Hypergastrinemia Promotes Adenoma Progression in the APC^Min−/+ Mouse Model of Familial Adenomatous Polyposis¹

Gastrin has been described as a growth factor for colorectal malignancy for well over a decade, and it is known to mediate proliferation via both endocrine and autocrine/paracrine mechanisms(1, 2, 3).

Elevated serum gastrin levels can be a feature of a number of clinical conditions, including ZES³(4, 5, 6, 7) and PA (8), but possibly more importantly as a result of infection with Helicobacter pylori (9, 10) and after the administration of proton pump inhibitors (11, 12).

In humans, increased proliferation of colonic crypts has been confirmed in patients with PA and ZES (13). However, the majority of epidemiological studies performed to date in patients with sustained hypergastrinemia have failed to show an enhanced risk of colorectal cancer. In patients with PA, although conflicting studies have been reported, the majority show that colorectal cancer incidence is not increased (14, 15). This has also been shown in patients with ZES (16). This failure may be attributed to lack of control of confounding factors, such as H. pylori in the control groups. A recent study, in which H. pyloriinfection was controlled for, showed that hypergastrinemia is associated with a 3.3-fold increase in the relative risk of developing colorectal cancer (17). It is conceivable,therefore, that hypergastrinemia may promote progression through the adenoma-carcinoma sequence. This hypothesis is strengthened by the preliminary findings of Smith and Watson (18) that human adenomas of all sizes and grades express an isoform of the cholecystokinin B/gastrin receptor.

Thus, the aim of the present study was to evaluate the effect of serum hypergastrinemia induced by omeprazole, a proton pump inhibitor that increases serum gastrin levels 2–4-fold (11, 12, 19), on colonic neoplastic progression in the multiple intestinal neoplasia APC^Min−/+ mouse model of APC.

APC^−/+ mice were bred within the Cancer Studies Unit at the University of Nottingham under Home Office Project License No. PPL 4011454. Heterozygous male APC^Min−/+mice were bred with wild-type C57/Bl6 female mice. The offspring were characterized for the Min genotype by PCR as described previously (20).

After results from the PCR analysis were obtained,APC^Min−/+-positive mice were randomized into each experimental group on an ongoing basis because of the limited number of mice positive for the Min mutation that are generated from established breeding colonies. Mice were randomized into groups on a litter basis with unequal numbers being a result of the distribution of Min positive and negative mice.

An initial study was performed to determine the effect of hypergastrinemia on tumor burden and survival of APC^Min−/+ mice. APC^Min−/+ mice 4 weeks of age were recruited to the following groups: group 1, oral vehicle for omeprazole [PBS (pH 7.2)]; group 2, omeprazole (75 mg/kg in a single oral dose; n = 10).

Tumor size was measured by image analysis by use of Leica Qwinn image processing and analysis system run on a Leica Q5001W PC.

APC^Min−/+ mice 4 weeks of age were recruited to the following groups: group 1, oral vehicle for omeprazole + control immunogen (n = 20); group 2,omeprazole (75 mg/kg in a single oral dose) + control immunogen (n = 20); group 3, omeprazole + Gastrimmune (n = 16).

Gastrimmune is a gastrin immunogen composed of the NH₂-terminal nine amino acids of gastrin-17 linked via a peptide spacer to diphtheria toxoid. Treatment with Gastrimmune generates antibodies that neutralize the amidated and glycine-extended forms of gastrin-17 (Aphton Corporation, CA;Ref. 21). A rat-specific form of the immunogen was formulated at 2 mg/ml and injected s.c. into mice in 100-μl volumes at week 4, and at three-week intervals thereafter. The rat sequence used within the immunogen differs only by one amino acid from the mouse sequence, and antibodies raised by rat Gastrimmune previously have been shown to neutralize mouse gastrin (data not shown). Control mice receiving immunogen constituents only were given the formulation without the active peptide.

APC^Min−/+ mice 4 weeks of age were recruited to the following groups: group 1, vehicle + control immunogen(n = 22 mice); group 2, omeprazole (75 mg/kg) + control immunogen (n = 30 mice); group 3, omeprazole + Gastrimmune(n = 30 mice); group 4, vehicle + Gastrimmune (n = 18 mice).

Gastrimmune was injected 2 weeks prior to omeprazole. (The latter was administered from week 6 when antibodies against Gastrimmune were known to be detectable.)

