| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Studies |
Digestive Diseases Branch, National Institute of Diabetes and Digestive and Kidney Diseases (J.S., S.U.G., P.L.P., F.G., R.T.J.), and Laboratory of Pathology, National Cancer Institute (I.A.L.), National Institutes of Health, Bethesda, Maryland 20892
Address all correspondence and requests for reprints to: Dr. Robert T. Jensen, Digestive Diseases Branch, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Building 10, Room 9C-103, 10 Center Drive, MSC 1804, Bethesda, Maryland 20892-1804. E-mail: robertj{at}bdg10.niddk.nih.gov
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In contrast to many nonendocrine tumors such as colorectal cancer and pancreatic cancer, the molecular pathogenesis of gastrinomas and other gastrointestinal neuroendocrine tumors remains largely unknown (7). Alterations of the common oncogenes or tumor suppressor genes (p53 and retinoblastoma) are uncommon in gastrinomas or other gastrointestinal neuroendocrine tumors (2, 7, 8, 9, 10). One to 40% of different pancreatic endocrine tumors and a small percentage (<1%) of gastrointestinal carcinoids are associated with the autosomal dominant syndrome, multiple endocrine neoplasia type 1 (MEN1) (1, 2, 5, 11). MEN1 has recently been shown to be caused by defects in a 10-exon gene on chromosome 11q13 that encodes for a 610-amino acid nuclear protein, MENIN (12, 13, 14). Mutations in the MEN1 gene have recently been reported in 16–42% of sporadic gastrointestinal neuroendocrine tumors (15, 16, 17, 18). Therefore, in the majority of gastrointestinal neuroendocrine tumors the molecular pathogenesis is unknown.
Recently, inactivation of p16INK4a/CDKN2A (p16INK4a), a tumor suppressor gene located on chromosome 9p21, has been reported in a number of neuroendocrine malignancies, including pancreatic cancers (19). The p16INK4a gene encodes for p16, an inhibitor of cyclin-dependent kinase 4, a cyclin-dependent kinase that, when active, inhibits cell progression at the G1S junction by phosphorylating Rb to its inactive form (19, 20, 21). The mechanisms of p16INK4a gene inactivation include homozygous deletions, mutation and hypermethylation of specific CpG islands in the promoter of the p16INK4a gene (19). Whether p16INK4a gene alterations are a frequent event in gastrointestinal neuroendocrine tumors is at present controversial and unclear. One study (22) reported homozygous deletions and/or hypermethylation of the p16INK4a gene in 92% of a small number of gastrinomas and nonfunctional tumors; however, another study (23) reported no homozygous deletions of the p16INK4a gene locus in 28 human gastrointestinal neuroendocrine tumors. Furthermore, recent studies in nonendocrine tumors have shown that the presence of a p16INK4a gene alteration and/or loss of the p16 protein can be an important determinant of prognosis, recurrence, tumor aggressiveness, and survival in some tumors (24, 25, 26, 27, 28).
The purpose of the present study was therefore to determine whether genetic alterations occurred in the p16INK4a gene in gastrinomas and to define the type of gene alteration. Furthermore, because all of these patients have long-term follow-up (mean, 5 yr postresection) with yearly reassessments of tumor growth and biological behavior, it was possible to correlate the presence or absence of a p16INK4a gene alteration with tumor growth, tumor biological behavior, and known prognostic factors (tumor size, curability, presence of lymph and/or liver metastases, and primary tumor location).
