Annals of Surgical Oncology, 11(7):644649

DOI: 10.1245/ASO.2004.11.025

Undetectable Preoperative Levels of Serum CA 19-9 Correlate

with Improved Survival for Patients with Resectable Pancreatic
Adenocarcinoma
Adam C. Berger, MD, Ingrid M. Meszoely, MD, Eric A. Ross, PhD, James C. Watson, MD, and
John P. Hoffman, MD

Background: Serum levels of CA19-9 have been shown to correlate with both recurrence and
survival in patients with pancreatic cancer. However, little is known about the prognosis for patients
with undetectable levels of serum CA19-9.
Methods: One hundred twenty-nine patients with pancreatic cancer who underwent preoperative
assessment of serum CA19-9 followed by resection with curative intent between 1990 and 2002
were retrospectively analyzed. Data collected included preoperative serum CA19-9 level (U/mL),
age, pathologic staging, and survival. Data were analyzed with the SAS system according to four
distinct preoperative serum CA19-9 levels: undetectable, normal (37), 38 200, and 200 U/mL.
Results: Serum CA19-9 levels ranged from undetectable to 16,300 U/mL. Stage III/IV disease
accounted for 86%, 67%, 59%, and 53% of patients in the four CA19-9 groups. The overall median
and 5-year survivals were 19 months and 11%, respectively. Survival was similar between nonsecretors and those with normal CA 19-9 levels. However, both groups had statistically significant
prolonged survival compared with the two groups with elevated CA 19-9 levels (P .003). The
only factors that were significant on univariate and multivariate analysis for overall survival were
lymph node positivity (P .015 and .002) and CA 19-9 grouping (P .003 and P .0001).
Although this group of patients presented with predominately advanced-stage disease, their overall
survival was superior.
Conclusions: These findings suggest that patients who present with undetectable preoperative
CA19-9 levels and potentially resectable pancreatic cancer, regardless of advanced stage, should be
considered candidates for aggressive therapy.
Key Words: CA 19-9 Lewis blood groupPancreatic cancerPrognosis.

The tumor-associated antigen CA 19-9 was first described by Koprowski et al. in 19791 as a marker for
colorectal cancer; since that time, it has become the most
important tumor marker for pancreatic adenocarcinoma.
Several hundred reports worldwide have attested to its

clinical usefulness in the diagnosis, prognosis, and monitoring of pancreatic cancer. Recent evidence has shown
that serum CA 19-9 is an independent predictor of recurrence and survival after resection,2 4 and its levels
correlate with tumor burden and metastatic disease. One
study demonstrated that patients with resectable tumors
had significantly lower CA 19-9 levels than those with
unresectable tumors.5 Finally, in patients being treated
with gemcitabine for advanced pancreatic cancer, a decrease of CA 19-9 by 20% was the strongest independent predictor of survival in multivariate analysis.6
CA 19-9 has also been identified as a monosialoganglioside/glycolipid and sialyl derivative of lacto-Nfucopenteose II (sialyl-Lewis[a], hapten of human
Lewis[a] blood-group determinant).7 As such, CA 19-9
levels detected by conventional antibody tests may be

Received November 14, 2003; accepted March 12, 2004.

From the Departments of Surgical Oncology (ACB, IMM, JCW,
JPH) and Biostatistics (EAR), Fox Chase Cancer Center, Philadelphia,
Pennsylvania.
Presented as a poster at the 56th Annual Cancer Symposium of the
Society of Surgical Oncology, Los Angeles, California, March 59,
2003.
Address correspondence and reprint requests to: Adam C. Berger,
MD, Department of Surgical Oncology, Fox Chase Cancer Center, 333
Cottman Avenue, Philadelphia, PA 19111; Fax: 215-728-2773; E-mail:
a_berger@fccc.edu.
Published by Lippincott Williams & Wilkins 2004 The Society of Surgical
Oncology, Inc.

