- Email: [email protected]
Determination of CA 19-9 antigen in serum and pancreatic juice for differential diagnosis of pancreatic adenocarcinoma from chronic pancreatitis
GASTROENTEROLOGY
1987;92:60-7
Determination of CA 19-9Antigen in Serum and Pancreatic Juice for Differential Diagnosis of Pancreatic Adenocarcinoma From Chronic Pancreatitis ALBERT0 MALESCI, MAURIZIO A. TOMMASINI, CLAUDIO BONATO, PAOLA BOCCHIA, MAURIZO BERSANI, ALESSANDRO ZERBI, EDOARDO BERETTA, and VALERIO DI CARLO Istituto Italy
di Medicina
Interna,
Osoedale
Policlinico
Serum CA 19-9 levels were measured in 63 patients with ductal pancreatic adenocarcinoma and in 49 patients with chronic pancreatitis. Concentrations were abnormally high (>40 U/ml) in 57 (90%) patients with cancer and only in 5 (10%) patients with chronic pancreatitis. All patients with falsely normal serum values had poorly differentiated carcinomas. Median CA 19-9 concentrations were progressively higher in patients with more advanced cancer. Fifteen of 16 (93%) patients with localized cancer has abnormal serum levels but only 5 (31%) of them had values >120 U/ml, which was the highest score observed in patients with chronic pancreatitis. Pure pancreatic juice was obtained endoscopically from 23 patients with pancreatic cancer and from 20 with chronic pancreatitis. CA 19-9 concentrations in pancreatic juice were significantly higher in patients with cancer than in nonneoplastic patients. AI1 11 patients with resectable cancer investigated had a ratio of CA 19-9 to secretory protein concentration in pancreatic juice above the range of patients with chronic pancreatitis. We conclude that serum CA 19-9 determination is highly sensitive and specific for the differential diagnosis of pancreatic cancer versus chronic pancreatitis. However, moderately increased values
Received July 23, 1985. Accepted June 23, 1986. Address requests for reprints to: Albert0 Malesci, M.D., Instituto di Medicina Interna, Via Pace, 9-20122 Milano, Italy. This paper was presented in part at a poster session at the American Gastroenterological Association meeting in New Orleans, Louisiana, during Digestive Disease Week, May 1984. 0 1987 by the American Gastroenterological Association 0016-5085/87/$3.50
and Istituto
Scientific0
S. Raffaele,
Milan,
(cl20 U/ml), as seen in patients with localized pancreatic adenocarcinoma, are not conclusive for malignancy. The measurement of CA 19-9 to total protein ratio in pure pancreatic juice is proposed as an adjunctive, accurate diagnostic marker for early stages of pancreatic adenocarcinoma. The differential diagnosis between cancer of the pancreas and chronic pancreatitis is still a major clinical problem, one often not solved even with invasive techniques such as endoscopic retrograde pancreatography or percutaneous fine-needle biopsy. Therefore, because of the obvious prognostic and therapeutic implications of correct and early diagnosis, a diagnostic tool able to discriminate between benign and neoplastic pancreatic lesions and recognize early stages of resectable cancer is needed. In 1981 Koprowsky and colleagues reported that elevated levels of CA 19-9, a carbohydrate antigenic determinant, defined by a monoclonal antibody (l-31, were present in sera of patients with gastrointestinal cancer (4,s). Subsequent clinical studies employing a radioimmunometric assay have shown abnormally high serum concentrations of CA 19-9 in about 80% of patients with pancreatic adenocarcinoma and, conversely, in very few patients with chronic pancreatitis (6-12). To assess the specificity and the sensitivity of this test for differential diagnosis of chronic pancreatic disease, we measured CA 19-9 levels in preoperative sera from two large series of patients with chronic pancreatitis or histologically ascertained pancreatic
CA 19-9 IN SERUM AND PANCREATIC JUICE
January 1987
adenocarcinoma. The potential role of the test for detecting localized cancer was specifically investigated. In addition, for those patients from whom pure pancreatic juice had been successfully collected, we looked to see whether detectable amounts of CA 19-9 immunoreactivity were secreted by the pancreas and whether there were different secretory patterns in patients with benign or neoplastic pancreatic disease.
Patients and Methods Sixty-three patients with pancreatic cancer and 49 patients with chronic pancreatitis referred to our units during the period from January 1983 to June 1985 were studied retrospectively. Pancreatic ductal adenocarcinomas were confirmed in all patients by histologic examination of surgical biopsy specimens. Histopathologic grading and TNM staging of the tumor (13) were also done at the time of surgery (Table 1). The diagnosis of chronic pancreatitis was based on (a) calcifications on plain x-ray of the abdomen in 16 of the 49 patients; (b) decreased 60-90-min output of bicarbonate (<6 mmol) or lipase (<30,000 IU) after secretin (0.5 CU/kg h) plus caerulein (75 ng/kg h) infusion in 19 of zz investigated patients; (c) pathological findings at endoscopic retrograde pancreatography (33 of 37); (d) low (~20 ngiml) plasma cathodal trypsinogen (9 of 49); and (e) histologic confirmation on tissue removed at laparotomy in 11 cases. A follow-up period of at least 6 mo was required for nonneoplastic patients to be included in the study. Venous blood samples obtained from all patients after 12 h fasting and before any surgical or endoscopic procedure were processed for serum CA 19-9 determination. Pure pancreatic juice was collected by endoscopic cannulation of the main pancreatic duct for 10 min after intravenous injection of 1 CU/kg of synthetic secretin (Hoechst AG, Frankfurt, Federal Republic of Germany) from 23 patients with final diagnoses of cancer (Table l), 20 patients with chronic pancreatitis, and 22 subjects referred for endoscopic retrograde cholangiopancreatography and subsequently found not to have any pan-
Table
1. Number of Patients With Pancreatic Included in the Study
Cancer
Stage” T, - T,, N,, Mo Grade
High Moderate Poor Total
Serum
Juice
Tz, N,, Mo
N, and/or M,
Serum
Juice
Serum
Juice
8 4 4
6 3 2
5 8 6
2 4 1
9 10 9
1 1 3
16
11
19
7
28
5
a T1, tumor limited to the pancreas; TZ, tumor extends beyond the pancreas; TB, tumor invades adjacent organs; No. no lymph node metastasis: N,, regional or juxtaregional lymph node involvement; MO, no distant metastasis; to Klapdor et al. (13)).