At the end-stage of their disease, APC^Min−/+mice bleed from their multiple polyps, giving rise to anemia, which was initially confirmed by hematocrit measurements [median hematocrit, 0.136 (range, 0.058–0.153) compared with 0.385 (range,0.381–0.391] in wild-type C57/Bl6 mice. Twelve days after the onset of anemia, mice were terminally anesthetized by a 300-μl injection containing 0.315 ng/ml Hypnorm and 5 ng/ml Hypnovel (Jannsen-Wickham,Bucks, United Kingdom, and Roche-Lewis, East Sussex, United Kingdom) by an experienced, independent observer who was blinded to the treatment groups. Anemia previously has been validated as a consistent end point for APC^Min−/+ mouse termination (20). After termination, the GI tracts of the APC^Min−/+ mice were removed, and the numbers of visible tumor nodules in both the small and large bowel were counted with the aid of a dissecting microscope. The mucosa was fixed in formal calcium (40% formaldehyde, 10% CaCl₂ in distilled water) prior to embedding in paraffin wax for histological analysis.

The United Kingdom Coordinating Committee for Cancer Research Guidelines were adhered to throughout all animal experimentation.

One h before termination, mice received i.p. injections of 10 mg/kg BrdUrd (Sigma). After paraffin-embedding, 4-μm sections were stained with H&E to confirm correct orientation of the crypts and with an antibody directed against BrdUrd (mouse monoclonal, used at 1:500 dilution; Dakopatts).

Two experienced observers, blinded to the treatment groups, counted the number of positive nuclei in the crypts from both the small and large intestine. Ten whole crypts from different areas of the colonic mucosa and the small intestinal mucosa of each mouse were identified where possible. The number of total positive nuclei within the epithelial cells was counted, together with the total number of cells per crypt. This was repeated with sections from 4–11 animals. The total number of cells per crypt in the large or small intestine were consistent both within and between experimental animals.

For scoring of the polyps, each section from each mouse was cut at the same depth (2 mm), and each polyp was analyzed by choosing a representative low-power field and by using image analysis to assess the area of BrdUrd staining as a percentage of the total tumor tissue. For assessment of the proliferation of adenoma and carcinoma samples that were taken from the large and small intestines, image analysis was performed by the use of the Leica Qwin image processing and analysis system. BrdUrd staining was specifically measured and expressed as a function of total tumor tissue area. Intraobserver variation, in which different fields of the same section were assessed, was 5.6%, and interobserver variation was 7.5%.

The GI tracts from end-stage APC^Min−/+ mice treated with either vehicle or omeprazole, as detailed in “Experiment 2,” were fixed in formal calcium, processed into paraffin, and stained with H&E. A consultant pathologist assessed the sections in a blinded manner detailing architectural detail and degree of dysplasia as well as necrotic and vascular appearance.

Prior to termination of the APC^Min−/+ mice, the animals were fasted overnight. Blood was collected by cardiac puncture,and serum was prepared by centrifugation after clotting at 4°C. The serum concentrations of total amidated gastrin were measured using the L2 antiserum and also antiserum directed against progastrin and glycine-extended gastrin, as described previously (22), or levels were measured by the use of an ELISA-based kit (Invitrogen,Newcastle, United Kingdom).

To measure “antibody-bound” serum gastrin concentrations in mice immunized with Gastrimmune, serum was collected as described above. The antibody:gastrin complexes were immunoprecipitated using newborn calf serum (Sigma) and 25% polyethylene glycol (8000; Sigma), with a final newborn calf serum-polyethylene glycol-test serum ratio of 1:4:5. The mixture was vortexed, and precipitates were pelleted by centrifugation at 460 × g for 30 min at 4°C. The precipitate was then resuspended in 200 μl of 0.02 m veronal buffer (pH 8.4; 4.12 g/liter sodium barbitone, 0.5 g/liter sodium azide, 1.65 g/liter serum BSA; all reagents from Sigma). The tubes containing the recovered antibody:antigen complexes were then tightly capped and placed in a boiling water bath for 5 min. Recovered gastrin was then measured from a standard gastrin-17 curve prepared with charcoal-stripped pooled mouse serum.