644

UNDETECTABLE CA 19-9 LEVELS IN PANCREATIC CANCER

affected by Lewis blood group phenotypes. In fact, pancreatic cancer patients with a Lewis negative (a-, b-)
phenotype will have an undetectable CA 19-9 level.8
Fortunately, this phenotype occurs in only 7% to 10% of
the population.9 Therefore, one cannot follow the serum
CA 19-9 levels in Lewis-negative patients with pancreatic cancer; these patients may have undetectable CA
19-9 levels even in the face of metastatic or recurrent
disease. Therefore, we set out to determine the characteristics and prognosis of pancreatic cancer in patients
who are CA 19-9 nonsecretors in comparison with patients with normal CA 19-9 levels at diagnosis and those
with elevated CA 19-9 levels. This is the first report in
the literature looking specifically at these patients.
MATERIALS AND METHODS
We retrospectively reviewed the password-protected
pancreatic cancer database at the Fox Chase Cancer
Center to identify 129 patients who underwent preoperative assessment of serum CA 19-9 levels followed by
resection with curative intent between 1990 and 2002.
Hospital, clinic, and physician records were reviewed for
pertinent information. Data collected included age, sex,
date of diagnosis, date of surgery, use of induction adjuvant chemotherapy and radiation, serum CA 19-9
level, pathologic T stage, lymph node status, overall
pathologic stage, margin status, and survival.
Serum CA 19-9 levels were determined with use of a
CA 19-9 radioimmunoassay kit manufactured by Abbott
Laboratories (Chicago, IL). The recommended normal
value of 37 U/mL was used for the serum assays. Patients
were divided into four groups based on their CA 19-9
level at diagnosis: undetectable, normal (37 U/mL),
38 200 U/mL, and 200 U/mL. These groups comprised 5%, 16%, 34%, and 44% of the entire population.
Lewis antigen testing was not performed for a couple of
reasons. First of all, an assay was not available for the
early part of this study. Second, our blood bank is unable
to retrospectively perform the test because it can be
performed on only fresh red blood cells. Overall survival
(OS) was defined as date of diagnosis until time of death
or last follow-up, and disease-free survival (DFS) was
measured from time of operation to date of recurrence.
Survival follow-up was complete for all patients, but 14
patients had incomplete data regarding time to recurrence and were thus left out of the DFS analysis.
Univariate analysis for categorical variables such as
stage, lymph node status, and margin status was performed with log-rank analysis. OS and DFS were analyzed by the Kaplan-Meier method, and significance was
determined by log-rank. Multivariate analysis was per-

645

formed with the Cox proportional hazards model to examine the effects of CA 19-9 group accounting for positive univariate predictors. All data were analyzed with
use of the SAS system (SPSS, Cary, NC).
RESULTS
Demographics
There were 59 males and 70 females, with an average
age of 67 years (range, 41 to 85 years); 114 (88%)
underwent pancreaticoduodenectomy, 4 (3%) underwent
distal pancreatectomy, and 11 (8%) underwent subtotal
or total pancreatectomy. On the basis of the American
Joint Committee on Cancer Staging System,10 patients
were pathologically staged as follows: 38 were stage I,
15 were stage II, 70 were stage III, and 6 were stage IV.
Tumor stages included 20 patients with T1 lesions, 48
with T2 tumors, 56 patients with T3 lesions, and 5 with
pathologic T4 tumors. Lymph nodes were positive in 78
patients (61%). Finally, margins were considered positive in 71 (55%) of the patients who underwent resection.
Several patients also underwent induction treatment with
chemotherapy and external beam radiotherapy (XRT);
33 patients received gemcitabine plus XRT; 13 received
5-FU, mitomycin C, and XRT as part of a phase II trial
at FCCC; 4 patients received 5-FU combined with XRT;
and 5 patients received other combinations (including
carboplatin and taxol). Finally, 80 patients underwent
some type of adjuvant therapy with 5-FU or gemcitabine,
with or without radiation therapy.
Prognostic Factors
Several factors were analyzed to determine whether
these impacted survival within these groups. Even
though there were a higher percentage of stage 3 and 4
tumors in groups 1 and 2 (71% vs. 51%), stage itself had
no impact on survival (P .384). Neoadjuvant chemotherapy and radiation were frequently used in this group
of patients, with 55 (43%) receiving treatment with this
modality. Only one patient in group 1 received induction
therapy, but otherwise these patients were evenly distributed between the groups; this modality had no impact on
survival (P .278). Adjuvant therapy was also frequently used, with 63% of patients receiving adjuvant
treatment. Patients receiving adjuvant therapy were
evenly distributed between the four CA 19-9 groups, and
this modality had no impact on survival (P .671).
Margin status was also evaluated as a possible factor
effecting survival. Overall, 71 patients (55%) had margins that were considered to be positive or close (1
mm). These patients were evenly distributed throughout
the CA 19-9 groups (43%, 55%, 52%, and 60%). By
Ann Surg Oncol, Vol. 11, No. 7, 2004