M,, distant
metastasis
[according
61
creatic, cholestatic, or malignant disease. Selective cannulation of the pancreatic duct at the time of pancreatic juice sampling was confirmed by subsequent injection of the contrast medium. Pancreatic juice was immediately frozen and stored at -75°C until assayed. Both serum and pancreatic juice CA 19-9 concentrations were measured by a solid-phase immunoradiometric assay (Centocor Sorin Biomedica, Saluggia, Italy). A value of 40 U/ml (1 U = 0.8 ng of purified antigen) was taken as the upper normal limit for serum CA 19-9 determinations, as suggested by Ritts et al. (9). Serial dilutions with normal human serum (up to 1:300) were assayed. The actual juice concentration of CA 19-9 was calculated as the mean of the results of three dilutions, all falling within the range of the standard curve, and after subtraction of the immunoreactivity of the diluting serum. The juice assay was validated by recovery studies after addition of CA 19-9 standards to a prediluted sample of pancreatic juice (mean percentage recovery f SD = 96.4 ? 2.7). The protein content of the pancreatic juice was measured by the Lowry method (14). The Lewis blood antigen type was characterized by an agglutination method employing anti-Lea and anti-Leb specific antisera (Ortho Diagnostic System, Milan, Italy). The data were analyzed statistically by the Wilcoxon test.
Results Serum Abnormally high (>40 U/ml) serum CA 19-9 levels were found in 57 of 63 (90%) patients with pancreatic adenocarcinoma and in 5 of 49 (10%) patients with chronic pancreatitis. One patient with cancer and undetectable CA 19-9 antigen in his serum was found to have a Lewis a-b- blood antigen type (Figure 1). Fifteen of 16 (93%) patients with localized pancreatic cancer had abnormal serum CA 19-9 concentrations. Median serum CA 19-9 values were progressively higher in patients with localized (87.5 U/ml, range 6-600), infiltrating (400 U/ml, range 3-2500), and metastasized (1200 U/ml, range l-24,000) tumors. Of the 5 neoplastic patients with serum CA 19-9 levels <40 U/ml, 4 were found to have infiltrating or metastatic cancer (Figure 2). The median serum CA 19-9 value was significantly lower (p < 0.01) in patients with poorly differentiated cancer (69.5 U/ml, range l-2400) than in those with highly (820 U/ml, range 78-24,000) or moderately (650 U/ml, range 58-12,000) differentiated adenocarcinoma. All 5 patients whose CA 19-9 levels were <4O U/ml had poorly differentiated tumors (Figure 3). Serum concentrations of CA 19-9 >120 U/ml, which was the highest value observed among subjects with chronic pancreatitis, were found in 64% (40 of 62) of all the cancer patients and only in 31% (5 of 16) of those with localized tumors. As 120 U/ml
GASTROENTEROLOGYVol. 92.No.l
62 MALESCIETAL.
U/ml
10' . . :: .. ..
Figure
1. Serum CA 19-9 in chronic pancreatitis (CP) and pancreatic cancer [K). As a reference value of 40 U/ml was chosen, the test yielded 90% sensitivity and 90% specificity in discriminating between benign and neoplastic disease. 0, Lewis a-bpatient.
... . . . i
lo3
: 8. : . 08
v . ‘2
.
78
A.
_----
L_____ . .
K was taken as a reference value, the sensitivity of the test was progressively lower in subjects having the same stage but a poorer grade of cancer (Table 2). Four of the 5 patients with chronic pancreatitis and abnormal serum CA 19-9 levels had clinical or biochemical evidence of relapse, or both, at the time of their test. On repeat testing in a nonrelapse status, they all had normal serum CA 19-9 scores. In 1 patient without any evidence of recurrence but with a persistent jaundice, the serum antigen concentration remained elevated on three determinations (Table 3). Pure Pancreatic Juice CA 19-9 immunoreactivity was measurable in all but one of the tested samples of pancreatic juice, including those from controls with no pancreatic disease. CA 19-9 was undetectable in pancreatic juice as well as in serum in the patient with a Lewis a-b- blood antigen type. Pancreatic juice concentrations were significantly higher (p < 0.001) in patients with pancreatic adenocarcinoma (median value, 18 Ulpl, range l-428) than in chronic pancreatitis patients (1.2 U/pi, 0.1-23) and in subjects
without pancreatic disease (1.7 Ulpl, 0.1-18).Individual values are shown in Figure 4. Concentrations exceeding 10 U/p1 were found in 17 of 22 (77%) patients with cancer, in 3 of 20 (15%) patients with chronic pancreatitis, and in 3 of 22 (13%) controls. Two of the 3 patients with chronic pancreatitis and juice concentrations >lO II/p1 had pancreatolithiasis (Figure 4). When the ratio of CA 19-9 to protein content of pure pancreatic juice was calculated, all but 1 patient with adenocarcinoma had higher (>2.5 U/pg) ratios than all the controls or chronic pancreatitis patients. The only cancer patient with a lower ratio (2.1 UIpg) also had a low circulating level of CA 19-9 and was found to have a metastatic undifferentiated tumor (Figure 5). All 11 investigated patients with surgically resectable cancer had a juice CA 19-g/protein ratio higher than the cutoff value of 2.5 U/pg. Eight of these patients had serum antigen concentrations still in the range of chronic pancreatitis patients (Figure 6). Discussion This study was originally designed to assess the clinical reliability of serum CA 19-9 determina-
CA 19-9 IN SERUM AND PANCREATIC JUICE
January 1987
63
.. .
10 U/ml
. . . .
t ...
.. . .. . .
10 .
+ :
: . 2.
10
40
____
:
t
.
V: .
. .
.
. ------._---------i
__-----
.
_----
??
.
.
.
