the Helsinki declaration and approved by the institutional review board.
124
K. Hotakainen et al. / Clinica Chimica Acta 400 (2009) 123–127
Table 1 Distribution of CA 19-9 concentrations (kU/l) determined with Architect CA 19-9XR and Elecsys assays in 315 apparently healthy individuals Architect
Elecsys
Mean Median Range
6.2 3.7 2–76.4
11.0 8.8 0.6–70.4
Percentiles 95th 97.5th 99th
17.5 25.9 51.0
29.5 36.4 53.8
2.2. Assays CA 19-9 was determined according to the manufacturers' instructions on Abbott Architect i2000 (CA 19-9XR assay), Elecsys 2010, and Immuno 1 analyzers, with functional detection limits of 2 kU/l, 3 kU/l and 5 kU/l, respectively. Samples were analyzed within 6 h with the Architect and Immuno 1 assays, and after one freeze–thaw cycle with the Elecsys assay. Reagents for the Architect were kindly provided by Abbott Diagnostics (Espoo, Finland). Reagents for the Immuno 1 method were no more available when reference values were determined. The recommended cut-off levels are 37 kU/l for the Architect and Immuno 1 assays, and 34 kU/l for Elecsys. 2.3. Effect of sample type, collection and storage The effect of sample type, collection vessel and freeze–thaw cycles was analyzed by determining CA 19-9 by the Architect method in plasma samples collected in Venoject EDTA-tubes (n = 20), plain serum tubes (Becton Dickinson VT-100SPZ; n = 30) and in serum gel separator tubes (Becton Dickinson Vacutainer SST™ II Advance; n = 30). Serum samples were analyzed within 6 h, after storage for 3 days at +4 °C (n = 30, range 5–715 kU/l), and at −20 °C for 3 weeks, 3 months (n = 22, range 7–928 kU/l) and 1 year (n = 21, range 5–3393 kU/l). Stability in plasma samples was not studied, since serum samples were used for routine analyses. 2.4. Statistical analyses The upper reference limit was calculated on the basis of the 97.5th percentile for healthy controls. Correlation between methods was analyzed using Passing–Bablok regression [26]. Sensitivity and specificity of the assays were compared using receiver operating characteristics (ROC) curve analysis using patients with benign GI diseases as the reference group. CA 19-9 levels before and after storage were compared using the Wilcoxon signed ranks test, and the correlation between the CA 19-9 concentration and ALP, ALT, AST and bilirubin was analyzed using Kendall's tau b test. All analyses were performed using SPSS statistical software (version 12.0.1; Chicago, IL).
3. Results 3.1. Effect of sample type, collection and storage The results for serum samples collected in plain tubes were not significantly different from those in gel tubes or plasma. Storage at +4 °C for 3 days, or at −20 °C for 3 weeks and 3 months had no significant effect on the result from serum samples (−3.4% after 72 h; +9.3% after 3 weeks, and +2.1% after 3 months). However, after 1 year of storage at −20 °C the mean concentration decreased by 20.6% (p b 0.0001; Wilcoxon signed ranks test).
Table 2 Age- and gender specific distribution of CA 19-9 concentrations (kU/l) in apparently healthy individuals determined with the Architect CA 19-9XR assay Females
Males
All
Age (years)
18–50
51–70
All females
18–50
51–70
All males
18–70
n Mean Median Range 97.5th percentile
102 7.9 5.1a 2–76.4 61.2
120 6.6 3.4 2–35 27.5
222 7.2 3.8 2–76.4 31.8
98 4.4 3.0b 2–24 21.4
192 6.7 4.3 2–39 24.1
190 5.9 3.6 2–39 23
512 6.5 3.7 2–76.4 25.8
The difference between all men and women and women of different age groups was not significant. a Mann–Whitney p = 0.002 against males aged 18–50 years. b p b 0.0001 against males aged 51–70 years.