646

A. C. BERGER ET AL.

univariate analysis, margin status showed a trend toward

significance but did not reach statistical difference (P
.093), with median survival being 22 months for those
with negative margins and 17 months for patients with
positive margins. Median DFS for patients with negative
margins was 15 months, compared with 10 months for
those with positive margins (P .059).
Lymph nodes were positive in the majority of patients
in this series, with 78 patients (61%) having positive
lymph nodes on final pathologic examination. It is interesting that six patients (86%) in the nonsecretor group
had positive lymph nodes. In the other three groups,
lymph nodes were positive in 60%, 59%, and 60% of
patients. In this analysis, lymph node status was a significant prognostic indicator by both univariate and multivariate analysis. The median and 5-year survivals of
patients with negative lymph nodes were 29 months and
16%. Patients with positive lymph nodes had a median
survival of 17 months and 5-year survival of 8% (P
.015 on univariate analysis and P .0022 by multivariate analysis). However, when disease-free survival was
analyzed, lymph node positivity did not reach statistical
significance. Patients with negative lymph nodes had a
median DFS of 18 months, compared with 11 months for
those with negative lymph nodes (P .061).
CA 19-9 Analysis
CA 19-9 levels ranged from 0 to 16,300 U/mL; patients were divided into four groups based on their preoperative CA 19-9 level. Group 1 consisted of 7 patients
(5%) and included those who were CA 19-9 nonsecretors; these patients always had an undetectable CA 19-9
level. Group 2 was composed of 21 patients (16%)
whose CA 19-9 level at diagnosis was within the normal
range (37 U/mL). Groups 3 and 4 encompassed patients with CA 19-9 levels 200 or 200; these groups
contained 44 patients (34%) and 57 patients (44%), respectively. Analysis of survival in these four groups
showed that the median and 5-year survivals for groups
1 and 2 were significantly improved in comparison with
groups 3 and 4 (P .0026, Table 1, Fig. 1). There were
no significant differences between the groups when patients were evaluated for recurrence. Even though the

FIG. 1. Overall survival of patients with resected pancreatic cancer in

relation to preoperative CA 19-9 value. Group 1 includes all patients
with undetectable CA 19-9 levels, group 2 includes all patients with a
normal CA 19-9 value (37 U/mL), group 3 contains patients with CA
19-9 values 200 U/mL, and group 4 is all patients with a preoperative
CA 19-9 level 200 U/mL. For these patients, P .005 on multivariate analysis with the Cox proportional hazards model.

median and 5-year DFS were longer in groups 1 and 2

than in groups 3 and 4, these levels did not reach statistical significance (P .232, Table 2, Fig. 2).
By multivariate analysis, CA 19-9 grouping continued
to have prognostic significance when other variables
were taken into account, such as positive lymph nodes
(P .005, Table 3). Patients with CA 19-9 levels 200
U/mL had the worst prognosis and were four times more
likely to die than those with undetectable CA 19-9 levels
(P .006). In addition, patients with CA 19-9 levels up
to 200 U/mL had a hazard ratio of 3, compared with
groups 1 and 2 (P .042). Finally, patients with positive
lymph nodes were twice as likely to die as those with
negative lymph nodes, no matter what their CA 19-9
level was (P .002).
When the Cox proportional hazards models were used
to evaluate multivariate predictors of DFS, there were no
significant factors. However, the CA 19-9 group did
have a definite impact on DFS, when margin and lymph
node status were taken into account. Patients in groups 2,
3, and 4 had a hazard ratio for recurrence of 1.6, 2.6, and
2.5, respectively, when compared to patients with undetectable CA 19-9 (P .125).
TABLE 2. Disease-free survival by CA 19-9 grouping

TABLE 1. Overall survival by CA 19-9 grouping

CA 19-9 value (U/ml)

Median
survival
(mo.)

Undetectable
37
38200
200

7
21
44
57

32
35
22
16

Ann Surg Oncol, Vol. 11, No. 7, 2004

5-year
survival
(%)
20
34
11
2

CA 19-9 value (U/ml)

Median
DFS
(mo.)

