Donanemab (LY3002813) Phase 1b Study in Alzheimer's Disease: Rapid and Sustained Reduction of Brain Amyloid Measured by Florbetapir F18 Imaging

BACKGROUND

Donanemab (LY3002813) is an IgG1 antibody directed at an N‑terminal pyroglutamate of amyloid beta epitope that is present only in brain amyloid plaques.

OBJECTIVES

To assess effects of donanemab on brain amyloid plaque load after single and multiple intravenous doses, as well as pharmacokinetics, safety/tolerability, and immunogenicity.

DESIGN

Phase 1b, investigator- and patient-blind, randomized, placebo-controlled study.

SETTING

Patients recruited at clinical research sites in the United States and Japan.

PARTICIPANTS

61 amyloid plaque-positive patients with mild cognitive impairment due to Alzheimer's disease and mild-to-moderate Alzheimer's disease dementia.

INTERVENTION

Six cohorts were dosed with donanemab: single dose 10-, 20- or 40- mg/kg (N = 18), multiple doses of 10-mg/kg every 2 weeks for 24 weeks (N = 10), and 10- or 20-mg/kg every 4 weeks for 72 weeks (N=18) or placebo (N = 15).

MEASUREMENTS

Brain amyloid plaque load, using florbetapir positron emission tomography, was assessed up to 72 weeks. Safety was evaluated by occurrence of adverse events, magnetic resonance imaging, electrocardiogram, vital signs, laboratory testing, neurological monitoring, and immunogenicity.

RESULTS

Treatment with donanemab resulted in rapid reduction of amyloid, even after a single dose. By 24 weeks, amyloid positron emission tomography mean changes from baseline for single donanemab doses in Centiloids were: -16.5 (standard error 11.22) 10-mg/kg intravenous; 40.0 (standard error 11.23) 20 mg/kg intravenous; and -49.6 (standard error 15.10) 40-mg/kg intravenous. Mean reduction of amyloid plaque in multiple dose cohorts by 24 weeks in Centiloids were: 55.8 (standard error 9.51) 10-mg/kg every 2 weeks; -50.2 (standard error 10.54) 10-mg/kg every 4 weeks; and -58.4 (standard error 9.66) 20-mg/kg every 4 weeks. Amyloid on average remained below baseline levels up to 72 weeks after a single dose of donanemab. Repeated dosing resulted in continued florbetapir positron emission tomography reductions over time compared to single dosing with 6 out of 28 patients attaining complete amyloid clearance within 24 weeks. Within these, 5 out of 10 patients in the 20 mg/kg every 4 weeks cohort attained complete amyloid clearance within 36 weeks. When dosing with donanemab was stopped after 24 weeks of repeat dosing in the 10 mg every 2 weeks cohort, florbetapir positron emission tomography reductions were sustained up to 72 weeks. For the single dose cohorts on day 1, dose proportional increases in donanemab pharmacokinetics were observed from 10 to 40 mg/kg. Dose proportional increases in pharmacokinetics were also observed at steady state with the multiple dose cohorts. Donanemab clearance was comparable across the dose levels. Mean donanemab elimination-half-life following 20 mg/kg single dose was 9.3 days with range of 5.6 to 16.2 days. Greater than 90% of patients had positive treatment-emergent antidrug antibodies with donanemab. However, overall, the treatment-emergent antidrug antibodies did not have a significant impact on pharmacokinetics. Donanemab was generally well tolerated. Amongst the 46 participants treated with donanemab, the following amyloid-related imaging abnormalities, common to the drug class, were observed: 12 vasogenic cerebral edema events (12 [19.7%] patients), 10 cerebral microhemorrhage events (6 [13.0%] patients), and 2 superficial siderosis events (2 [4.3%] patients).

CONCLUSIONS

Single and multiple doses of donanemab demonstrated a rapid, robust, and sustained reduction up to 72 weeks in brain amyloid plaque despite treatment-emergent antidrug antibodies detected in most patients. Amyloid-related imaging abnormalities were the most common treatment-emergent event.