Fig. 1. Correlation between (A), Architect CA 19-9XR and Elecsys; (B), Architect and Immuno 1 and (C), Immuno 1 and Elecsys assays in patients under examinations or followup for GI diseases (n = 610) analyzed by Passing–Bablok regression (solid line); dotted line, x =y identity line. (A), Elecsys= 0.83 Architect + 7.0 kU/l (95% CI 0.75–0.92 for slope and 6.51– 7.98 for y-intercept, respectively); (B), Immuno 1 =0.77 Architect +3.46 kU/l (95% CI 0.67– 0.84 for slope and 3.31–3.67 for y-intercept, respectively); (C), Elecsys= 0.76 Immuno 1 + 6.19 kU/l (95% CI 0.0.74–0.79 for slope and 5.47–6.32 for y-intercept, respectively).
K. Hotakainen et al. / Clinica Chimica Acta 400 (2009) 123–127 Table 3 Diagnostic concordance of Architect (Arc), Immuno 1 (Imm) and Elecsys assays using various cut-offs in benign diseases, pancreatic cancer and colorectal cancer
125
did not. The AUC values were also higher for the Architect (AUC = 0.67) than for Immuno-1 (0.63) and Elecsys (0.65) in the group of other cancers (Fig. 2 and Table 4).
Concordance Benign GI-diseases Arc 37 kU/l vs. Imm 37 kU/l Arc 37 kU/l vs. Elecsys 34 kU/l Imm 37 kU/l vs. Elecsys 34 kU/l Arc 26 kU/la vs. Imm 22 kU/lb Arc 26 kU/la vs. Elecsys 36 kU/la Imm 22 kU/lb vs. Elecsys 36 kU/la Pancreatic cancer Arc 37 kU/l vs. Imm 37 kU/l Arc 37 kU/l vs. Elecsys 34 kU/l Imm 37 kU/l vs. Elecsys 34 kU/l Arc 26 kU/la vs. Imm 22 kU/lb Arc 26 kU/la vs. Elecsys 36 kU/la Imm 22 kU/lb vs. Elecsys 36 kU/la Colorectal cancer Arc 37 kU/l vs. Imm 37 kU/l Arc 37 kU/l vs. Elecsys 34 kU/l Imm 37 kU/l vs. Elecsys 34 kU/l Arc 26 kU/la vs. Imm 22 kU/lb Arc 26 kU/la vs. Elecsys 36 kU/la Imm 22 kU/lb vs. Elecsys 36 kU/la
n
%
35/68 39/68 56/68 38/68 47/68 55/68
51 57 82 56 69 81
27/30 25/30 28/30 36/30 26/30 27/30
90 83 93 87 87 90
30/43 30/43 34/43 31/43 31/43 37/43
70 70 79 72 72 86
Cut-off recommended by the manufacturers of Architect and Immuno 1 assays is 37 kU/l, and that for Elecsys is 34 kU/l. a Upper reference limits (97.5th percentile) determined in the present study. b 97.5th percentile given by the manufacturer of Immuno 1.
3.2. Reference values The upper reference limit in 315 apparently healthy individuals based on the 97.5th percentile was 25.9 kU/l with the Architect assay and 36.4 kU/l with Elecsys (Table 1). The concentrations in females tended to be higher than in males but the difference was not significant. To further study the effect of age and gender, 197 additional samples were analyzed with the Architect assay. Inclusion of the additional samples did not change the age- and gender specific distribution or reference values for the Architect (Table 2). In males, the CA 19-9 concentration tended to increase with age, whereas it decreased in females: the concentrations in females below 50 years of age were significantly higher than in males of the same age group (p = 0.002) (Table 2). 3.3. Between-method correlation and concordance Overall correlation between the methods was satisfactory but at low and moderately elevated concentrations typical of controls and patients with benign disease, the correlation was poor (Fig. 1). The diagnostic concordance between the Architect CA 19-9XR and the other methods was low (51–57%) in benign controls and higher (83– 90%) in pancreatic cancer using the cut-offs given by the manufacturers. In colorectal cancer, the concordance ranged from 70 to 86%. In all conditions, the concordance was similar when different cut-offs in the range of 22–37 kU/l were used and better between Immuno 1 and Elecsys than between these and Architect (Table 3). 3.4. Clinical validity With the Architect assay, the AUC value for differentiation between pancreatic cancer and benign GI disease (AUC = 0.90; 95% CI 0.83–0.98) was significantly higher than for Immuno-1 (AUC = 0.76, 95% CI 0.65– 0.87, p b 0.0002) and Elecsys (AUC = 0.78; 95% CI 0.68–0.89, p = 0.0012). In colorectal cancer, the AUC values were lower, but the AUC for Architect (0.71) was significantly larger than those of Immuno 1 and Elecsys (0.54 and 0.56, respectively), and provided significant discrimination from benign disease (p b 0.0001), while the other methods
Fig. 2. ROC analysis of CA 19-9 methods for differentiation between benign GI disease and (A), pancreatic cancer (B), colorectal cancer and (C), other cancers.