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Forty-four patients who underwent exploratory laparotomy for Zollinger-Ellison syndrome at the NIH between 1990 and 1998 were included in this study. The study protocol was approved by the clinical research committee of the NIDDK, and all patients gave informed consent. Preoperative measurement of basal acid output and maximal acid output were performed as previously described (29). Preoperative serum gastrin levels were analyzed by RIA by Bioscience Laboratories (New York, NY) or Mayo Clinic Laboratories (Rochester, MN). The diagnosis of Zollinger-Ellison syndrome was established as previously reported (30). The duration of disease was defined by the clinical history from the time of diagnosis or from the time of disease onset as previously described (6). Detailed conventional imaging studies [computed tomography (CT) with oral and iv contrast, magnetic resonance imaging (MRI), ultrasound, and bone scan], selective abdominal angiography with secretin stimulation and hepatic vein gastrin sampling, and somatostatin receptor scintigraphy (SRS) were performed as previously reported (31, 32, 33) to locate the primary tumor and evaluate the extent of disease. All patients underwent an exploratory laparotomy with an extensive intraoperative evaluation for attempted curative resection as described previously (31, 34, 35). The patients were then reassessed within 2 weeks of surgery and 3–6 months postoperatively to determine disease-free status and annually to monitor progression of disease as previously described (30). Disease-free status was defined by normal fasting gastrin levels (<200 pg/mL), negative results on gastrin provocative testing with secretin (<200 pg/mL increase) and calcium (<395 pg/mL increase), and no evidence of tumor on any imaging study (30, 31, 36). Patients were classified as to whether they were disease free immediately postoperatively as well as whether they were disease free at the last follow-up. In those patients that were not disease free postoperatively, annual detailed imaging studies (CT, MRI, ultrasound, and SRS) and, if the results were unclear, selective angiography provided the basis for assessment of tumor growth or progression. Consistent absence of imaged lesions or lack of increase in size or number of lesions over the follow-up period was defined as a tumor not demonstrating growth. An increase in size or number of lesions on imaging studies was defined as evidence of tumor growth as described previously (37). The development of liver metastases identified by imaging studies during follow-up served as the definition for the liver metastases group (4, 37). Liver metastases were confirmed by percutaneous liver biopsy.
Tumor cell lines were obtained from the American Type Tissue Collection
(Manassas, VA). Four nonsmall cell lung cancer cells (SK-LU-1
adenocarcinoma cells, SK-MES-1 lung cancer, A-427 lung cancer, and
SW900 squamous cell lung cancer) (38) and three breast
cancer cell lines (MDA-MB-231, MCF-7, and Hs578t) (39),
which are reported to have homozygous deletions of the
p16INK4a gene, were used as positive controls for
the semiquantitative PCR-based deletion analysis. DU145 prostate
adenocarcinoma cells, which are reported (40) to have a
missense mutation (codon 76, GAC
TAC) in exon 2, and Calu-3 nonsmall
cell lung adenocarcinoma cells, which are reported to have a mutation
in codon 83, exon 2, resulting in a His to Tyr substitution
(38), were used as positive controls for sequencing. All
cancer cells were grown in Eagle’s MEM supplemented with 10% FCS,
except Hs578t, SW900, and MDA-MB-231, which were grown in DMEM with
10% calf serum.
Tumor samples were immediately snap-frozen in liquid nitrogen either during surgery or after harvesting and stored at -70 C. The histological diagnosis of gastrinoma was made using standard histological and immunohistological stains (chromogranin A and gastrin). Tumors from the two patients with MEN1 included in the present study were both from the duodenum, and both demonstrated gastrin and chromogranin staining. Tumor genomic DNA was extracted from 8-µm cryosections of the gastrinomas after analyzing an adjacent slide with hematoxylin and eosin staining to determine that 80% or more of the section contained tumor tissue. Cancer cell line genomic DNA was extracted from cell pellets after centrifugation (9000 x g for 10 min). Tumor DNA was extracted using a commercial kit (QiAmp Blood Kit, QIAGEN, Santa Clarita, CA). Germline DNA was extracted from leukocytes of these patients using the same kit.
Semiquantitative PCR-based deletion analysis
Primers for amplification of exon 2 of the
p16INK4a gene were designed according to the
method reported by Chen et al. (41). Primers
for amplification of an internal control marker, hypoxanthine-guanine
phosphoribosyl transferase (HPRT), were designed as described
previously (22). PCR was carried out with and without
[
-32P]deoxy (d)-CTP in a final volume of 25
µL with 5–10 ng DNA and 0.5 µU DNA polymerase (AmpliTaq Gold,
Perkin-Elmer Corp., Foster City, CA) in a DNA thermal
cycler (Perkin-Elmer Corp., 9600 Thermocycler) according
to the protocol described previously (22). The number of
PCR cycles per primer set was adjusted to the signals generated within
the linear range. PCR products were resolved at 270 V during 30 min in
10% polyacrylamide-Tris-Borate-EDTA gels (Novex,
San Diego, CA). The bands were visualized with ethidium bromide and
excised with a razor blade. The gel slices were placed in 10 mL
scintillation fluid, and the incorporated radioactivity was counted in
a 2500 TR liquid scintillation analyzer (Packard, Downers Grove, IL).