Florbetapir F 18 amyloid PET and 36-month cognitive decline: a prospective multicenter study

This study was designed to evaluate whether subjects with amyloid beta (Aβ) pathology, detected using florbetapir positron emission tomorgraphy (PET), demonstrated greater cognitive decline than subjects without Aβ pathology. Sixty-nine cognitively normal (CN) controls, 52 with recently diagnosed mild cognitive impairment (MCI) and 31 with probable Alzheimer's disease (AD) dementia were included in the study. PET images obtained in these subjects were visually rated as positive (Aβ+) or negative (Aβ-), blind to diagnosis. Fourteen percent (10/69) of CN, 37% (19/52) of MCI and 68% (21/31) of AD were Aβ+. The primary outcome was change in ADAS-Cog score in MCI subjects after 36 months; however, additional outcomes included change on measures of cognition, function and diagnostic status. Aβ+ MCI subjects demonstrated greater worsening compared with Aβ- subjects on the ADAS-Cog over 36 months (5.66 ± 1.47 vs -0.71 ± 1.09, P = 0.0014) as well as on the mini-mental state exam (MMSE), digit symbol substitution (DSS) test, and a verbal fluency test (P < 0.05). Similar to MCI subjects, Aβ+ CN subjects showed greater decline on the ADAS-Cog, digit-symbol-substitution test and verbal fluency (P<0.05), whereas Aβ+ AD patients showed greater declines in verbal fluency and the MMSE (P < 0.05). Aβ+ subjects in all diagnostic groups also showed greater decline on the CDR-SB (P<0.04), a global clinical assessment. Aβ+ subjects did not show significantly greater declines on the ADCS-ADL or Wechsler Memory Scale. Overall, these findings suggest that in CN, MCI and AD subjects, florbetapir PET Aβ+ subjects show greater cognitive and global deterioration over a 3-year follow-up than Aβ- subjects do.

Alterations of directional connectivity among resting-state networks in Alzheimer disease

BACKGROUND AND PURPOSE

AD has been documented as a kind of disconnection syndrome by functional neuroimaging studies. The primary focus of this study was to examine, with the use of resting-state fMRI, whether AD would impact connectivity among RSNs.

MATERIALS AND METHODS

Fourteen patients with AD and 16 NC were recruited and scanned by using resting-state fMRI. Group independent-component analysis and the BN learning approach were used, respectively, to separate the RSNs and construct the network-to-network connectivity patterns for each group. The convergence index for the special network DMN was measured.

RESULTS

Three of the 4 connections were significantly lower in AD compared with NC. Although numerically the AD group had more connections, none was statistically different from that in the NC group except for 1 increased connection from the DMN to the DAN. The convergence index for the DMN node was lower in AD than in NC.

CONCLUSIONS

Connections among cognitive networks in AD were more vulnerable to impairment than sensory networks. The DMN decreased its integration function for other RSNs but may also play a role in compensating for the disrupted connections in AD.

Chronic divalproex sodium use and brain atrophy in Alzheimer disease

OBJECTIVE

We evaluated the effect of the divalproex sodium formulation of valproic acid on brain volumes using MRI in people with mild to moderate Alzheimer disease (AD) and assessed for changes associated with behavioral and cognitive effects.

METHODS

Eighty-nine of 313 participants randomized to divalproex or placebo in a 24-month, parallel-group trial received MRI scans at baseline and 12 months. Interval MRI annual percent changes in whole brain, ventricular, and hippocampal volumes were the primary outcomes of interest. Change from baseline in clinical outcomes was assessed at 6-month intervals.

RESULTS

There were no baseline differences between active treatment and placebo groups in age, education, brain volumes, clinical rating scores, or APOE ε4 carrier status. The group treated with divalproex showed a greater rate of decline in left and right hippocampal and brain volumes (-10.9% and -12.4% vs -5.6% and -6.3%, and -3.5% vs -1.4%, respectively), and a greater rate of ventricular expansion (24.5% vs 9.9%) (p < 0.001). Mini-Mental State Examination scores showed a more rapid decline with divalproex through month 12 (placebo = -2.0 ± 4.3, divalproex = -3.9 ± 4.0) (p = 0.037), although there were no changes on other cognitive, behavioral, or functional ratings at 12 and 24 months.

CONCLUSIONS

Divalproex treatment was associated with accelerated brain volume loss over 1 year and perhaps with greater cognitive impairment. The long-term clinical effects of these changes are not known.

Volumetric MRI vs clinical predictors of Alzheimer disease in mild cognitive impairment

OBJECTIVE

To compare volumetric MRI of whole brain and medial temporal lobe structures to clinical measures for predicting progression from amnestic mild cognitive impairment (MCI) to Alzheimer disease (AD).