126
K. Hotakainen et al. / Clinica Chimica Acta 400 (2009) 123–127
Table 4 Distribution of CA 19-9 values in benign GI diseases and various cancers using the Architect CA 19-9XR (Arc), Immuno 1 (Imm) and Elecsys (E) assays Diagnosis
n
Method
Median (range) kU/l
IQRa (kU/l)
AUCb
95% CIc
P
Sensitivity at 95% specificity (%)
Benign GI disease
68
Pancreatic cancer
30
Colorectal cancer
43
Other cancers
33
Arc Imm E Arc Imm E Arc Imm E Arc Imm E
16 (2–210) 59 (5–506) 44 (3–1000) 194 (5.3–306,000) 236 (19–60,500) 134 (6–12,700) 49 (2–6060) 57 (5–2350) 52 (3–2480) 26 (2–1170) 83 (5–7500) 67 (5–2690)
6–31 18–123 20–85 49–3685 43–1220 53–834 16–140 25–187 28–115 15–150 47–198 38–134
0.90d 0.76 0.78 0.71e 0.54 0.56 0.67 0.63 0.65
0.83–0.98 0.65–0.87 0.68–0.89 0.61–0.82 0.43–0.65 0.45–0.68 0.56–0.79 0.51–0.74 0.53–0.76
b0.0001 b0.0001 b0.0001 b0.0001 0.47 0.26 0.005 0.04 0.02
60 43 47 35 9 23 27 15 21
The AUC values denote differentiation between malignant and benign GI diseases. a IQR, interquartile range 25th to 75th percentile. b AUC for differentiation between cancer and benign GI-disease. c CI, confidence interval of the AUC. d p = 0.0002 against Immuno 1 and p = 0.0012 against Elecsys. e p = 0.002 against Immuno 1.
Based on the ROC analysis, a cut-off value of 37 kU/l for Architect gave the best combination of sensitivity (87%) and specificity (80%) for pancreatic cancer. With the same cut-off for Immuno 1 the corresponding values were 83% and 37%, and 90% and 43% for Elecsys, respectively. With cut-off values based on the reference limit (26 kU/l) for Architect the sensitivity increased to 90% but specificity decreased to 65%. To investigate how liver diseases affect the results for CA 19-9 obtained with the various assays, we studied the correlation with markers of liver function. In benign diseases, CA 19-9 determined with Immuno 1 (n = 58–73) was significantly associated with ALP (p = 0.008, r = 0.21), bilirubin (p b 0.0001, r = 0.32), AST (p = 0.002, r = 0.29) and ALT (p b 0.0001; r = 0.31), and those of Elecsys with ALT (n = 43, p = 0.039, r = 0.22), while the correlations with the Architect assay were not significant (p N 0.05, r b 0.1). 4. Discussion This study shows that the three CA 19-9 assays give quite variable results especially at low and moderately elevated concentrations. Results obtained with the Architect CA 19-9XR assay are significantly lower than those with the other assays in patients with benign conditions (Table 4) and lower than with the Elecsys in apparently healthy controls (Table 1), while the concentrations in cancer patients are more similar with all assays studied. Thus, the upper reference limit for the Architect assay based on the 97.5th percentile was lower, 26 kU/l, than the traditional cut-off of 37 kU/l, while the corresponding limit for the Elecsys assay was 36.4 kU/l. The upper reference limit could not be determined for Immuno 1; the manufacturer reports an upper reference limit of 22 kU/l but recommends a cut-off of 37 kU/l. However, based on the correlation (Fig. 1), the calibration of the Immuno 1 assay appears to be more similar to that of the Elecsys than the Architect assay. Our upper reference limit for the Architect assay is virtually identical to that determined by La'Ulu and Roberts [27], 26.4 kU/l. The corresponding value for Elecsys determined