Similar gel slices from DNA-absent PCR reactions were used to assess
background. The ratio of radioactive incorporation into the p16 exon 2
fragment (numerator) with the corresponding incorporated into the HPRT
fragment (denominator) was calculated for each gastrinoma using the
protocol described previously (22). p16 exon 2/HPRT
radioactivity incorporation ratios were also determined from leukocyte
genomic DNA obtained from 20 of the patients.
Automatic sequencing of p16INK4a exons 1 and 2
The sequences of exons 1 and 2 of the p16INK4a gene were analyzed in 24 tumors. PCR fragments from exons 1 and 2 were amplified using primers as described previously (22). Amplified exons were processed as recommended and sequenced twice using an ABI PRISM dRodamine Terminator Cycle Sequencing Ready Reaction Kit and an ABI Prism 377 Automatic Sequencer (Perkin-Elmer Corp.).
Methylation-specific PCR of a CpG island in the p16INK4a promoter
Genomic DNA from all 44 tumors and from 38 leukocyte samples were evaluated for the presence of hypermethylation of CpG in position +167 of the p16INK4a gene promoter as described previously (42). Briefly, the chemical modification of cytosine to uracil by bisulfite treatment provides a method for the study of DNA methylation that avoids the use of restriction enzymes (43). In this reaction, cytosines are converted to uracil, but methylated cytosines are not converted to uracils. The altered DNA can be amplified with primers specific to the modified resulting sequence depending upon whether the primers are complementary to modified methylated or unmethylated sequences in the target DNA region. Sodium bisulfite-modified genomic DNA was amplified with methylated and unmethylated specific primers according to the conditions described previously (42).
Statistics
The 
2 test, Fisher’s exact test, and
Mann-Whitney rank test were used to compare the effects of genetic
alterations of the p16INK4A gene on clinical and
laboratory parameters and tumor characteristics. P <
0.05 was considered significantly different.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
). This cohort is similar to other large
series of patients with Zollinger-Ellison syndrome with respect to a
slight male overrepresentation (64%), mean age of patients (48 yr),
duration of disease from onset until the surgery date (9.3 yr), mean
serum gastrin, and preoperative basal acid and maximal acid output
values (5, 44, 45, 46) (Table 1). The number of patients with
MEN1 is less than normally seen in all patients with gastrinoma (4.5%
vs. 20%, respectively) (5, 47), because in our
institution these patients do not routinely undergo exploratory
laparotomy unless tumors are at least 3 cm in diameter (47, 48). Inhibition of gastric acid hypersecretion at the time of
surgery was achieved predominantly (91%) with
H+,K+-adenosine
triphosphatase inhibitors. In contrast to older series (5, 49, 50, 51), but similar to other recent studies (3, 34, 52), 43% of all identified primary gastrinomas were located in
the duodenum, and only 20% were found in the pancreas (Table 1).
|
). In contrast, 18% of patients had
metastases only, with 11% having liver metastases found at surgery and
7% having metastases found only in lymph nodes (Table 1).
The 44 gastrinomas were evaluated for the presence of homozygous
deletions, mutations, and 5'-CpG promoter methylation of the
p16INK4a gene. The analysis of the 5'-CpG island
methylation at position +167 (the location of the 5'-nucleotide of the
sense primer in relation to the major transcriptional start site) of
the promoter of the p16INK4a gene, using
methylation-specific PCR (42) showed methylation in 23 of
the 44 (52%) gastrinomas (Fig. 1 and
Table 2). To assess whether homozygous
deletions were occurring in the p16INK4a gene, we
performed semiquantitative PCR in which the patterns of
[-32P]dCTP incorporation into the PCR-
amplified p16INK4a gene and that incorporated
into the housekeeping gene HPRT were determined, as previously
described (22). Results from 13 patients are shown in Fig. 2, and the counts per min incorporated
into each PCR product and the ratio for each patient are shown in Table 3. To determine the normal range, we performed a similar analysis on
genomic DNA extracted from the white blood cells (WBC) of 20 of the
patients with gastrinoma. Results from 5 representative patients are
shown in Fig. 3 for both gastrinomas and
WBC DNAs. The p16INK4a to HPRT ratio in the
normal WBC DNA ranged from 0.7–2.7 (Table 3) with a mean ±
SEM of 1.43 ± 0.13. Either using the WBC range as a
normal range or tumor ratios below the mean - 2 SD
found in the 20 WBC samples (0.3), none of the 44 gastrinoma samples
had reduced p16INK4a gene copy number, which
would be seen if a homozygous deletion had occurred (22).