.
were progressively higher in patients with Figure 2. Serum CA 19-9 at different stages of pancreatic cancer. Median values (-1 localized (T1--T2, NO, M,), infiltrating (T3, N,, M& and metastasized (N, and/or M,) tumors. However, 4 of 5 patients with normal values already had advanced, unresectable cancer.
tion for differentiating pancreatic cancer from chronic pancreatitis. Using a reference cutoff of 40 U/ml, the sensitivity of the serum test in our large series of patients with histologically ascertained ductal adenocarcinoma was high (90%) and somewhat greater than previously reported for less precisely characterized patients (Figure 1) (5-12). Subjects lacking the Lewis gene, -5% of the normal population, are unable to synthesize the complete CA 19-9 antigen (15). Thus the finding of only one Lewis a-b- subject among
Table
2. Sensitivity’
of Serum
CA 19-9 Assay
to Detect
the 61 cancer patients may have helped to enhance the sensitivity of the test in our series as well. In the present study all falsely low levels were seen in cancer patients with poorly differentiated tumors (Figure 3). Histochemical studies employing the CA 19-9 monoclonal antibody confirmed that very few of these undifferentiated carcinomas produced the antigen. Our data confirmed also that the serum CA 19-9 assay is highly specific (90%) for pancreatic cancer when patients with chronic pancreatitis are chosen
Pancreatic Cancer According to Stage and Grade Moderate
High >120 U/ml +/nb T1 - Tz, NO, MO TB,No,Mo N1 and/or M1
418 415 919
(%I (50) (801 (1001
Median (U/ml) 135 800 2,000
Range (U/ml)
>120 U/ml +/n
(96)
(25) 78-600 114 114~2,500 518 (62) 840-24,000 lo/10 (100)
Median
(U/ml) 91 380 1,700
Poor Range (U/ml) 81-160 58-2,000 470-12,000
>120 U/ml -t/n
(%I
o/4 216 4/8
(0) (33) (501
Median (U/ml) 54 101.5 83.5
Range (U/ml) 3-75 3-980 l-2,400
See Table 1 for definition of symbols. QThe CA 19-9 reference value of 120 U/ml was the one necessary to include all patients chronic pancreatitis. b +/n is the ratio of patients with a positive (>12O U/ml) test to all patients of each subgroup.
with
64
MALESCI ET AL.
GASTROENTEROLOGY Vol. 92, No. 1
!
.
lo'
U/ml . . V .
..
! :
10'
Figure
.----____-_
3. Serum CA 19-9 in pancreatic
value (-) was significantly differentiated carcinoma.
.. .
! .
-A. . .
40
.
.. .
ld
.
...
. 8 :
..*. . I_________ .
?? ??
??
? ? ? ? ?A_____ ? :
cancer according to the grade of histologic differentiation. I, high; II, moderate; III, poor. Median lower in cases with poorly differentiated cancer. All patients with falsely normal levels had poorly
as controls (Figure 1). Patients with evidence of relapsing pancreatitis at the time of the test, however, may have values higher than 40 U/ml. Therefore, in these patients, a positive serum test cannot be of any help to the diagnosis until nonrelapse sera are reassayed [Table 3). Elevated serum CA 19-9 levels in chronic pancreatitis patients with a concomitant jaundice also must be evaluated cautiously because abnormal values can be found in patients with biliary obstruction, even in the absence of pancreatic disease (7,12). The serum test, although very sensitive (93%) for early resectable pancreatic cancer (Figure 2), was not able to discriminate between this condition and chronic pancreatitis, which is the crucial clinical dilemma. Only about one-third of patients with localized tumors had values >12O U/ml, which was the highest score observed in subjects with benign chronic pancreatic disease (Table 2). Thus, although in patients with focal pancreatic lesions moderately increased serum CA 19-9 values are highly suggestive, they cannot be considered to be diagnostic for malignancy. To see whether an abhormal pancreatic secretion of CA 19-9 could represent a marker of localized
pancreatic adenocarcinoma more specific than its serum concentration, we also assayed pure pancreatic juices obtained by endoscopic cannulation of the main pancreatic duct. CA 19-9 concentrations in
Table
3. Longitudinal
Serum
Patients With Chronic Abnormal Tesf
Patient
Determination
1
2
3 4 5
a > 40 U/ml.
Elapsed time (wk)
CA 19-9 Levels in the 5 Pancreatitis Having an
Clinical status
CA 19-9 (U/ml)
First Second
4
Relapse Nonrelapse
60 28
First Second Third
Relapse Nonrelapse Relapse
120
5 20
First Second
2
Relapse Nonrelapse
42 21
First Second
Relapse Nonrelapse
61
3
60
4 12
Nonrelapse Nonrelapse Nonrelapse
First Second Third
-
15
46
33 102
57
CA 19-9 IN SERUM
lanuary 1987
C
AND PANCREATIC
JUICE
65
..
6a Figure
3a
4. CA 19-9 concentration in pancreatic juice. Patients with pancreatic cancer (K) had significantly (p < 0.001) higher juice coucentrations than those with chronic pancreatitis (CP) or subjects with no pancreatic disease (C). Still a considerable overlap between cancer and benign disease was present. LI, patients with pancreatolithiasis.
. . .
i 0.’ . .
”
i .
C Figure
.
f .
:
.
CP
5. Ratio of CA 19-9 to secretory protein concentration in pancreatic same range as those with chronic pancreatitis (c2.5 Uipg).
juice. Only 1 of 22 patients
with cancer (K) had a ratio in the
66
MALESCI
ET AL.
GASTROENTEROLOGY
.
Vol. 92, No. 1
‘c;
I I
10
I I
’
b
I
2
I I I Figure
6. Comparison of serum CA 19-9 levels and pancreatic juice CA 19S/protein ratio in detecting pancreatic cancer. Cutoff values including all patients with chronic pancreatitis are indicated by dotted lines. The juice test was more accurate than the serum determination in discriminating between cancer and chronic pancreatitis. 0, localized; 0, infiltrating; *, metastatic carcinoma.
.
5 ,
. 0
-------??