Undetectable
37
38200
200

7
21
44
57

29
14
12
12

DFS, disease-free survival.

5-year
DFS
27%
27%
0%
0%

UNDETECTABLE CA 19-9 LEVELS IN PANCREATIC CANCER

FIG. 2. Disease-free survival of patients after resection of pancreatic

adenocarcinoma in relation to preoperative CA 19-9 value. Group 1
includes all patients with undetectable CA 19-9, group 2 includes all
patients with a normal CA 19-9 value (37 U/mL), group 3 contains
patients with CA 19-9 values (200 U/mL), and group 4 is all patients
with a preoperative CA 19-9 values 200 U/mL. For these patients, P
.125 on multivariate analysis with the Cox proportional hazards
model.

DISCUSSION
Carbohydrate antigen (CA) 19-9 is a tumor-associated
antigen found on the cell surface of many gastrointestinal
tumors. It was first described by Koprowski et al.,1 who
immunized mice with a colorectal carcinoma cell line
and then isolated the CA 19-9 monoclonal antibody from
mouse splenocytes. Further research has shown that CA
19-9 is normally present in salivary mucus and exocrine
pancreatic secretions.10,11 Although it originated from a
colorectal carcinoma cell line, CA 19-9 has found its
greatest utility in the diagnosis and evaluation of patients
with pancreatic adenocarcinoma.
Safi et al.12 prospectively evaluated the utility of CA
19-9 in patients with pancreatic cancer versus those with
benign diseases. They found that 92% of patients with
pancreatic adenocarcinoma had a CA 19-9 level 37
U/mL, and 77% had levels 120 U/mL.12 The median
CA 19-9 level in patients with pancreatic cancer was 528
TABLE 3. Effect of variable on survival by univariate and
multivariate analysis
Variable

Univariate

Multivariate

Hazard ratio

Stage
Margins
Neoadjuvant therapy
Adjuvant therapy
Positive Lymph nodes
CA 19-9 grouping
Compared to group 1

0.384
0.093
0.278
0.671
0.015
0.003

n/a
n/a
n/a
n/a
0.002
0.005

n/a
n/a
n/a
n/a
1.988
group 21.35
group 33.03
group 44.44

647

U/mL, compared with 18, 14, and 7 U/mL for patients

with chronic pancreatitis, acute pancreatitis, and benign
surgical diseases, respectively. In another group of patients with colorectal, gastric, esophageal, or extragastrointestinal malignancies, the majority of patients had
CA 19-9 levels in the normal range.12 In this series, the
overall sensitivity, specificity, and accuracy were 92%,
85%, and 82%.
Several investigators have demonstrated the effectiveness of serial CA 19-9 measurements in following the
course of patients with pancreatic adenocarcinoma. One
of the earliest studies was at the National Cancer Institute
in 1986. These investigators found that a return to normal
CA 19-9 level after resection was associated with longer
survival than among patients whose CA 19-9 level never
returned to normal.2 They also found that elevation of
CA 19-9 to 95 U/mL or a fourfold increase from the
lowest postoperative level predicted disease progression.
Other investigators found that secondary elevations of
CA 19-9 in the postoperative period preceded detectable
CT scan or clinical examination changes by 2 to 9
months.13
In another study, Montgomery et al.4 found that postoperative CA 19-9 measurements were the best predictors of disease-free and overall survival. They found that
patients whose CA 19-9 levels returned to normal by 3 to
6 months had a longer DFS and OS than those whose
levels did not. In addition, the survival of patients whose
postoperative CA 19-9 level was 180 U/mL at 1 to 3
months was similar to that of patients whose values
normalized at 3 to 6 months.
Finally, CA 19-9 has been a useful predictor of response in patients with advanced pancreatic cancer undergoing chemotherapy with gemcitabine-containing
regimens. In one study, the investigators found a significant decline in CA 19-9 levels in patients achieving a
radiographic complete response and a decrease in CA
19-9 levels in those patients who had a partial response.14
Another study showed that a decrease of 20% of the
baseline CA 19-9 level after 8 weeks of treatment resulted in a significantly better median survival. This CA
19-9 response was the strongest independent predictor of
survival by multivariate analysis.6
Since the time that it was discovered, investigators
have shown that the CA 19-9 antigen is also a sialylated
Lea blood group antigen.7 Magnani et al.15 discovered
that the antigen is found in pancreatic tissue and salivary
mucin from most normal individuals belonging to the
Le(ab-) or Le(a-b) blood group and is not found in
salivary mucin from normal individuals belonging to the
Le(a-b-) blood group. About 5% to 7% of the population
belong to the Le(a-b-) blood group and thus lack the
Ann Surg Oncol, Vol. 11, No. 7, 2004