The results were similar regardless of whether the ratio between the
p16INK4a and HPRT genes was calculated using
[-32P]dCTP incorporation or using optical
density reading of the amplified DNA bands after agarose
electrophoresis and ethidium bromide staining (Table 3 and Figs. 2 and 3). The fact that this semiquantitative PCR-based method could reliably
detect homozygous deletions of the p16INK4a gene
if it was present is shown in Fig. 4 with
nine cancer cell lines. Seven of these cancer cell lines are known to
have homozygous deletions of the p16INK4a gene
[SK-LU-1, SK-MES-1, A-427, SW900 nonsmall cell lung cancer cells
(38), and MDA-MB-231, MCF-7, Hs578t breast cancer cells
(39)], and they had a mean p16/HPRT ratio of 0.17 ±
0.028 (range, 0.09–0.3; Fig. 4). In contrast, two cancer cell lines
with mutations in the p16INK4a gene without
deletions [Calu-3 lung cancer cells (38) and DU-145
prostate cancer cells (40)] had p16/HPRT ratios well
above these values (Fig. 4).
|
|
|
|
|
|
ccA; PP)
was found that did not result in an amino acid change (Table 3). In each of the 13 patients with a
polymorphism in codon 41 the sequence was heterozygous. To verify that
p16INK4a gene mutations could be deleted if
present with the methods used in this study, two cancer cell lines
[DU-145 prostate cancer cells (40) and Calu-3 (nonsmall
cell lung cancer cells (38)] reported to have mutations
in the p16INK4a gene were analyzed in the same
manner. The reported (40) missense mutation (codon 76,
GACTAC) in exon 2 of the p16INK4a gene in
DU-145 cells was confirmed as well as the mutation in codon 83, exon 2
of Calu-3 nonsmall cell lung cancer cells, resulting in a His to Tyr
substitution (38).
To determine whether methylation of the 5'-CpG island of the of
p16INK4a gene correlated with tumor growth or
other tumor characteristics of gastrinomas, as reported in some
nonendocrine malignancies (39, 53, 54, 55, 56, 57, 58, 59), gastrinoma growth,
primary tumor location, tumor extent, tumor size, and ability to render
patients disease free postresection were correlated with tumor
methylation status (Table 2).
Methylation of the promoter of the p16INK4a gene
did not correlate with the tumor growth pattern. Specifically, the
percentage of methylation found in gastrinomas that demonstrated no
growth over a follow-up period of 5.0 ± 0.4 yr (range, 1.2–11.7
yr) was similar to that found in gastrinomas that demonstrated growth
during follow-up (P = 0.33) and in gastrinomas that
demonstrated an aggressive growth pattern with the patient developing
liver metastases (Table 2). Similarly, a lack of effect of methylation
status on tumor aggressiveness was supported by finding a similar
percentage of gastrinomas demonstrating methylation of the
p16INK4a gene 5'-CpG island in patients with
different tumor extents at surgery (i.e. primary only or
primary with lymph node involvement or with liver metastases; Table 2).
The frequency of methylated p16INK4a gene in
pancreatic gastrinomas with lymph node metastases compared with the
frequency of methylation in gastrinomas in other nonpancreatic
locations with lymph node metastases almost reached significance [4 of
4 (100%) vs. 5 of 13 (38%); P = 0.053].
Furthermore, tumor size, which has been correlated to gastrinoma’s
metastatic potential (4, 6, 60), did not correlate to p16
methylation status, and the percentage of gastrinomas 1 cm or larger
and the percentage of those less than 1 cm in size with or without p16
methylation (P = 0.37; Table 2) were not significantly
different. Gastrinomas in different locations (4, 6, 61)
are reported to vary in aggressiveness, with pancreatic gastrinomas
more frequently associated with liver metastases. The percentage of
pancreatic gastrinomas with p16 methylation did not differ from that of
duodenal gastrinomas (P = 0.29). Furthermore, the
percentage of duodenal gastrinomas demonstrating methylation of
p16INK4a gene did not differ from that of lymph
node primary gastrinomas or primary gastrinomas in other locations
(P = 0.60; Table 2).