I
c
f I I I ?? I I 1 01 I L____--------___ I I I I I
100 SERUM pancreatic juice were significantly higher in patients with pancreatic cancer than in nonneopiastic patients, including those with chronic pancreatitis. These data are in partial accordance with those reported by Schmiegel et al. (16) who, however, observed a considerably wider range of juice concentrations in patients with chronic pancreatitis and in controls. In contrast, Tatsuta et al. (17) recently reported much lower levels of antigen in the pancreatic juice of patients with cancer. As the commercially available CA 19-9 kit is specifically designed for serum measurement, the lack of a thorough validation of the juice assayed by these investigators might account for such discrepancies. Despite the significantly different distribution of results in the two classes of patients, the absolute CA 19-9 juice concentration was not apt to discriminate between patients with pancreatic cancer and those with benign chronic pancreatic disease. We then decided to refer the juice antigen concentration to some simple secretory parameter, such as the protein concentration. The choice of this reference parameter was also supported by the finding of high antigen levels in the juice of patients with pancreatolithiasis, a condition that seems to be associated with pancreatic protein hypersecretion (18). When ratios of CA 19-9 to total protein were calculated, the juice assay became very accurate for
E
C
.
cr z
I
r’
200
300-
(U/ml)
differentiating pancreatic cancer from chronic pancreatitis (Figure 5). In particular, juice CA 19-9 immunoreactivity, as corrected for the protein content, was much more sensitive than circulating antigen levels in detecting localized pancreatic adenocarcinoma, if cutoff values including all patients with chronic pancreatitis were applied (Figure 6). We propose the determination of CA 19-9 to protein ratio in pure pancreatic juice obtained by endoscopic cannulation as adjunctive and accurate diagnostic test in those patients with focal pancreatic lesions having only slightly increased serum CA 19-9 levels, which are not definitive for the presence of a pancreatic adenocarcinoma.
References Koprowsky H, Steplewski Z, Mitchell K, Herlyn M, Herlyn D, Fulner P. Colorectal carcinoma antigens detected by hybridoma antibodies. Somatic Cell Genet 1979;5:957-72. Magnani JL, Brockhause M, Smith DF, et al. A monosialoganglioside is a monoclonal antibody-defined antigen of colon carcinoma. Science 1981;212:55-6. Magnani JL, Nilsson B, Brockhaus M, et al. A monoclonal antibody-defined antigen associated with gastrointestinal cancer is a ganglioside containing sialylated lacto-N-fucopentaose ll.J Biol Chem 1982;257:14365-9. Koprowski H, Herlyn M, Steplewski Z, Sears HF. Specific antigen in serum of patients with colon carcinoma. Science 1981;212:53-5.
January
CA 19-9 IN SERUM
1987
5. Herlyn M, Sears HF, Steplewski Z, Koprowski H. Monoclonal antibody detection of a circulating tumor-associated antigen. I. Presence of antigen in sera of patients with colorectal, gastric and pancreatic carcinoma. J Clin Chem 1982;2:13540. 6. Del Villano BC, Brennan S, Brock P, et al. Radioimmunometric assay for a monoclonal antibody-defined tumor marker. ClinChem 1983;29:549-52. 7. Malesci A, Tommasini M, Bocchia P, et al. Differential diagnosis of pancreatic cancer and chronic pancreatitis by a monoclonal antibody detecting a new cancer associated antigen [CA 19-9).La Ricerca Clin Lab 1984;14:303-6. 8. Jalanko H, Kuusela P, Roberts P, Sipponer P, Haglund C, Makela 0. Comparison of a new tumor marker, CA 19-STM, with alpha-fetoprotein and carcinoembryonic antigen in patients with upper gastrointestinal disease. J Clin Path01 1984; 37:218-22. 9. Ritts RE Jr, Del Villano BC, Go VLW, Herzerman RB, Klug TL, Zurawski VR Jr. Initial clinical evaluation of an immunoradiametric assay for CA 19-9 using the NC1 serum bank. Int J Cancer 1984;33:339-45. 10. Gupta MK, Arciaga R, Bocci L, Tubbs R, Bukowski R, Deodhar SD. Measurement of a monoclonal-antibody-defined antigen (CA 19-9) in the sera of patients with malignant and nonmalignant disease: comparison with carcinoembryonic antigen. Cancer 1985;56:277-83. 11. Yoskikiwa T, Nishida K. Taniqawa M, Fukumoto K, Kondo
AND PANCREATIC
IUICE
67
M. Carbohydrate antigen determinant (CA 19-9) and other tumor markers in gastrointestinal malignancies. Digestion 1985;31:67-76. 12. Steinberg WM, Gelfand R, Anderson KK, et al. Comparison of the sensitivity and specificity of the CA 19-9 and carcinoma embryonic antigen assays in detecting cancer of the pancreas. Gastroenterology 13. Klapdor
1986;90:343-9.
R, Lehmann
for classification Invest
U, Klopplel
of exocrine
G. Necessity
pancreatic
and proposal
carcinoma.
Eur J Clin
1983;13:AZO.
14. Lowry
OH,
Rosebrough
measurement
with
NJ, Fazz
the
Folin
RL, Randall
reagent.
RJ. Protein
J Biol
Chem
1951;
193:265-75. 15. Itzkowitz
SH, Kim
Gastroenterology 16. Schmiegel body
YS. New
carbohydrate
tumor
markers.
1986;90:491-4.
WH, Kreiker
defines
C, Eberl
CA 19-9
W. et al. Monoclonal
in pancreatic
juices
and
anti-
sera.
Gut
1985;26:456-60. 17. Tatsuta
M, Yamamura
the serum, material
pure in the
Cancer
diagnosis
juice,
and
of malignant
of CA 19-9 in
aspirated
pancreatic
pancreatic
tumor.
1985;56:2669-73.
18. Multigner stone
H, Iishi H, et al. Values
pancreatic
L, Montalto
formation
MV. Sarles Amsterdam:
G, Sarles
in chronic
M. De Caro A. New aspects
pancreatitis.