648

A. C. BERGER ET AL.

fucosyltransferase that catalyzes the synthesis of the antigen detected by the CA 19-9 monoclonal antibody. As
a consequence, Koprowski has hypothesized that patients
with a Le(a-b-) phenotype should be unable to synthesize
CA 19-9 and thus not express it in their secretions.16
However, there have been reports of some exceptional
Le(a-b-) patients who demonstrated positive CA 19-9
values.17 Unfortunately, the Le blood group phenotype
can be mistyped in patients with conditions such as
pregnancy, alcoholic cirrhosis and pancreatitis, hydatid
cysts, and cancer.18 This is probably because the absorption of Le-activated glycolipids to RBCs is inhibited in
these patients, and the standard hemagglutination test
results in a false-negative Le grouping. One possible
explanation for this is that anti-Le antibodies are elevated
in the sera of patients with cancer; these antibodies may
perturb the glycolipid uptake by RBCs.19 Additionally, in
many of these conditions, the serum lipoprotein level is
elevated, and the Le antigen may be adsorbed onto the
lipoprotein, resulting in a reduced amount of antigen and
thus a false-negative Le(a-b-).19 This is the reason why
some Le(a-b-) cancer patients are not genuinely Le-negative and thus have an elevated CA 19-9 value.20 These
false-negative Le(a-b-) patients usually become Le-positive after surgical resection and thus can be followed
with regard to CA 19-9 values.
Because of this mistyping phenomenon, the Le-negative phenotype is more common among cancer patients
(20%) than among healthy individuals (8%). Individuals who are genuinely Le-negative and genetically
lack the enzyme never have a positive serum CA 19-9
value. This was clearly demonstrated in two different
studies in Japan and Denmark.20,21 It is important, therefore, to note that patients who do not secrete CA 19-9 are
always Le-negative. On the other hand, Le-negative individuals may have elevated CA 19-9 levels in the conditions described above.
There have been very few studies reported in the
literature concerning the prognosis of patients who are
CA 19-9 nonsecretors. One would hypothesize that these
patients may do worse because we lose the ability to
monitor their levels and thus predict response to therapy
and disease progression or recurrence. However, we
have demonstrated in this study that these patients do just
as well as patients with normal CA 19-9 levels at diagnosis. The median survivals of these two groups of
patients were 32 and 35 months, respectively. Additionally, these patients had a statistically significantly better
survival than patients with elevated CA 19-9 levels. This
factor was significant by both univariate and multivariate
analysis.
Ann Surg Oncol, Vol. 11, No. 7, 2004