Mouse-specific primers were designed to assess classical CCKBR. The primers MCK1 (5′-CCCTCCTCAACAGCAGTAGC-3′) and MCK2(5′-GGGTGATTCGAATGGTCAAC-3′), specific for exons 1 and 2, respectively,were used to amplify cDNA prepared from tumor and nonmalignant mucosa from both large and small intestine. The 50-μl PCR reactions were set up using 3 μl of sample cDNA in 1× PCR Buffer (PE Applied Biosystems), 40 μm deoxynucleotide triphosphates, 20μ m each primer, and 1 unit of AmpliTaq Gold (PE Applied Biosystems). The PCR cycle consisted of a 95°C activation stage for 10 min followed by 40 cycles of 94°C for 30 s, 60°C for 60 s, and 72°C for 30 s. After a final extension stage of 72°C for 10 min, the PCR products were separated on a 2% (w/v)agarose gel and stained with ethidium bromide. The mouse colon cancer cell line MC26 was used as a positive control. A negative control was also included that was not reverse transcribed and confirmed the absence of genomic DNA.

RNA was extracted from tissue and reverse transcribed from random hexamer primers (Pharmacia) using Supercript reverse transcriptase(Life Technologies, United Kingdom). Real-time PCR was performed using the 5700 Sequence Detection System (PE Applied Biosystems,Warrington, United Kingdom). Each PCR was performed according to manufacturer’s instructions using 1 μl of sample cDNA in a 25-μl reaction volume. The reaction buffer was prepared from the SYBR Green PCR Core Kit (PE Applied Biosystems) and consisted of 1× SYBR Green PCR Buffer (PE Applied Biosystems), 3 mmMgCl₂, 0.2 mm dATP, 0.2 mm dCTP, 0.2 mm dGTP, 0.4 mm dUTP,0.25 units of AmpErase UNG (PE Applied Biosystems), and 0.625 units of AmpliTaq Gold (PE Applied Biosystems). The murine CCKBRprimers MCK1 and MCK2 and the GAPDH primers (GAPU,5′-GGTGAAGGTCGGAGTCAACGGA-3′; GAPL, 5′-GAGGGATCTCGCTCCTGGAAGA-3′)were included in parallel reactions at a final concentration of 100 nm each. The fluorescence of the SYBR Green dye bound to the CCKBR and GAPDH PCR products was measured after each cycle by the 5700 System, and the cycle number was recorded when the accumulated signal crossed an arbitrary threshold (Ct value). The relative gene expression for each sample was determined using the formula 2^{Ct(GAPDH) − Ct(CCKBR)}, and reflected CCKBR gene expression normalized to GAPDH levels.

Chilled methanol (0.4 ml) was added to 0.1 ml of serum. After mixing,samples were centrifuged at 2000 × g for 10 min at 4°C. The supernatant was then analyzed by high-performance liquid chromatography together with standard samples prepared by adding known amounts of omeprazole to control serum. The high-performance liquid chromatography chromatographic conditions were as follows: Prodigy OGS (250 × 4.5 mm) column; mobile phase, 42% acetonitrile in 7.5 mm diammonium hydrogen phosphate (Sigma); flow rate, 1.0 ml/min; UV detection at 304 nm; injection volume, 0.05 ml; analysis time, 30 min; detection time,6.3 min. The linearity of the standard curve was confirmed over a sample concentration of 1–20 μg/ml (r = 0.99).

The measurement of antibodies raised against Gastrimmune was performed according to a method described previously (21).

Survival analysis was performed by a log-rank test using a multiple group comparison, and the in vitro results were compared using a one-way ANOVA, a Student’s t test, or the Mann-Whitney test with the Minitab statistical package.

Elevation in serum gastrin concentrations was achieved by administration of the proton pump inhibitor omeprazole with the dose chosen so that it reflected the elevation of serum gastrin seen in patients on a maintenance dose of omeprazole (80 pm; Refs.15, 16). In an initial pilot study, mice were dosed with 12.5, 37.5, and 75.0 mg/kg of omeprazole daily by oral gavage for a period of 7 days and were fasted prior to cardiac puncture. Mice treated with vehicle had a median serum total amidated gastrin concentration of 15.5 pm. When omeprazole was dosed at 12.5 mg/kg, the median serum amidated gastrin concentrations increased to 36.5 pm compared with medians of 52.5 and 102.0 pm in mice treated with omeprazole doses of 37.5 and 75 mg/kg, respectively. In the group treated with 37.5 mg/kg, 4 of 16 mice achieved a serum amidated gastrin concentration of 80.0 pmor greater compared with 12 of 16 in the 75 mg/kg omeprazole-treated group. The highest dose of omeprazole was therefore chosen for the in vivo studies, which generated median serum omeprazole levels of 5.82 μg/ml (interquartile range, 1.77–17.3; n = 10 mice).

No glycine-extended gastrin or progastrin was detected by RIA in either C57/Bl6 or Balb/C mouse strains after omeprazole treatment.