M, eds. Pancreatitis-concepts Elsevier
Science
Publishers,
of
In: Gyr KE. Singer and classification. 1984:171-5.
1987;92:60-7
Determination of CA 19-9Antigen in Serum and Pancreatic Juice for Differential Diagnosis of Pancreatic Adenocarcinoma From Chronic Pancreatitis ALBERT0 MALESCI, MAURIZIO A. TOMMASINI, CLAUDIO BONATO, PAOLA BOCCHIA, MAURIZO BERSANI, ALESSANDRO ZERBI, EDOARDO BERETTA, and VALERIO DI CARLO Istituto Italy
di Medicina
Interna,
Osoedale
Policlinico
Serum CA 19-9 levels were measured in 63 patients with ductal pancreatic adenocarcinoma and in 49 patients with chronic pancreatitis. Concentrations were abnormally high (>40 U/ml) in 57 (90%) patients with cancer and only in 5 (10%) patients with chronic pancreatitis. All patients with falsely normal serum values had poorly differentiated carcinomas. Median CA 19-9 concentrations were progressively higher in patients with more advanced cancer. Fifteen of 16 (93%) patients with localized cancer has abnormal serum levels but only 5 (31%) of them had values >120 U/ml, which was the highest score observed in patients with chronic pancreatitis. Pure pancreatic juice was obtained endoscopically from 23 patients with pancreatic cancer and from 20 with chronic pancreatitis. CA 19-9 concentrations in pancreatic juice were significantly higher in patients with cancer than in nonneoplastic patients. AI1 11 patients with resectable cancer investigated had a ratio of CA 19-9 to secretory protein concentration in pancreatic juice above the range of patients with chronic pancreatitis. We conclude that serum CA 19-9 determination is highly sensitive and specific for the differential diagnosis of pancreatic cancer versus chronic pancreatitis. However, moderately increased values
Received July 23, 1985. Accepted June 23, 1986. Address requests for reprints to: Albert0 Malesci, M.D., Instituto di Medicina Interna, Via Pace, 9-20122 Milano, Italy. This paper was presented in part at a poster session at the American Gastroenterological Association meeting in New Orleans, Louisiana, during Digestive Disease Week, May 1984. 0 1987 by the American Gastroenterological Association 0016-5085/87/$3.50
and Istituto
Scientific0
S. Raffaele,
Milan,
(cl20 U/ml), as seen in patients with localized pancreatic adenocarcinoma, are not conclusive for malignancy. The measurement of CA 19-9 to total protein ratio in pure pancreatic juice is proposed as an adjunctive, accurate diagnostic marker for early stages of pancreatic adenocarcinoma. The differential diagnosis between cancer of the pancreas and chronic pancreatitis is still a major clinical problem, one often not solved even with invasive techniques such as endoscopic retrograde pancreatography or percutaneous fine-needle biopsy. Therefore, because of the obvious prognostic and therapeutic implications of correct and early diagnosis, a diagnostic tool able to discriminate between benign and neoplastic pancreatic lesions and recognize early stages of resectable cancer is needed. In 1981 Koprowsky and colleagues reported that elevated levels of CA 19-9, a carbohydrate antigenic determinant, defined by a monoclonal antibody (l-31, were present in sera of patients with gastrointestinal cancer (4,s). Subsequent clinical studies employing a radioimmunometric assay have shown abnormally high serum concentrations of CA 19-9 in about 80% of patients with pancreatic adenocarcinoma and, conversely, in very few patients with chronic pancreatitis (6-12). To assess the specificity and the sensitivity of this test for differential diagnosis of chronic pancreatic disease, we measured CA 19-9 levels in preoperative sera from two large series of patients with chronic pancreatitis or histologically ascertained pancreatic
CA 19-9 IN SERUM AND PANCREATIC JUICE
January 1987
adenocarcinoma. The potential role of the test for detecting localized cancer was specifically investigated. In addition, for those patients from whom pure pancreatic juice had been successfully collected, we looked to see whether detectable amounts of CA 19-9 immunoreactivity were secreted by the pancreas and whether there were different secretory patterns in patients with benign or neoplastic pancreatic disease.
Patients and Methods Sixty-three patients with pancreatic cancer and 49 patients with chronic pancreatitis referred to our units during the period from January 1983 to June 1985 were studied retrospectively. Pancreatic ductal adenocarcinomas were confirmed in all patients by histologic examination of surgical biopsy specimens. Histopathologic grading and TNM staging of the tumor (13) were also done at the time of surgery (Table 1). The diagnosis of chronic pancreatitis was based on (a) calcifications on plain x-ray of the abdomen in 16 of the 49 patients; (b) decreased 60-90-min output of bicarbonate (<6 mmol) or lipase (<30,000 IU) after secretin (0.5 CU/kg h) plus caerulein (75 ng/kg h) infusion in 19 of zz investigated patients; (c) pathological findings at endoscopic retrograde pancreatography (33 of 37); (d) low (~20 ngiml) plasma cathodal trypsinogen (9 of 49); and (e) histologic confirmation on tissue removed at laparotomy in 11 cases. A follow-up period of at least 6 mo was required for nonneoplastic patients to be included in the study. Venous blood samples obtained from all patients after 12 h fasting and before any surgical or endoscopic procedure were processed for serum CA 19-9 determination. Pure pancreatic juice was collected by endoscopic cannulation of the main pancreatic duct for 10 min after intravenous injection of 1 CU/kg of synthetic secretin (Hoechst AG, Frankfurt, Federal Republic of Germany) from 23 patients with final diagnoses of cancer (Table l), 20 patients with chronic pancreatitis, and 22 subjects referred for endoscopic retrograde cholangiopancreatography and subsequently found not to have any pan-
Table
1. Number of Patients With Pancreatic Included in the Study
Cancer
Stage” T, - T,, N,, Mo Grade
High Moderate Poor Total
Serum
Juice
Tz, N,, Mo
N, and/or M,
Serum
Juice
Serum
Juice
8 4 4
6 3 2
5 8 6
2 4 1
9 10 9
1 1 3
16
11
19
7
28
5
a T1, tumor limited to the pancreas; TZ, tumor extends beyond the pancreas; TB, tumor invades adjacent organs; No. no lymph node metastasis: N,, regional or juxtaregional lymph node involvement; MO, no distant metastasis; to Klapdor et al. (13)).