In fact, patients with undetectable CA 19-9 had a

longer median disease-free survival than all other
groups27 months versus 14, 10, and 10, respectively
although this did not reach statistical significance. The
only other positive predictor of overall survival by multivariate analysis was lymph node positivity, which has
been seen in previous studies.22 Unfortunately, there
were only seven patients in our nonsecretor groups,
which comprised 5% of our study population and thus
fits in with the frequency of Le(a-b-) patients in the
population. Therefore, it is difficult to make any definitive conclusions based on this small number of patients.
It is unclear why patients with undetectable CA 19-9
levels had better prognoses despite their lymph node
positivity. It may be that Le-negativity is a marker for
decreased metastatic potential or aggressiveness. However, these speculations will have to be borne out in
larger studies before any conclusions can be drawn.
Future studies will attempt to investigate the significance
of an undetectable CA 19-9 level in the context of larger,
multiinstitutional studies for which larger patient samples are available.
REFERENCES
1. Koprowski H, Steplewski Z, Mitchell K, et al. Colorectal carcinoma antigens detected by hybridoma antibodies. Somatic Cell
Genetics 1979;5:95771.
2. Glenn J, Steinberg WM, Kurtzman SH, et al. 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.
3. Sperti C, Pasquali C, Catalini S, et al. CA 19-9 as a prognostic
index after resection for pancreatic cancer. J Surg Oncol 1993;52:
137 41.
4. Montgomery RC, Hoffman JP, Riley LB, et al. Prediction of
recurrence and survival by post-resection CA 19-9 values in patients with adenocarcinoma of the pancreas. Ann Surg Oncol
1997;4:551 6.
5. Safi F, Schlosser W, Falkenreck S, Beger HG. Prognostic values of
CA 19-9 serum course in pancreatic cancer. Hepatogastroenterology 1998;45:2539.
6. Halm U, Schumann T, Schiefke I, et al. Decrease of CA 19-9
during chemotherapy with gemcitabine predicts survival time in
patients with advanced pancreatic cancer. Br J Cancer 2000;82:
1013 6.
7. Magnani J, Nilsson B, Brockhaus M, et al. A monoclonal antibodydefined antigen associated with gastrointestinal cancer is a ganglioside containing sialylated lacto-N-fucopenteose II. J Biological
Chem 1982;257:143659.
8. Ritts RE, Pitt HA. CA 19-9 in pancreatic cancer. Surg Oncol Clin
North Am 1998;7:93100.
9. Lamerz R. Role of tumour markers, cytogenetics. Ann Oncol
1990;10(Supp. 4):1459.
10. Brockhaus M, Wysocka M, Magnani JL, Steplewski Z, Koprowski
H, Ginsburg V. Normal salivary mucin contains the gastrointestinal cancer associated antigen detected by monoclonal antibody CA
19-9 in the serum mucin of patients. Vox Sang 1985;48:34 8.
11. Kalthoff H, Kreiker C, Schmiegel WH, Greten H, Thiele HG.
Characterization of CA 19-9 bearing mucins as physiological exocrine pancreatic secretion products. Cancer Res 1986;46:36057.

UNDETECTABLE CA 19-9 LEVELS IN PANCREATIC CANCER

12. Safi F, Beger HG, Bittner R, Buchler M, Krautzberger W. CA 19-9
and pancreatic adenocarcinoma. Cancer 1986;57:779 83.
13. Tian F, Appert HE, Myles J, Howard JM. Prognostic value of
serum CA 19-9 levels in pancreatic adenocarcinoma. Ann Surg
1992;215:350 5.
14. Heinemann V, Schermuly MM, Stieber P, Schulz L, Jungst D,
Wilkowski R, Schalhorn A CA 19-9: A predictor of response in
pancreatic cancer treated with gemcitabine and cisplatin. Anticancer Res 1999;19:2433 6.
15. 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.
16. Koprowski H, Brockhaus M, Blaszczyk M, Magnani J, Steplewski
Z, Ginsburg V. Lewis blood type may affect the incidence of
gastrointestinal cancer. Lancet 1982;1:13323.
17. Kawa S, Oguchi H, Kobayashi T, et al. Elevated serum levels of
Dupan-2 in pancreatic cancer patients negative for Lewis blood
group phenotype. Br J Cancer 1991;64:899 902.

649

18. Yazawa S, Nishihara S, Iwasaki H, et al. Genetic and enzymatic

evidence for Lewis enzyme expression in Lewis-negative cancer
patients. Cancer Res 1995;55:1473 8.
19. Hirano K, Kawa S, Oguchi H, et al. Loss of Lewis antigen
expression on erythrocytes in some cancer patients with high
serum CA 19-9 levels. JNCI 1987;79:1261 8.
20. Narimatsu H, Iwasaki H, Nakayama F, et al. Lewis and secretor
gene dosages affect CA 19-9 and DU-PAN-2 serum levels in
normal individuals and colorectal cancer patients. Cancer Res
1998;58:512 8.
21. Orntoft TF, Vestergaard EM, Holmes E, et al. Influence of Lewis
13/4-L-fucosyltransferase (FUT3) gene mutations on enzyme
activity, erythrocyte phenotyping, and circulating tumor marker
sialyl-Lewis a levels. J Biol Chem 1996;271:32260 8.
22. Lim JE, Chien MW, Earle CC. Prognostic factors following curative resection for pancreatic adenocarcinoma: a population-based,
linked database analysis of 396 patients. Ann Surg 2003;237:74
85.

Ann Surg Oncol, Vol. 11, No. 7, 2004