METHODS

Baseline MRI scans from 129 subjects with amnestic MCI were obtained from participants in the Alzheimer's Disease Cooperative Study group's randomized, placebo-controlled clinical drug trial of donepezil, vitamin E, or placebo. Measures of whole brain, ventricular, hippocampal, and entorhinal cortex volumes were acquired. Participants were followed with clinical and cognitive evaluations until formal criteria for AD were met, or completion of 36 months of follow-up. Logistic regression modeling was done to assess the predictive value of all MRI measures, risk factors such as APOE genotype, age, family history of AD, education, sex, and cognitive test scores for progression to AD. Least angle regression modeling was used to determine which variables would produce an optimal predictive model, and whether adding MRI measures to a model with only clinical measures would improve predictive accuracy.

RESULTS

Of the four MRI measures evaluated, only ventricular volumes and hippocampal volumes were predictive of progression to AD. Maximal predictive accuracy using only MRI measures was obtained by hippocampal volumes by themselves (60.4%). When clinical variables were added to the model, the predictive accuracy increased to 78.8%. Use of MRI measures did not improve predictive accuracy beyond that obtained by cognitive measures alone. APOE status, MRI, or demographic variables were not necessary for the optimal predictive model. This optimal model included the Delayed 10-word list recall, New York University Delayed Paragraph Recall, and the Alzheimer's Disease Assessment Scale-Cognitive Subscale total score.

CONCLUSION

In moderate stages of amnestic mild cognitive impairment, common cognitive tests provide better predictive accuracy than measures of whole brain, ventricular, entorhinal cortex, or hippocampal volumes for assessing progression to Alzheimer disease.

Clinical predictors of progression to Alzheimer disease in amnestic mild cognitive impairment

OBJECTIVE

To investigate the neurocognitive measures that best predict progression from amnestic mild cognitive impairment (aMCI) to Alzheimer disease (AD).

METHODS

We evaluated 539 participants with aMCI from the Alzheimer's Disease Cooperative Study clinical drug trial of donepezil, vitamin E, or placebo. During the study period of 36 months, 212 aMCI participants progressed to AD. Using progression from aMCI to AD within 36 months as the dependent variable, a generalized linear model was fit to the data using the least absolute shrinkage and selection operator. Independent variables included in this analysis were age, sex, education, APOE-e4 (APOE4) status, family history of dementia, Mini-Mental State Examination score, Digits Backwards (Wechsler Memory Scale), Maze Time and Errors, Number Cancellation, Delayed Recall of Alzheimer's Disease Assessment Scale Word List, New York University Paragraph Recall Test (Immediate and Delayed), Boston Naming Test, Category Fluency, Clock Drawing Test, and the Alzheimer's Disease Assessment Scale-Cognitive subscale (ADAS-cog).

RESULTS

The model that best predicted progression from aMCI to AD over 36 months included APOE4 status, the Symbol Digit Modalities Test, Delayed 10-Word List Recall, New York University Paragraph Recall Test (Delayed), and the ADAS-cog total score. When APOE4 was removed from the analysis the resulting model had a similar estimated predictive accuracy as the full model. As determined by cross-validation, the estimated predictive accuracy of the final model was 80%.

CONCLUSION

Progression from amnestic mild cognitive impairment to Alzheimer disease in this cohort was best determined by combining four common, easily administered, cognitive measures.

Hypermethylation of the hMLH1 gene promoter is associated with microsatellite instability in early human gastric neoplasia

A significant portion of gastric cancers exhibit defective DNA mismatch repair, manifested as microsatellite instability (MSI). High-frequency MSI (MSI-H) is associated with hypermethylation of the human mut-L homologue 1 (hMLH1) mismatch repair gene promoter and diminished hMLH1 expression in advanced gastric cancers. However, the relationship between MSI and hMLH1 hypermethylation has not been studied in early gastric neoplasms. We therefore investigated hMLH1 hypermethylation, hMLH1 expression and MSI in a group of early gastric cancers and gastric adenomas. Sixty-four early gastric neoplasms were evaluated, comprising 28 adenomas, 18 mucosal carcinomas, and 18 carcinomas with superficial submucosal invasion but clear margins. MSI was evaluated using multiplex fluorescent PCR to amplify loci D2S123, D5S346, D17S250, BAT 25 and BAT 26. Methylation-specific PCR was performed to determine the methylation status of hMLH1. In two hypermethylated MSI-H cancers, hMLH1 protein expression was also evaluated by immunohistochemistry. Six of sixty-four early gastric lesions were MSI-H, comprising 1 adenoma, 4 mucosal carcinomas, and 1 carcinoma with superficial submucosal invasion. Two lesions (one adenoma and one mucosal carcinoma) demonstrated low-frequency MSI (MSI-L). The remaining 56 neoplasms were MSI-stable (MSI-S). Six of six MSI-H, one of two MSI-L, and none of thirty MSI-S lesions showed hMLH1 hypermethylation (P<0.001). Diminished hMLH1 protein expression was demonstrated by immunohistochemistry in two of two MSI-H hypermethylated lesions. hMLH1 promoter hypermethylation is significantly associated with MSI and diminished hMLH1 expression in early gastric neoplasms. MSI and hypermethylation-associated inactivation of hMLH1 are more prevalent in early gastric cancers than in gastric adenomas. Thus, hypermethylation-associated inactivation of the hMLH1 gene can occur early in gastric carcinogenesis.