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Recent studies involving small numbers of patients with sporadic gastrointestinal neuroendocrine tumors report that loss of heterozygosity (LOH) of the MEN1 locus at 11q13 occurs in 36–93% of sporadic gastrinomas (15, 67, 68), in 16–50% of insulinomas (15, 68), and in 0–78% of sporadic gastrointestinal carcinoid tumors (17, 69, 70, 71), whereas mutations in the MEN1 gene occur in 27–39% of sporadic gastrinomas (15, 16, 18, 72) and 17% of insulinomas (15). Genetic alterations of the DPC4/Smad4 tumor suppressor gene were described in 55% of nonfunctional PETs, but were not found in other PETs (i.e. insulinomas, gastrinomas, and vasoactive intestinal polypeptide-secreting tumors) (73). Furthermore, genomic-wide alletopying (23) in 28 gastrointestinal neuroendocrine tumors provided evidence for allele deletions on 3q, 11p, 16p, and 22q, suggesting that important tumor suppressor loci could exist in these areas where inactivation could contribute to their pathogenesis. The results of the above studies suggest that in the majority of cases the molecular pathogenesis of gastrointestinal neuroendocrine tumors is still unknown.
Inactivation of the p16INK4a tumor suppressor gene located on 9p21 has been demonstrated in a large number of malignancies (19, 74). In many tumors the frequency of inactivation of the p16INK4a is exceeded only by the frequency of p53 inactivation (20). Inactivation of the p16INK4a suppressor gene has been demonstrated in a great variety of nonendocrine gastrointestinal malignancies, including cancers of the esophagus (53, 54, 75), stomach (55, 56), pancreas (57), bile duct (58), colon (39, 59), and liver (76, 77). In recent studies inactivation of p16INK4a has been reported in a few neuroendocrine tumors, including parathyroid (78), thyroid (79, 80), small cell and nonsmall cell lung cancer (81, 82, 83, 84), pancreas (22), and pituitary (85, 86, 87, 88). Recently, p16INK4a gene alterations were reported (22) in four of four nonfunctional pancreatic endocrine tumors and seven of eight gastrinomas, including hypermethylation of the 5'-CpG island (50%) and homozygous deletions (42%). These results suggested that genetic alterations in the p16INK4a could be a frequent event in pancreatic endocrine tumors and perhaps an important event in their molecular pathogenesis.
The present study was designed to answer this question by determining both the frequency and mechanism as well as the possible clinical consequences of p16INK4a gene alterations in a large number of gastrinomas, which are the most common symptomatic, malignant, pancreatic endocrine tumor (2, 5). With some nonendocrine tumors inactivation of the p16INK4a gene and/or the level of p16 protein expression are associated with tumor growth, early recurrence, and/or poor prognosis (24, 25, 26, 27, 28, 89), whereas in others alterations in the p16INK4a gene are not (24, 90). Because the patients included in this study had long-term follow-up (mean, 5 yr), it was possible to correlate the presence or absence of p16INK4a gene alterations to subsequent tumor growth.
Our results demonstrate that p16INK4a gene alterations occur in 52% of gastrinomas and therefore are the most frequent genetic alterations described to date in these tumors (7, 11, 15, 18, 23). No homozygous deletions or p16INK4a gene mutations were found, and all of the p16INK4a gene alterations were due to methylation of a 5'-CpG island of the promoter of the p16INK4a gene. These results have both similarities and differences from those reported previously in both endocrine and nonendocrine tissues. Our results differ from those recently reported in 1 study (22) involving a small number of nonfunctional PETs and gastrinomas, in which homozygous deletions of the p16INK4a gene were reported in 37% (3 gastrinomas) and 50% (2 nonfunctional PETs), respectively. This difference was not due to methodological differences in the 2 studies, because the same semiquantitative PCR analysis was used for the deletion of homozygous deletion in both studies. Furthermore, in the present study the failure to detect homozygous deletions was not due to either the method of quantitation used or the controls used.