in our study (36.4 kU/l) was close to that determined by the manufacturer (34 kU/l). Vestergaard et al. determined the CA 19-9 concentrations in 497 healthy blood donors using the Immulite Automated Analyzer (Diagnostic Products), and reported a median CA 19-9 concentration of 4.1 kU/l and a 97.5th percentile of 28.7 kU/l [28]. Interestingly, Del Villano et al. [24] determined the CA 19-9 values by the original immunoradiometric assay in 1020 healthy blood donors, and found levels above 37 kU/l in only 0.6% of these. They also measured nearly 2-fold higher concentrations in females younger than 30 years of age as compared to men of the same age. This is in accordance with our
results; the median CA 19-9 concentration in apparently healthy females between 18 and 30 years was 4.9 kU/l vs. 2.2 kU/l in males (p = 0.003). However, we found no significant overall difference between the genders. This justifies the use of common reference limits for males and females. In ROC analysis, the Architect gave higher AUC values than the other assays and the difference was significant for pancreatic and colon cancer. The best discrimination between malignant and benign GI disease was obtained with a cut-off of 37 kU/l. The same decision limit could also be used for the two other methods. The AUC values for discrimination between pancreatic cancer and healthy controls were very high for the Architect and Elecsys assays (0.98 and 0.97, respectively, not shown), and for this purpose the optimal cut-off corresponded to the upper reference limit, 26 kU/l, for the Architect and 37 kU/l for Elecsys assay. Since we studied a cross-sectional sample of patients under examination and follow-up for GI diseases rather than a screening population the AUC values reflect the utility of the assays for this purpose, but they do not indicate the validity for detecting cancer in a screening setting. Various benign GI diseases causing obstructive jaundice [13] and impaired hepatic function tend to cause falsely elevated tumor marker levels. Elevated CA 19-9 concentrations were more often observed in benign hepatobiliary diseases (with or without obstructive jaundice) with the Elecsys and Immuno 1 than with the Architect assay. With the Architect assay the results were not significantly correlated with liver function tests, which they were with the other two assays. Elevated CA 19-9 concentrations have also been reported in benign respiratory diseases [11,14,29] and we observed increased concentrations (up to 210 kU/l) with all assays in patients with severe chronic obstructive pulmonary disease (n = 3) and three pneumonia leading to multi-organ failure. Because all the assays studied use the same 1116-NS-19-9 antibody, the between-assay variation is obviously caused by differences in assay design and calibration. However, the differences in clinical specificity are probably not explained by calibration alone. The Architect CA 19-9XR assay uses two-step incubation, while Elecsys and Immuno 1 assays are single-step assays. Furthermore the labels and solid phases are different. Mucins of variable size expressing the CA 19-9 epitope occur in different diseases [30,31] and it is possible that detection of the various forms of CA 19-9 is dependent on assay design. The sialylated lacto-N-fucopentaose II-epitope defined by the 1116-NS-19-9 antibody is related to the Lewis A blood group determinant, and Lewis-negative persons (5–9% of a Caucasian population) cannot express the antigen. The upper reference limit of CA 19-9 varies six-fold (7.9–50.6 kU/l) according to the difference in blood group