M,, distant
metastasis
[according
61
creatic, cholestatic, or malignant disease. Selective cannulation of the pancreatic duct at the time of pancreatic juice sampling was confirmed by subsequent injection of the contrast medium. Pancreatic juice was immediately frozen and stored at -75°C until assayed. Both serum and pancreatic juice CA 19-9 concentrations were measured by a solid-phase immunoradiometric assay (Centocor Sorin Biomedica, Saluggia, Italy). A value of 40 U/ml (1 U = 0.8 ng of purified antigen) was taken as the upper normal limit for serum CA 19-9 determinations, as suggested by Ritts et al. (9). Serial dilutions with normal human serum (up to 1:300) were assayed. The actual juice concentration of CA 19-9 was calculated as the mean of the results of three dilutions, all falling within the range of the standard curve, and after subtraction of the immunoreactivity of the diluting serum. The juice assay was validated by recovery studies after addition of CA 19-9 standards to a prediluted sample of pancreatic juice (mean percentage recovery f SD = 96.4 ? 2.7). The protein content of the pancreatic juice was measured by the Lowry method (14). The Lewis blood antigen type was characterized by an agglutination method employing anti-Lea and anti-Leb specific antisera (Ortho Diagnostic System, Milan, Italy). The data were analyzed statistically by the Wilcoxon test.
Results Serum Abnormally high (>40 U/ml) serum CA 19-9 levels were found in 57 of 63 (90%) patients with pancreatic adenocarcinoma and in 5 of 49 (10%) patients with chronic pancreatitis. One patient with cancer and undetectable CA 19-9 antigen in his serum was found to have a Lewis a-b- blood antigen type (Figure 1). Fifteen of 16 (93%) patients with localized pancreatic cancer had abnormal serum CA 19-9 concentrations. Median serum CA 19-9 values were progressively higher in patients with localized (87.5 U/ml, range 6-600), infiltrating (400 U/ml, range 3-2500), and metastasized (1200 U/ml, range l-24,000) tumors. Of the 5 neoplastic patients with serum CA 19-9 levels <40 U/ml, 4 were found to have infiltrating or metastatic cancer (Figure 2). The median serum CA 19-9 value was significantly lower (p < 0.01) in patients with poorly differentiated cancer (69.5 U/ml, range l-2400) than in those with highly (820 U/ml, range 78-24,000) or moderately (650 U/ml, range 58-12,000) differentiated adenocarcinoma. All 5 patients whose CA 19-9 levels were <4O U/ml had poorly differentiated tumors (Figure 3). Serum concentrations of CA 19-9 >120 U/ml, which was the highest value observed among subjects with chronic pancreatitis, were found in 64% (40 of 62) of all the cancer patients and only in 31% (5 of 16) of those with localized tumors. As 120 U/ml
GASTROENTEROLOGYVol. 92.No.l
62 MALESCIETAL.
U/ml
10' . . :: .. ..
Figure
1. Serum CA 19-9 in chronic pancreatitis (CP) and pancreatic cancer [K). As a reference value of 40 U/ml was chosen, the test yielded 90% sensitivity and 90% specificity in discriminating between benign and neoplastic disease. 0, Lewis a-bpatient.
... . . . i
lo3
: 8. : . 08
v . ‘2
.
78
A.
_----
L_____ . .
K was taken as a reference value, the sensitivity of the test was progressively lower in subjects having the same stage but a poorer grade of cancer (Table 2). Four of the 5 patients with chronic pancreatitis and abnormal serum CA 19-9 levels had clinical or biochemical evidence of relapse, or both, at the time of their test. On repeat testing in a nonrelapse status, they all had normal serum CA 19-9 scores. In 1 patient without any evidence of recurrence but with a persistent jaundice, the serum antigen concentration remained elevated on three determinations (Table 3). Pure Pancreatic Juice CA 19-9 immunoreactivity was measurable in all but one of the tested samples of pancreatic juice, including those from controls with no pancreatic disease. CA 19-9 was undetectable in pancreatic juice as well as in serum in the patient with a Lewis a-b- blood antigen type. Pancreatic juice concentrations were significantly higher (p < 0.001) in patients with pancreatic adenocarcinoma (median value, 18 Ulpl, range l-428) than in chronic pancreatitis patients (1.2 U/pi, 0.1-23) and in subjects
without pancreatic disease (1.7 Ulpl, 0.1-18).Individual values are shown in Figure 4. Concentrations exceeding 10 U/p1 were found in 17 of 22 (77%) patients with cancer, in 3 of 20 (15%) patients with chronic pancreatitis, and in 3 of 22 (13%) controls. Two of the 3 patients with chronic pancreatitis and juice concentrations >lO II/p1 had pancreatolithiasis (Figure 4). When the ratio of CA 19-9 to protein content of pure pancreatic juice was calculated, all but 1 patient with adenocarcinoma had higher (>2.5 U/pg) ratios than all the controls or chronic pancreatitis patients. The only cancer patient with a lower ratio (2.1 UIpg) also had a low circulating level of CA 19-9 and was found to have a metastatic undifferentiated tumor (Figure 5). All 11 investigated patients with surgically resectable cancer had a juice CA 19-g/protein ratio higher than the cutoff value of 2.5 U/pg. Eight of these patients had serum antigen concentrations still in the range of chronic pancreatitis patients (Figure 6). Discussion This study was originally designed to assess the clinical reliability of serum CA 19-9 determina-
CA 19-9 IN SERUM AND PANCREATIC JUICE
January 1987
63
.. .
10 U/ml
. . . .
t ...
.. . .. . .
10 .
+ :
: . 2.
10
40
____
:
t
.
V: .
. .
.
. ------._---------i
__-----
.
_----
??
.
.
.