The results are shown in Table 1. There was no increase in either small or large intestinal tumor number at termination (Table 1). However, when mean total tumor size(cumulative size of large and small tumors within either the small or large intestine) was assessed, there was a significant increase in large (2.3-fold; P < 0.05) but not small intestinal tumor load. Survival was significantly reduced in the omeprazole group (mean survival of 9.6 weeks compared with 13.9 weeks in the vehicle control-treated group; P < 0.0001). Tumor burden/time also revealed significant differences between both the small and large intestine (1.46- and 2.6-fold increases, respectively, in the omeprazole-treated groups; P < 0.025 for both).

The first therapy study involved immunization with Gastrimmune at the same time as treatment with omeprazole (both at week 4). An initial study was performed to determine the time of onset of antigastrin antibody titers induced by Gastrimmune immunization, and by day 14,antibodies were measurable in the serum (data not shown).

Omeprazole, together with administration of the control immunogen, was shown to significantly reduce the survival of APC^Min−/+ mice compared with the vehicle control (P = 0.00001, log-rank test) with a median 50% survival of 10 weeks compared with 13 weeks (Fig. 1). When Gastrimmune immunization was combined with omeprazole treatment,there was a increase in the median 50% survival to 14 weeks, which was significantly different from omeprazole alone (P = 0.0002, log-rank test) but not significantly different from the vehicle control (P = 0.2602, log-rank test).

In the second study, mice were preimmunized with Gastrimmune at week 4,and omeprazole treatment was delayed until week 6. This allowed Gastrimmune-induced antibodies to be generated before serum hypergastrinemia was initiated. Fig. 2 shows the results of this study. Omeprazole significantly reduced survival (P = 0.0038, log-rank test), with a median 50% survival of 11 weeks, despite a treatment delay. The vehicle group had a median 50% survival of 13 weeks, as did the group receiving the Gastrimmune-omeprazole combination, and there was no significant difference between these two groups (P = 0.1103, log-rank test). However, the survival of the group treated with the Gastrimmune-omeprazole combination was significantly increased from the group treated with omeprazole alone(P = 0.0002, log-rank test). Gastrimmune alone was also shown to significantly increase survival when compared with the vehicle control (P = 0.0017) with a 50% median survival of 19 weeks.

In the Gastrimmune + vehicle-treated group, terminal antibody levels had a mean absorbance of 0.447 (SD, 0.43) compared with a mean of 0.330 (SD, 0.016) for the mice treated with Gastrimmune + omeprazole, a 26% reduction, which was significant(P < 0.02, Student’s t test). Animals treated with either vehicle or omeprazole only had no detectable antibody-specific absorbance.

Antibody-bound gastrin levels were measured after processing of sera as detailed in “Materials and Methods.” Animals receiving Gastrimmune + vehicle had a mean bound gastrin level of 18.35 pmol/ml (SD, 2.32 pmol/ml), which approximates the fasting serum gastrin levels measured previously. In the Gastrimmune + omeprazole-treated mice, the bound serum gastrin levels had increased to 55.6 pmol/ml (SD, 9.65 pmol/ml), a 3-fold significant increase over the levels shown in the Gastrimmune + vehicle-treated mice(P < 0.0001).

To determine adenoma proliferation at earlier stages of disease, a third study was initiated in which APC^Min−/+mice were treated with the same treatment regimes as study 2 but sacrificed at week 12. The number of tumors macroscopically visible in the small and large bowel was counted and the LI measured.

Measurement of small intestinal tumors numbers revealed no significant difference between any of the groups. With large intestinal tumor number, there was a 2.5-fold increase, from 0.88 in the vehicle control-treated group to 2.22 in the omeprazole-treated group(P = 0.030). This appeared to be partially reduced when omeprazole was combined with Gastrimmune to give a mean number of 1.65, which was not significantly different from the vehicle control (P = 0.075). The group treated with Gastrimmune alone had a mean large intestinal tumor number of 0.76, a 14% reduction, which again was not significantly different from the vehicle control.

The LI as measured by BrdUrd uptake from the same study is shown in Table 1. Omeprazole treatment significantly increased the LI of normal large intestine (53%; P < 0.020) and large intestinal tumors (29%; P < 0.025) compared with the vehicle control. Histological assessment of the large intestinal crypts after omeprazole treatment revealed a distortion in their shape attributable to the increased proliferation together with an increase of the proliferative zone toward the top of the crypt (Fig. 3,A). This was not observed in the large intestinal crypts from vehicle-treated APC^Min−/+ mice (Fig. 3,A). There was also an obvious increase in BrdUrd incorporation in large intestinal adenomas/carcinomas. Fig. 3 B shows a polypoid carcinoma from a control mouse and an omeprazole-treated mouse at the end-stage of disease progression. The LI of the small intestinal normal mucosa (from lower and upper regions)was not significantly affected by omeprazole treatment, whereas that of small intestinal tumors was increased (35% increase; P < 0.05).