K. Hotakainen et al. / Clinica Chimica Acta 400 (2009) 123–127
genotype [28]. It is possible, that CA 19-9-negative pancreatic cancer patients represent Lewis negative cases. However, Lewis blood group type is not routinely considered in the interpretation of CA 19-9 levels in our hospital, and we did not have access to this information. In conclusion, the assays studied show large variation in results for individual patients and thus new baseline concentrations need to be defined when changing assay method during patient follow-up. The Architect assay gave the lowest results in benign conditions and similar results as the other two assays in cancer patients, and thus it provides the best differentiation between benign and malignant disease. Acknowledgments The study was funded, in part, by the Stockmann foundation, the Finska Läkaresällskapet and the Helsinki University Research Funds. We thank Abbott Laboratories, Diagnostics Division, for providing reagents. References [1] Koprowski H, Steplewski Z, Mitchell K, Herlyn M, Herlyn D, Fuhrer P. Colorectal carcinoma antigens detected by hybridoma antibodies. Somatic Cell Genet 1979;5: 957–71. [2] Magnani JL, Steplewski Z, Koprowski H, Ginsburg V. Identification of the gastrointestinal and pancreatic cancer-associated antigen detected by monoclonal antibody 19-9 in the sera of patients as a mucin. Cancer Res 1983;43:5489–92. [3] Victorzon M, Haglund C, Lundin J, Roberts P. A prognostic value of CA 19-9 but not of CEA in patients with gastric cancer. Eur J Surg Oncol 1995;21:379–84. [4] Berthiot G, Marechal F, Cattan A, Deltour G. Serum levels of CA-50, CA-19.9, CA125, neuron specific enolase and carcinoembryonic antigen in lung cancer and benign diseases of the lung. Biomed Pharmacother 1989;43:613–20. [5] Yasumoto K, Takahashi Y, Mai M, Kawashima A. Meningeal carcinomatosis preceded by a rapid increase in serum CA19-9 levels in a patient with breast cancer. Int J Clin Oncol 2005;10:276–80. [6] Papantoniou V, Tsiouris S, Koutsikos J, Ptohis N, Lazaris D, Zerva C. Increased serum carbohydrate antigen 19-9 in relapsed ductal breast carcinoma. Hell J Nucl Med 2006;9:36–8. [7] Molina R, Ojeda B, Filella X, et al. A prospective study of tumor markers CA 125 and CA 19.9 in patients with epithelial ovarian carcinomas. Tumour Biol 1992;13:278–86. [8] Borras G, Molina R, Xercavins J, Ballesta A, Iglesias J. Tumor antigens CA 19.9, CA 125, and CEA in carcinoma of the uterine cervix. Gynecol Oncol 1995;57:205–11. [9] Dietel M, Arps H, Klapdor R, Muller-Hagen S, Sieck M, Hoffmann L. Antigen detection by the monoclonal antibodies CA 19-9 and CA 125 in normal and tumor tissue and patients' sera. J Cancer Res Clin Oncol 1986;111:257–65. [10] Aybek H, Aybek Z, Sinik Z, Demir S, Sancak B, Tuncay L. Elevation of serum and urinary carbohydrate antigen 19-9 in benign hydronephrosis. Int J Urol 2006;13: 1380–4. [11] Marechal F, Berthiot G, Deltour G. Serum levels of CA-50, CA-19.9, CA-125, CA-15.3, enolase and carcino-embryonic antigen in non neoplastic diseases of the lung. Anticancer Res 1988;8:677–80.