Hypermethylated APC DNA in plasma and prognosis of patients with esophageal adenocarcinoma

BACKGROUND

The adenomatous polyposis coli (APC) locus on chromosome 5q21-22 shows frequent loss of heterozygosity (LOH) in esophageal carcinomas. However, the prevalence of truncating mutations in the APC gene in esophageal carcinomas is low. Because hypermethylation of promoter regions is known to affect several other tumor suppressor genes, we investigated whether the APC promoter region is hypermethylated in esophageal cancer patients and whether this abnormality could serve as a prognostic plasma biomarker.

METHODS

We assayed DNA from tumor tissue and matched plasma from esophageal cancer patients for hypermethylation of the promoter region of the APC gene. We used the maximal chi-square statistic to identify a discriminatory cutoff value for hypermethylated APC DNA levels in plasma and used bootstrap-like simulations to determine the P: value to test for the strength of this association. This cutoff value was used to generate Kaplan-Meier survival curves. All P values were based on two-sided tests.

RESULTS

Hypermethylation of the promoter region of the APC gene occurred in abnormal esophageal tissue in 48 (92%) of 52 patients with esophageal adenocarcinoma, in 16 (50%) of 32 patients with esophageal squamous cell carcinoma, and in 17 (39.5%) of 43 patients with Barrett's metaplasia but not in matching normal esophageal tissues. Hypermethylated APC DNA was observed in the plasma of 13 (25%) of 52 adenocarcinoma patients and in two (6.3%) of 32 squamous carcinoma patients. High plasma levels of methylated APC DNA were statistically significantly associated with reduced patient survival (P =.016).

CONCLUSION

The APC promoter region was hypermethylated in tumors of the majority of patients with primary esophageal adenocarcinomas. Levels of hypermethylated APC gene DNA in the plasma may be a useful biomarker of biologically aggressive disease in esophageal adenocarcinoma patients and should be evaluated as a potential biomarker in additional tumor types.

Microsatellite instability in inflammatory bowel disease-associated neoplastic lesions is associated with hypermethylation and diminished expression of the DNA mismatch repair gene, hMLH1

Twelve to 15% of sporadic colorectal cancers display defective DNA mismatch repair (MMR), manifested as microsatellite instability (MSI). In this group of cancers, promoter hypermethylation of the MMR gene hMLH1 is strongly associated with, and believed to be the cause of, MSI. A subset of colorectal neoplastic lesions arising in inflammatory bowel disease (IBD) is also characterized by MSI. We wished to determine whether hMLH1 hypermethylation was associated with diminished hMLH1 protein expression and MSI in IBD neoplasms. We studied 148 patients with IBD neoplasms, defined as carcinoma or dysplasia occurring in patients with ulcerative colitis or Crohn's disease. MSI was evaluated using multiplex fluorescent PCR to amplify loci D2S123, BAT-25, BAT-26, D5S346, and D17S250 in all cases. Lesions were characterized as high-frequency MSI (MSI-H) if they manifested instability at two or more loci, low-frequency MSI (MSI-L) if unstable at only one locus, or MS-stable (MSS) if showing no instability at any loci. Methylation-specific PCR was performed to determine the methylation status of the hMLH1 promoter region. hMLH1 protein expression was also evaluated by immunohistochemistry. Thirteen (9%) of 148 neoplasms arising in IBD were MSI-H, comprising 11 carcinomas and 2 dysplastic lesions. Sixteen additional lesions (11%) were MSI-L, comprising 11 carcinomas and 5 dysplastic lesions. The remaining 118 neoplasms (80%) were MSS. Six (46%) of 13 MSI-H, 1 (6%) of 16 MSI-L, and 4 (15%) of 27 MSS lesions showed hMLH1 hypermethylation (P = 0.013). Diminished hMLH1 protein expression in neoplastic cell nuclei relative to surrounding normal cell nuclei was demonstrated immunohistochemically in four of four (100%) hypermethylated lesions tested. In IBD neoplasia, hMLH1 promoter hypermethylation occurs frequently in the setting of MSI, particularly MSI-H. Furthermore, hMLH1 hypermethylation and MSI are strongly associated with diminished hMLH1 protein expression in IBD neoplasms. These findings suggest that hMLH1 hypermethylation causes defective DNA MMR in at least a subset of IBD neoplasms.