In the present study homozygous deletions were assessed by determining
the ratio of [-32P]dCTP incorporated into
either p16INK4a or HPRT, to control for genomic
input. Identical results were obtained using PCR without radioactivity
and calculating the ratio by assessing ethidium bromide staining
intensity. The control values of the assay for homozygous deletion are
probably reliable, because the control range of the ratio without
homozygous deletions was determined from the same patients’ WBC
genomic DNA, processed and analyzed in an identical manner to that of
the gastrinoma DNA. Contamination by nontumoral DNA can interfere with
the determination of homozygous deletions; therefore, in the present
study semiquantitative analyses were only performed on DNA from tissue
sections in which the gastrinomas comprised more than 80% of the
sample by histological analysis. Our failure to detect homozygous
deletions in gastrinomas are compatible with results from a recent
study (23) assessing genomic-wide allelotyping in 28 human
pancreatic endocrine tumors, including 7 gastrinomas in which no
deletions were found at the 9p21 locus, which contains the
p16INK4a gene. Our results, finding 13 patients
with a polymorphism in codon 41 and none with LOH at this site, provide
further support for lack of a deletion in 1 or both alleles. More
extensive LOH studies need to be performed to exclude small deletions
on 1 allele. However, inactivation of tumor suppressor genes typically
involves a large deletion of 1 of the 2 alleles (23).
Our results, demonstrating no mutations in the p16INK4a gene, but a high rate (52% of the gastrinomas) of hypermethylation of the p16INK4a gene promoter, are similar to those reported for a few other endocrine tumors and some nonendocrine tumors. Specifically, in 4 nonfunctional pancreatic endocrine tumors and 8 gastrinomas (22), no mutations were found; however, methylation of the p16INK4a gene promoter was found in 50% (2 nonfunctional PETs) and 58% (5 gastrinomas). In studies of pituitary adenomas (85, 86, 87, 88), 83–90% had p16INK4a gene promoter methylation, but either no mutations or infrequent mutations were detected in the p16INK4a gene. In two studies of follicular and papillary thyroid cancer (80, 91) no homozygous deletions were found, and mutations were rare; however, 30% had methylation of the p16INK4a gene. Similar findings were found in other gastrointestinal cancers, such as colorectal cancer (39, 59) and hepatocellular cancer (76, 77). In contrast in malignancies such as pancreatic cancer and nasopharyngeal cancers, p16INK4a gene deletions, mutations, as well as hypermethylation are frequently found (19). In numerous tumors, evidence shows that p16INK4a promoter hypermethylation per se can lead to transcriptional silencing without an accompanying p16INK4a deletion or mutation detected in either allele (19, 39, 40, 92, 93, 94). This conclusion is supported by the observation that treatment with the demethylating agent, 5- deoxyazacytidine, can reactivate p16INK4a transcription (19, 40, 92). The present study does not provide any data on the exact mechanism of how this transcriptional inactivation occurs. Our studies do not discriminate whether one or both alleles are methylated. Furthermore, multiple CpG sites are present in the p16INK4a promoter region that might be methylated, and whether all are detected by the primers used is unknown (92, 94).