Gastrimmune treatment significantly reduced the LI of the normal large intestinal mucosa (20% reduction; P < 0.050) and large intestinal tumors (42% reduction; P < 0.002). Gastrimmune had no effect on the growth of normal small intestinal mucosa, but significantly reduced the growth of tumors from the small intestine (22% reduction; P < 0.025).

Gastrimmune and omeprazole in combination significantly inhibited the LI increase induced by omeprazole treatment in all tissues responding to omeprazole (significance levels for normal and malignant large intestinal mucosa and small intestinal tumors are shown in Table 2).

When Gastrimmune and omeprazole were coadministered, the LI was significantly reduced compared with the vehicle control in large intestinal normal and malignant mucosa and in small intestinal normal but not malignant mucosa (significance levels shown in Table 2).

In most groups in which Gastrimmune significantly inhibited the mucosal LI, the difference was nonsignificant when compared with the respective Gastrimmune-omeprazole combination. However, in large normal intestinal mucosa, there was a significant 30% reduction in the LI after Gastrimmune-omeprazole treatment compared with the LI in the mucosa of the group treated with Gastrimmune alone (P < 0.010).

Polyps in both vehicle control- and omeprazole-treated groups showed moderate dysplasia without significant pseudostratification of epithelial cells. Small polyps involved few glands and appeared as sessile tubular adenomas, with very little villous formation. As adenomas enlarged, they became polypoid with surface ulceration and vascular stroma.

Subtle changes in dysplasia between adenomas was difficult to quantify in an objective manner. However, no gross changes in dysplasia were evident between the two groups.

CCKBR was expressed at the gene level on polyps from both the small and large intestine. Fig. 4 shows a gel in which a specific 99-bp band is shown for the positive control cell line, MC26, together with polyps and normal mucosa from the large and small intestines. CCKBR mRNA levels were measured in both large and small intestinal polyps with and without treatment with omeprazole. There was a significant 6-fold increase in CCKBR mRNA levels in hypergastrinemic mice (P < 0.001;Fig. 5).

The role of serum hypergastrinemia in colorectal carcinogenesis remains controversial; most epidemiological studies show that hypergastrinemia does not increase colonic tumor incidence(14, 15, 16), but does increase colonic mucosal proliferation(13). Past studies evaluating the effect of hypergastrinemia on tumor incidence in humans did not control for confounding factors such as H. pylori in the control population. In a recent study in which H. pylori infection was taken into account, a temporal relationship between elevated gastrin levels and development of colorectal malignancy was shown(17).

What is not known is how hypergastrinemia affects premalignant change in the colon. This question was addressed in the present study by use of the APC^Min−/+ mouse model of familial adenomatous polyposis. A state of hypergastrinemia was induced by the proton pump inhibitor, omeprazole, such that serum gastrin concentrations in the mice were elevated to the same level as patients on a maintenance dose of omeprazole (23, 24). The variations in serum gastrin concentrations were large and may have included a postprandial element, despite fasting, because mice are known to be coprophagic. The high omeprazole dose may have been necessary because, unlike rats, which are acutely sensitive to omeprazole in terms of serum gastrin elevation, mice are more refractory, which may relate to enterochromaffin-like cell density and differences in omeprazole metabolism (25).

Omeprazole-induced hypergastrinemia significantly decreased the survival of APC^Min−/+ mice. Examination of the normal and malignant mucosa from both the small and large bowels of these mice revealed an enhanced proliferation in the tumors from both areas. Thus, omeprazole appeared to increase proliferation of the tumors arising within the GI mucosa, which caused the terminal stage of the disease to be reached at an earlier time point. This was confirmed in the first experiment when tumor burden/time was significantly enhanced in both small and large intestinal adenomas from omeprazole-treated mice. In mice terminated at 12 weeks of age,omeprazole had increased the macroscopic tumor number 2.5-fold. This is possibly a result of enhanced proliferation, resulting in earlier visualization of the tumors rather than an effect on tumor incidence. Histological evaluation of the adenomas revealed that smaller lesions were sessile tubular adenomas, becoming polypoid with increasing size. All showed moderate dysplasia, and after treatment with omeprazole,there was no gross change in the dysplasia exhibited by the lesions. Thus, in the time frame of these experiments, an increase in proliferation rather than malignant potential was observed.