Distinct methylation patterns of two APC gene promoters in normal and cancerous gastric epithelia

The adenomatous polyposis coli (APC) tumor suppressor gene is mutationally inactivated in both familial and sporadic forms of colorectal cancers. In addition, hypermethylation of CpG islands in the upstream portion of APC, a potential alternative mechanism of tumor suppressor gene inactivation, has been described in colorectal cancer. Because a subset of both gastric and colorectal cancers display the CpG island methylator phenotype, we hypothesized that epigenetic inactivation of APC was likely to occur in at least some gastric cancers. APC exhibits two forms of transcripts from exons 1A and 1B in the stomach. Therefore, we investigated CpG island methylation in the sequences upstream of exons 1A and 1B, i.e., promoters 1A and 1B, respectively. We evaluated DNAs from 10 gastric cancer cell lines, 40 primary gastric cancers, and 40 matching non-cancerous gastric mucosae. Methylated alleles of promoter 1A were present in 10 (100%) of 10 gastric cancer cell lines, 33 (82.5%) of 40 primary gastric cancers, and 39 (97.5%) of 40 noncancerous gastric mucosae. In contrast, promoter 1B was unmethylated in all of these same samples. APC transcripts from exon 1A were not expressed in nine of the 10 methylated gastric cancer cell lines, whereas APC transcripts were expressed from exon 1B. Thus, expression from a given promoter correlated well with its methylation status. We conclude that in contrast to the colon, methylation of promoter 1A is a normal event in the stomach; moreover, promoter 1B is protected from methylation in the stomach and thus probably does not participate in this form of epigenetic APC inactivation.

E-Cadherin gene promoter hypermethylation in primary human gastric carcinomas

BACKGROUND

E (epithelial)-cadherin, the cell adhesion molecule also considered a potential invasion/metastasis suppressor, is mutationally inactivated in nearly half of all undifferentiated-scattered (diffuse-type) gastric carcinomas. In addition, silencing of E-cadherin by CpG methylation within its promoter region has been reported in several gastric carcinoma cell lines. We investigated the methylation status of the E-cadherin promoter region in 53 primary human gastric carcinomas.

METHODS

Hypermethylation of the E-cadherin promoter was determined by utilizing methylation-specific polymerase chain reaction (PCR)-single-strand conformation polymorphism (MSP-SSCP) analysis followed by direct sequencing of PCR products. Expression of E-cadherin was studied by western blot analysis. All statistical tests were two-sided.

RESULTS

Hypermethylation of the E-cadherin promoter was evident in 27 (51%) of 53 primary gastric carcinomas examined by MSP-SSCP. It occurred more frequently in carcinomas of the undifferentiated-scattered type (in 15 [83%] of 18) than in other histologic subtypes (in 12 [34%] of 35) (P =.0011, Fisher's exact test), and it was present at similar rates in early (in six [60%] of 10) versus advanced (in 21 [49%] of 43) carcinomas (P =.73, Fisher's exact test). Methylation occurring at all cytosine-guanosine sequences (CpGs) near the transcriptional start site was confirmed in six of six tumors examined by bisulfite-DNA sequencing, including two early gastric carcinomas. In addition, loss or diminished expression of E-cadherin was confirmed by western blotting in four of the six tumor tissues demonstrating hypermethylation.

CONCLUSIONS

The E-cadherin promoter frequently undergoes hypermethylation in human gastric cancers, particularly those of the undifferentiated-scattered histologic subtype. E-cadherin promoter hypermethylation is associated with decreased expression and may occur early in gastric carcinogenesis.