Gastrinomas, similar to carcinoid tumors and other pancreatic endocrine tumors, are not homogenous in their biological behavior, location, or growth behavior (1, 2, 3, 4, 60). Gastrinomas differ in primary tumor location (duodenum, pancreas, or other) (5, 34) and clinical and biological behavior, with 25% pursuing an aggressive growth pattern, and 75% demonstrating either no growth or an indolent growth pattern with long-term follow-up (4, 6). Even the biological behavior of metastatic gastrinoma to the liver in different patients varies, with 30% demonstrating no growth, 30% slow growth, and 40% aggressive growth during a 5- to 6-yr follow-up period (37). At present it is not possible to predict the different biological phenotypes in a given patient with gastrinoma, nor is the molecular basis for the differences in biological behavior known. In previous studies in non-Hodgkin’s lymphomas (27, 28), leukemias (24), melanomas (25, 26, 89), nonsmall cell lung cancer (82), and gliomas (95), the presence of mutations, deletions, or methylation of the p16INK4a gene and/or loss of the p16INK4a protein resulted in a worse prognosis, earlier recurrence, and/or more aggressive growth behavior. In contrast, in some studies p16INK4a gene alterations and/or protein expression have not had prognostic value, including human lymphomas and leukemias (24) and head and neck carcinomas (90). There is only one prior study (88) of p16INK4a gene mutation and/or protein expression in a neuroendocrine tumor (i.e. pituitary adenomas) of its prognostic value to compare to the results of our findings in the present study. In that study (88) methylation of p16 was present in invasive and noninvasive nonfunctional adenomas in a similar percentage of tumors. In our study the presence of p16INK4a gene alterations (i.e. methylation in 52%) did not correlate with measures of biological behavior of the gastrinoma (fasting gastrin level or basal or maximal acid output, which assess the magnitude and effect of biologically active gastrin), with the presence of factors known to have important prognostic value for the development of metastases (tumor size, curability, and primary tumor location) (4, 5, 6, 60, 61), or with growth behavior (development of metastases and postresection growth pattern) (2, 4, 5, 6). The failure to detect a correlation of tumor growth behavior with the presence of a p16INK4a gene alteration was unlikely to be due to other factors that might alter the ability to detect such an association. Such factors include the slow growth behavior of many gastrinomas, inadequate numbers of patients in the different subgroups analyzed, or failure to detect significant changes in tumor extent, size, or growth in some patients. The mean follow-up time in the present study postresection was 5 yr, with a maximal up to 11.7 yr, which is sufficiently long to assess tumor growth in patients with gastrinomas (4, 6, 37). Furthermore, all patients postresection underwent yearly imaging with at least three imaging modalities (ultrasound, CT scan, and MRI), and since 1994 all patients also had yearly SRS with [111In-diethylenetriamine pentaacetic acid,Phe (1)]octreotide with single photon emission computed tomography imaging, which is at present the most sensitive modality to detect gastrinoma extent (31, 33, 96). Because patients with the most aggressive forms of the disease who present with diffuse liver metastases usually do not undergo surgical exploration (34, 97), and there is usually insufficient tumor tissue available from the percutaneous biopsies, they are not included in this type of study. Therefore, the possibility exists that this subset could differ in their frequency of p16 gene mutations. Lastly, gastrinomas from 44 patients were included in the present study, and 23 patients (52%) demonstrated a p16INK4a gene alteration. Therefore, tumor characteristics could be compared between approximately equal groups of patients with and without p16INK4a gene alteration.
The fact that methylation of the p16INK4a gene promoter occurred in gastrinomas independently of tumor size, the presence or absence of regional metastases to lymph nodes, or the presence or absence of distant metastases to the liver suggests that alterations in p16 gene function are probably an early event in gastrinoma tumorigenesis. A similar temporal location of p16 gene alterations has been proposed in the development of colorectal cancer in patients with ulcerative colitis (75) and in patients with pancreatic ductular adenocarcinoma (98). Furthermore, recent studies have proposed that duodenal and pancreatic gastrinomas have a different origin (99, 100). The fact that an equal percentage of both duodenal and pancreatic gastrinomas as well as primary gastrinomas from other locations had similar percentages of methylation of the p16INK4a gene promoter suggest that this process plays an important central role in the molecular pathogenesis of gastrinomas that is independent of the site of origin.
In conclusion, our study demonstrates that a p16INK4a gene alteration occurs in 52% of gastrinomas. In all cases the p16INK4a gene alteration was hypermethylation of the p16INK4a gene promoter, with no p16INK4a gene mutations or homozygous deletions detected. The presence or absence of hypermethylation of the p16INK4a gene promoter did not correlate with clinical characteristics, biological behavior of the tumor, tumor location, or growth pattern of the gastrinoma. The occurrence of an equal percentage of p16INK4a gene promoter methylation in gastrinomas with different tumor extents supports the conclusion that this is an early event in gastrinoma tumorigenesis. The equal occurrence in primary gastrinomas in different locations, which are now thought to have different cells of origin (99, 100), supports the conclusion that a p16INK4a gene alteration is a central event in the molecular pathogenesis of a proportion of gastrinomas.
Received October 21, 1999.
Revised April 5, 2000.
Accepted July 20, 2000.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
X174 Hae fragments were used as molecular weight
standards (shown on the left). Patient numbers are
identical to those listed in Table 3