127
[12] Tsavaris N, Mavragani CP, Pikazios D. Rheumatoid arthritis: correlation between rheumatoid factor levels and CA-125 tumour marker elevation. Ann Rheum Dis 2007;66:980. [13] Duraker N, Hot S, Polat Y, Hobek A, Gencler N, Urhan N. CEA, CA 19-9, and CA 125 in the differential diagnosis of benign and malignant pancreatic diseases with or without jaundice. J Surg Oncol 2007;95:142–7. [14] Kodama T, Satoh H, Ishikawa H, Ohtsuka M. Serum levels of CA19-9 in patients with nonmalignant respiratory diseases. J Clin Lab Anal 2007;21:103–6. [15] Steinberg W. The clinical utility of the CA 19-9 tumor-associated antigen. Am J Gastroenterol 1990;85:350–5. [16] Glenn J, Steinberg W, Kurtzman S, Steinberg S, Sindelar W. Evaluation of the utility of a radioimmunoassay for serum CA 19-9 levels in patients before and after treatment of carcinoma of the pancreas. J Clin Oncol 1988;6:462–8. [17] Lundin J, Roberts P, Kuusela P, Haglund C. The prognostic value of preoperative serum levels of CA 19-9 and CEA in patients with pancreatic cancer. Br J Cancer 1994;69:515–9. [18] Boeck S, Stieber P, Holdenrieder S, Wilkowski R, Heinemann V. Prognostic and therapeutic significance of carbohydrate antigen 19-9 as tumor marker in patients with pancreatic cancer. Oncology 2006;70:255–64. [19] Safi F, Schlosser W, Kolb G, Beger H. Diagnostic value of CA 19-9 in patients with pancreatic cancer and nonspecific gastrointestinal symptoms. J Gastrointest Surg 1997;1:106–12. [20] Maisey N, Norman A, Hill A, Massey A, Oates J, Cunningham D. CA19-9 as a prognostic factor in inoperable pancreatic cancer: the implication for clinical trials. Br J Cancer 2005;93:740–3. [21] Stemmler J, Stieber P, Szymala A, et al. Are serial CA 19-9 kinetics helpful in predicting survival in patients with advanced or metastatic pancreatic cancer treated with gemcitabine and cisplatin? Onkologie 2003;26:462–7. [22] Micke O, Bruns F, Schafer U, Kurowski R, Horst E, Willich N. CA 19-9 in the therapy monitoring and follow-up of locally advanced cancer of the exocrine pancreas treated with radiochemotherapy. Anticancer Res 2003;23:835–40. [23] Safi F, Schlosser W, Falkenreck S, Beger H. Prognostic value of CA 19-9 serum course in pancreatic cancer. Hepatogastroenterology 1998;45:253–9. [24] Del Villano B, Brennan S, Brock P, et al. Radioimmunometric assay for a monoclonal antibody-defined tumor marker, CA 19-9. Clin Chem 1983;29:549–52. [25] Passerini R, Riggio D, Salvatici M, Zorzino L, Radice D, Sandri M. Interchangeability of measurements of CA 19-9 in serum with four frequently used assays: an update. Clin Chem Lab Med 2007;45:100–4. [26] Passing H, Bablok. A new biometrical procedure for testing the equality of measurements from two different analytical methods. Application of linear regression procedures for method comparison studies in clinical chemistry, Part I. J Clin Chem Clin Biochem 1983;21:709–20. [27] La'ulu S, Roberts W. Performance characteristics of five automated CA 19-9 assays. Am J Clin Pathol 2007;127:436–40. [28] Vestergaard EM, Hein HO, Meyer H, et al. Reference values and biological variation for tumor marker CA 19-9 in serum for different Lewis and secretor genotypes and evaluation of secretor and Lewis genotyping in a Caucasian population. Clin Chem 1999;45:54–61. [29] Fontana D, Della Beffa V, Fusca M, Loddo F. Extralobar pulmonary sequestration showing high CA 19-19 levels. Minerva Chir 2007;62:39–42. [30] Wu J, Olson J, Walker K. Tumor markers CA 19-9 and CA 195 are also useful as markers for cystic fibrosis. J Clin Lab Anal 1992;6:151–61. [31] Fujita J, Dobashi N, Ohtsuki Y, et al. Detection of large molecular weight cytokeratin 8 as carrier protein of CA19-9 in non-small-cell lung cancer cell lines. Br J Cancer 1999;81:769–73.