Low prevalence of the APC I1307K sequence in Jewish and non-Jewish patients with inflammatory bowel disease

A germline sequence alteration at codon 1307 of the APC gene (I1307K) has been reported in 6-7% of the Ashkenazi Jewish population in the United States. This alteration is believed to predispose the APC gene to a secondary mutation at the same locus, resulting in an increased risk of colorectal carcinoma. There is an increased risk of colorectal carcinoma in patients with inflammatory bowel disease (IBD), a relatively large proportion of whom are Ashkenazi Jews. We therefore sought to determine whether the I1307K sequence variant occurred in the germline DNA of IBD patients. To our surprise, we found this sequence in only two of 267 patients with IBD (0.7%), occurring in only 1.5% of Jewish IBD patients. The I1307K sequence variant was not found in 67 patients with esophageal cancer, 53 patients with gastric carcinoma (13 MSI-H and 44 MSI-negative), or ten patients with sporadic MSI-H colon cancer. These findings suggest that the I1307K sequence is relatively rare in the germline of Jewish as well as non-Jewish IBD patients. It does not appear to contribute to the increased colorectal cancer risk present in these patients.

Hypermethylation of the hMLH1 gene promoter in human gastric cancers with microsatellite instability

Human gastric carcinoma shows a higher prevalence of microsatellite instability (MSI) than does any other type of sporadic human cancer. The reasons for this high frequency of MSI are not yet known. In contrast to endometrial and colorectal carcinoma, mutations of the DNA mismatch repair (MMR) genes hMLH1 or hMSH2 have not been described in gastric carcinoma. However, hypermethylation of the hMLH1 MMR gene promoter is quite common in MSI-positive endometrial and colorectal cancers. This hypermethylation has been associated with hMLH1 transcriptional blockade, which is reversible with demethylation, suggesting that an epigenetic mechanism underlies hMLH1 gene inactivation and MMR deficiency. Therefore, we studied the prevalence of hMLH1 promoter hypermethylation in a total of 65 gastric tumors: 18 with frequent MSI (MSI-H), 8 with infrequent MSI (MSI-L), and 39 that were MSI negative. We found a striking association between hMLH1 promoter hypermethylation and MSI; of 18 MSI-H tumors, 14 (77.8%) showed hypermethylation, whereas 6 of 8 MSI-L tumors (75%) were hypermethylated at hMLH1. In contrast, only 1 of 39 (2.6%) MSI-negative tumors demonstrated hMLH1 hypermethylation (P<0.0001 for MSI-H or MSI-L versus MSI-negative). Moreover, hypermethylated cancers demonstrated diminished expression of hMLH1 protein by both immunohistochemistry and Western blotting, whereas nonhypermethylated tumors expressed abundant hMLH1 protein. These data indicate that hypermethylation of hMLH1 is strongly associated with MSI in gastric cancers and suggest an epigenetic mechanism by which defective MMR occurs in this group of cancers.

Sequence alterations of insulin-like growth factor binding protein 3 in neoplastic and normal gastrointestinal tissues

Insulin-like growth factor binding protein 3 (IGFBP-3) is an important regulator of normal and malignant cell growth. It modulates the mitogenic effects of insulin-like growth factors (IGFs) by inhibiting growth through mechanisms both dependent on and independent of IGF binding. IGF-I and IGF-II levels are regulated by binding to the IGF-II receptor, which is inactivated by mutation in human gastrointestinal (GI) tumors. We have previously demonstrated elevated IGF-II ligand expression in IGF-II receptor-mutant GI tumors, implicating the IGF signaling system in GI tumorigenesis. Therefore, to investigate the potential involvement of IGFBP-3 in human GI carcinogenesis, direct DNA sequencing of exons 1-4 and intron-exon boundaries of the IGFBP-3 gene was performed in 10 colorectal cancers, 10 gastric cancers, and 10 esophageal cancers. Four distinct sequence alterations were identified: (a) in one gastric and one esophageal tumor, an A to C transversion occurred at nucleotide 5795 (CAC-->CCC), leading to a His-->Pro substitution at codon 179; (b) a second esophageal tumor had a C to T transition at nucleotide 8291 (ACC-->ATC), leading to a Thr-->Ile substitution at codon 277 of IGFBP-3; (c) one alteration comprised a G to C transversion in exon 1 at nucleotide 2132 (GGG-->GCG), leading to a Gly-->Ala substitution at codon 32 in two gastric cancers, seven esophageal cancers, and nine colon cancers; and (d) a C to G transversion located 17 nucleotides from the 3' splice site in intron 1 was observed in three colon cancers and four esophageal cancers. All of these DNA sequence alterations were present in matched normal DNA from the same subjects, which suggests that some or all of them may represent polymorphisms. However, we cannot exclude the possibility that the germ-line nonconservative amino acid substitutions predicted to occur as a result of these alterations result in subtle changes to IGFBP-3 protein function and a predisposition to developing GI malignancy.