In the present study, large but not small bowel mucosal crypts were shown to respond proliferatively in omeprazole-treated mice, which may reflect the differential sensitivity that the GI tract has to gastrin peptides (26). In the study by Thorburn et al.(17), the association between gastrin and malignancy was more closely associated in rectal cancer, which lends support to the finding that gastrin has a greater proliferative role in the distal GI tract. In the present study, the large intestinal crypts from omeprazole-treated APC^Min−/+ mice had a distorted structure and an increase in the proliferative zone from the lower third to the whole length of the crypt. This has been shown in patients with an increased risk of colon cancer (27, 28).

The effect on survival was confirmed as gastrin-mediated by use of the gastrin immunogen, Gastrimmune, which produces antibodies that neutralize both amidated and glycine-extended gastrin-17 with high affinity (21). Antibodies raised by Gastrimmune do not bind cholecystokinin or other gastrin species such as G34 and pentagastrin. Antigastrin antibodies reversed the reduced survival observed in omeprazole-treated APC^Min−/+ mice. Antibodies raised by Gastrimmune were shown to be effective in binding serum gastrin. Antibody-bound gastrin levels were increased in mice treated with omeprazole compared with the vehicle-treated mice. This resulted in a significant reduction in free Gastrimmune-induced antibodies in the omeprazole-treated group. When Gastrimmune was administered at the same time as omeprazole, the decreased survival appeared to be partially reversed at the late stages of the study. This is likely a result of the timing of production of Gastrimmune-induced antibodies, which began to be detectable from week 2 after the initial immunization. Thus, mice had a period of elevated serum gastrin in the initial stages, which could account for only a partial reversal in survival. Omeprazole administration was therefore delayed for a 2-week period to allow anti-Gastrimmune antibodies to be generated, which resulted in the attainment of complete reversal of survival in the Gastrimmune-omeprazole-treated group compared with the mice treated with vehicle.

This study does not delineate the gastrin species that may be inducing the proliferative action on the GI tract because Gastrimmune inhibits only amidated and glycine-extended G17. Omeprazole induced amidated but not glycine-extended gastrin species in the mouse strain used in this study. CCKBR gene and protein expression of the APC^Min−/+ tumors has been reported previously(29, 30). After omeprazole treatment, classical CCKBR gene expression was increased by intestinal polyps. The role of CCKBR in mediating the effects of hypergastrinemia in APC^Min−/+ mice is being tested at present by the use of specific CCKBR antagonists.

Gastrimmune alone had a significant effect in enhancing survival of the APC^Min−/+ mice. This indicates that the tumors may have responded in an endocrine manner to basal gastrin levels. Antibody-bound gastrin was indeed confirmed to be present in normogastrinemic APC^Min−/+ mice. It previously has been shown that the gastrin gene is activated in normal mucosa and tumors in the APC^Min−/+ mouse model and that gastrin immunoreactivity can be detected (31). If the gastrin is increasing proliferation in an autocrine manner, this may partially explain the increased LI in the normal GI mucosa of APC^Min−/+ mice compared with Min-negative mice (31). Gastrin gene expression and progastrin and glycine-extended gastrin previously have been shown to be expressed in polyps arising within both small and large intestinal mucosa of mice with a mutated APC gene (29, 32). Because antibodies raised by Gastrimmune neutralize glycine-extended G17, the antibodies may therefore partially inhibit autocrine pathways established by the tumors. This is relevant because a number of recent publications have shown that in mice in which there is overexpression of precursor gastrin molecules, there is a predisposition to aberrant colonic crypt foci and adenomas and carcinomas in response to a chemical carcinogen (31, 33).

In conclusion, gastrin may enhance progression to malignancy when premalignant changes have occurred in the large and small intestines. The findings of these studies highlight an urgent need to investigate the possibility that gastrin may increase progression of existing premalignant colorectal lesions in a clinical setting.

We would like to acknowledge T. Morris, Unit Manager, Academic Unit of Cancer Studies, for performing the in vivo studies,Dr. P. D. James, Pathologist, for histological evaluations of the APC^Min−/+ intestinal sections, Dr. D. McWilliams for performing the molecular biology on the APC^Min−/+ mouse samples, and P. Clarke for performing the image analysis and the BrdUrd staining.