.
were progressively higher in patients with Figure 2. Serum CA 19-9 at different stages of pancreatic cancer. Median values (-1 localized (T1--T2, NO, M,), infiltrating (T3, N,, M& and metastasized (N, and/or M,) tumors. However, 4 of 5 patients with normal values already had advanced, unresectable cancer.
tion for differentiating pancreatic cancer from chronic pancreatitis. Using a reference cutoff of 40 U/ml, the sensitivity of the serum test in our large series of patients with histologically ascertained ductal adenocarcinoma was high (90%) and somewhat greater than previously reported for less precisely characterized patients (Figure 1) (5-12). Subjects lacking the Lewis gene, -5% of the normal population, are unable to synthesize the complete CA 19-9 antigen (15). Thus the finding of only one Lewis a-b- subject among
Table
2. Sensitivity’
of Serum
CA 19-9 Assay
to Detect
the 61 cancer patients may have helped to enhance the sensitivity of the test in our series as well. In the present study all falsely low levels were seen in cancer patients with poorly differentiated tumors (Figure 3). Histochemical studies employing the CA 19-9 monoclonal antibody confirmed that very few of these undifferentiated carcinomas produced the antigen. Our data confirmed also that the serum CA 19-9 assay is highly specific (90%) for pancreatic cancer when patients with chronic pancreatitis are chosen
Pancreatic Cancer According to Stage and Grade Moderate
High >120 U/ml +/nb T1 - Tz, NO, MO TB,No,Mo N1 and/or M1
418 415 919
(%I (50) (801 (1001
Median (U/ml) 135 800 2,000
Range (U/ml)
>120 U/ml +/n
(96)
(25) 78-600 114 114~2,500 518 (62) 840-24,000 lo/10 (100)
Median
(U/ml) 91 380 1,700
Poor Range (U/ml) 81-160 58-2,000 470-12,000
>120 U/ml -t/n
(%I
o/4 216 4/8
(0) (33) (501
Median (U/ml) 54 101.5 83.5
Range (U/ml) 3-75 3-980 l-2,400
See Table 1 for definition of symbols. QThe CA 19-9 reference value of 120 U/ml was the one necessary to include all patients chronic pancreatitis. b +/n is the ratio of patients with a positive (>12O U/ml) test to all patients of each subgroup.
with
64
MALESCI ET AL.
GASTROENTEROLOGY Vol. 92, No. 1
!
.
lo'
U/ml . . V .
..
! :
10'
Figure
.----____-_
3. Serum CA 19-9 in pancreatic
value (-) was significantly differentiated carcinoma.
.. .
! .
-A. . .
40
.
.. .
ld
.
...
. 8 :
..*. . I_________ .
?? ??
??
? ? ? ? ?A_____ ? :
cancer according to the grade of histologic differentiation. I, high; II, moderate; III, poor. Median lower in cases with poorly differentiated cancer. All patients with falsely normal levels had poorly
as controls (Figure 1). Patients with evidence of relapsing pancreatitis at the time of the test, however, may have values higher than 40 U/ml. Therefore, in these patients, a positive serum test cannot be of any help to the diagnosis until nonrelapse sera are reassayed [Table 3). Elevated serum CA 19-9 levels in chronic pancreatitis patients with a concomitant jaundice also must be evaluated cautiously because abnormal values can be found in patients with biliary obstruction, even in the absence of pancreatic disease (7,12). The serum test, although very sensitive (93%) for early resectable pancreatic cancer (Figure 2), was not able to discriminate between this condition and chronic pancreatitis, which is the crucial clinical dilemma. Only about one-third of patients with localized tumors had values >12O U/ml, which was the highest score observed in subjects with benign chronic pancreatic disease (Table 2). Thus, although in patients with focal pancreatic lesions moderately increased serum CA 19-9 values are highly suggestive, they cannot be considered to be diagnostic for malignancy. To see whether an abhormal pancreatic secretion of CA 19-9 could represent a marker of localized
pancreatic adenocarcinoma more specific than its serum concentration, we also assayed pure pancreatic juices obtained by endoscopic cannulation of the main pancreatic duct. CA 19-9 concentrations in
Table
3. Longitudinal
Serum
Patients With Chronic Abnormal Tesf
Patient
Determination
1
2
3 4 5
a > 40 U/ml.
Elapsed time (wk)
CA 19-9 Levels in the 5 Pancreatitis Having an
Clinical status
CA 19-9 (U/ml)
First Second
4
Relapse Nonrelapse
60 28
First Second Third
Relapse Nonrelapse Relapse
120
5 20
First Second
2
Relapse Nonrelapse
42 21
First Second
Relapse Nonrelapse
61
3
60
4 12
Nonrelapse Nonrelapse Nonrelapse
First Second Third
-
15
46
33 102
57
CA 19-9 IN SERUM
lanuary 1987
C
AND PANCREATIC
JUICE
65
..
6a Figure
3a
4. CA 19-9 concentration in pancreatic juice. Patients with pancreatic cancer (K) had significantly (p < 0.001) higher juice coucentrations than those with chronic pancreatitis (CP) or subjects with no pancreatic disease (C). Still a considerable overlap between cancer and benign disease was present. LI, patients with pancreatolithiasis.
. . .
i 0.’ . .
”
i .
C Figure
.
f .
:
.
CP
5. Ratio of CA 19-9 to secretory protein concentration in pancreatic same range as those with chronic pancreatitis (c2.5 Uipg).
juice. Only 1 of 22 patients
with cancer (K) had a ratio in the
66
MALESCI
ET AL.
GASTROENTEROLOGY
.
Vol. 92, No. 1
‘c;
I I
10
I I
’
b
I
2
I I I Figure
6. Comparison of serum CA 19-9 levels and pancreatic juice CA 19S/protein ratio in detecting pancreatic cancer. Cutoff values including all patients with chronic pancreatitis are indicated by dotted lines. The juice test was more accurate than the serum determination in discriminating between cancer and chronic pancreatitis. 0, localized; 0, infiltrating; *, metastatic carcinoma.
.
5 ,
. 0
-------??
I
c
f I I I ?? I I 1 01 I L____--------___ I I I I I
100 SERUM pancreatic juice were significantly higher in patients with pancreatic cancer than in nonneopiastic patients, including those with chronic pancreatitis. These data are in partial accordance with those reported by Schmiegel et al. (16) who, however, observed a considerably wider range of juice concentrations in patients with chronic pancreatitis and in controls. In contrast, Tatsuta et al. (17) recently reported much lower levels of antigen in the pancreatic juice of patients with cancer. As the commercially available CA 19-9 kit is specifically designed for serum measurement, the lack of a thorough validation of the juice assayed by these investigators might account for such discrepancies. Despite the significantly different distribution of results in the two classes of patients, the absolute CA 19-9 juice concentration was not apt to discriminate between patients with pancreatic cancer and those with benign chronic pancreatic disease. We then decided to refer the juice antigen concentration to some simple secretory parameter, such as the protein concentration. The choice of this reference parameter was also supported by the finding of high antigen levels in the juice of patients with pancreatolithiasis, a condition that seems to be associated with pancreatic protein hypersecretion (18). When ratios of CA 19-9 to total protein were calculated, the juice assay became very accurate for
E
C
.
cr z
I
r’
200
300-
(U/ml)
differentiating pancreatic cancer from chronic pancreatitis (Figure 5). In particular, juice CA 19-9 immunoreactivity, as corrected for the protein content, was much more sensitive than circulating antigen levels in detecting localized pancreatic adenocarcinoma, if cutoff values including all patients with chronic pancreatitis were applied (Figure 6). We propose the determination of CA 19-9 to protein ratio in pure pancreatic juice obtained by endoscopic cannulation as adjunctive and accurate diagnostic test in those patients with focal pancreatic lesions having only slightly increased serum CA 19-9 levels, which are not definitive for the presence of a pancreatic adenocarcinoma.
References Koprowsky H, Steplewski Z, Mitchell K, Herlyn M, Herlyn D, Fulner P. Colorectal carcinoma antigens detected by hybridoma antibodies. Somatic Cell Genet 1979;5:957-72. Magnani JL, Brockhause M, Smith DF, et al. A monosialoganglioside is a monoclonal antibody-defined antigen of colon carcinoma. Science 1981;212:55-6. Magnani JL, Nilsson B, Brockhaus M, et al. A monoclonal antibody-defined antigen associated with gastrointestinal cancer is a ganglioside containing sialylated lacto-N-fucopentaose ll.J Biol Chem 1982;257:14365-9. Koprowski H, Herlyn M, Steplewski Z, Sears HF. Specific antigen in serum of patients with colon carcinoma. Science 1981;212:53-5.
January
CA 19-9 IN SERUM
1987
5. Herlyn M, Sears HF, Steplewski Z, Koprowski H. Monoclonal antibody detection of a circulating tumor-associated antigen. I. Presence of antigen in sera of patients with colorectal, gastric and pancreatic carcinoma. J Clin Chem 1982;2:13540. 6. Del Villano BC, Brennan S, Brock P, et al. Radioimmunometric assay for a monoclonal antibody-defined tumor marker. ClinChem 1983;29:549-52. 7. Malesci A, Tommasini M, Bocchia P, et al. Differential diagnosis of pancreatic cancer and chronic pancreatitis by a monoclonal antibody detecting a new cancer associated antigen [CA 19-9).La Ricerca Clin Lab 1984;14:303-6. 8. Jalanko H, Kuusela P, Roberts P, Sipponer P, Haglund C, Makela 0. Comparison of a new tumor marker, CA 19-STM, with alpha-fetoprotein and carcinoembryonic antigen in patients with upper gastrointestinal disease. J Clin Path01 1984; 37:218-22. 9. Ritts RE Jr, Del Villano BC, Go VLW, Herzerman RB, Klug TL, Zurawski VR Jr. Initial clinical evaluation of an immunoradiametric assay for CA 19-9 using the NC1 serum bank. Int J Cancer 1984;33:339-45. 10. Gupta MK, Arciaga R, Bocci L, Tubbs R, Bukowski R, Deodhar SD. Measurement of a monoclonal-antibody-defined antigen (CA 19-9) in the sera of patients with malignant and nonmalignant disease: comparison with carcinoembryonic antigen. Cancer 1985;56:277-83. 11. Yoskikiwa T, Nishida K. Taniqawa M, Fukumoto K, Kondo
AND PANCREATIC
IUICE
67
M. Carbohydrate antigen determinant (CA 19-9) and other tumor markers in gastrointestinal malignancies. Digestion 1985;31:67-76. 12. Steinberg WM, Gelfand R, Anderson KK, et al. Comparison of the sensitivity and specificity of the CA 19-9 and carcinoma embryonic antigen assays in detecting cancer of the pancreas. Gastroenterology 13. Klapdor
1986;90:343-9.
R, Lehmann
for classification Invest
U, Klopplel
of exocrine
G. Necessity
pancreatic
and proposal
carcinoma.
Eur J Clin
1983;13:AZO.
14. Lowry
OH,
Rosebrough
measurement
with
NJ, Fazz
the
Folin
RL, Randall
reagent.
RJ. Protein
J Biol
Chem
1951;
193:265-75. 15. Itzkowitz
SH, Kim
Gastroenterology 16. Schmiegel body
YS. New
carbohydrate
tumor
markers.
1986;90:491-4.
WH, Kreiker
defines
C, Eberl
CA 19-9
W. et al. Monoclonal
in pancreatic
juices
and
anti-
sera.
Gut
1985;26:456-60. 17. Tatsuta
M, Yamamura
the serum, material
pure in the
Cancer
diagnosis
juice,
and
of malignant
of CA 19-9 in
aspirated
pancreatic
pancreatic
tumor.
1985;56:2669-73.
18. Multigner stone
H, Iishi H, et al. Values
pancreatic
L, Montalto
formation
MV. Sarles Amsterdam:
G, Sarles
in chronic
M. De Caro A. New aspects
pancreatitis.
M, eds. Pancreatitis-concepts Elsevier
Science
Publishers,
of
In: Gyr KE. Singer and classification. 1984:171-5.