Background: Familial pancreatic cancer (FPC) describes a group of families where the inheritance of pancreatic cancer is consistent with an autosomal-dominant mode of inheritance. The 4q32-34 region has been previously identified as a potential locus for FPC in a large American family.

Methods: The region was allelotyped in 231 individuals from 77 European families using nine microsatellite markers, and haplotyping was possible in 191 individuals from 41 families. Families were selected based on at least two affected first-degree relatives with no other cancer syndromes.

Results: Linkage to most of the locus was excluded based on LOD scores less than −2.0. Eight families were excluded from linkage to 4q32-34 based on haplotypes not segregating with the disease compared with a predicted six to seven families. Two groups of families were identified, which seem to share common alleles within the minimal disease-associated region of 4q32-34, one group with an apparently earlier age of cancer death than the other pancreatic cancer families. Four genes were identified with potential tumor suppressor roles within the locus in regions that could not be excluded based on the LOD score. These were HMGB2, PPID, MORF4, and SPOCK3. DNA sequence analysis of exons of these genes in affected individuals and in pancreatic cancer cell lines did not reveal any mutations.

Conclusion: This locus is unlikely to harbor a FPC gene in the majority of our European families. (Cancer Epidemiol Biomarkers Prev 2006;15(10):1948–55)

Pancreatic ductal adenocarcinoma is one of the most common causes of cancer death in the Western world (1-4). An estimated 2.7% to 10% of pancreatic cancers occur in families where there is at least one other case (5-8). In some instances, this is associated with a general familial cancer syndrome, such as familial atypical multiple-mole melanoma syndrome, familial adenomatous polyposis, hereditary nonpolyposis colorectal cancer, hereditary breast-ovarian cancer syndrome, and Peutz-Jeghers syndrome (9-13), although in most cases, these syndromes are associated with isolated rather than multiple cases of pancreatic cancer. Germ line mutations in various mismatch repair genes, including both MutS homologues (hMSH2) and MutL homologues (hMLH1), are associated with pancreatic cancer in some patients in hereditary nonpolyposis colorectal cancer families (14). Similarly, germ line mutations in genes involved in recombinational repair, such as BRCA2, can be associated with pancreatic cancer in patients with familial breast-ovarian cancer syndromes. Germ line mutations of p16INK4a are implicated in pancreatic cancer in familial atypical multiple-mole melanoma syndrome (15) and STK11 mutations in Peutz-Jeghers syndrome (16). Other germ line mutations associated with sporadic incidences of pancreatic cancer include those of the FANCG gene (17). In the majority of families with multiple cases of pancreatic cancer, the genetic cause is unknown. This includes a group of families where the inheritance of pancreatic cancer is consistent with an autosomal-dominant mode of inheritance; these are grouped together under the umbrella term familial pancreatic cancer (FPC). Registries of such families have been established in North America (18) and in Europe (19). Germ line BRCA2 mutations may occur in up to 20% of FPC (20, 21). However, other candidate genes, such as STK11 (22) and p16INK4a (23), have been ruled out as causative in the majority of FPC families. Much has been documented of the association of DNA repair genes and FPC, consistent with our previous observation of BRCA2 mutations in some of our families (20). We and others have previously reported anticipation in FPC families (19, 24, 25), and it is of interest that anticipation in cancer syndromes may be the result of mutations in DNA repair genes (26). It has also been reported that fanconi anemia patients with mutations in DNA repair genes FANCC and FANCG exhibit young onset pancreatic cancer (27).

Segregation analysis of FPC families from the United States has shown that a major gene must be involved in the etiology of pancreatic cancer (28). Linkage analysis to identify the disease locus in qualitative conditions (patients are either affected or unaffected) with a late onset is complicated by the problem of distinguishing noncarriers from nonpenetrants. This was overcome by Eberle et al. by using a large family where unaffected individuals were screened for preneoplastic lesions, which could then be used as a marker for gene carriers (29). In this way, they were able to establish that a 4q32-34 locus segregated with pancreatic cancer in a single large FPC kindred. Within the 4q32-34 region, there are 36 characterized genes and many putative genes. There are >120 different single nucleotide polymorphisms linked to these genes (http://www.ncbi.nlm.nih.gov/mapview).

In this article, we will describe allelotyping within the 4q32-34 region in families with an apparently autosomal-dominant inheritance of pancreatic ductal adenocarcinoma, registered by the European Registry of Hereditary Pancreatitis and Familial Pancreatic Cancer and the German National Case Collection for Familial Pancreatic Cancer. Two approaches were used to examine whether these allelotypes were consistent with an association of the 4q32-34 locus with the disease: haplotyping and two-point linkage analysis. As will be described, for both forms of analysis, all individuals with pancreatic ductal adenocarcinoma were assumed to be carriers of an autosomal-dominant disease mutation, and the probability that other members of the family were carriers was estimated according to their age and relationship to affected individuals: this required a model of penetrance that we based on the assumption of anticipation. Evidence for anticipation in FPC has been presented by several groups (24, 25) and has recently been rigorously tested using the families described in this article (19).

Patients

Families were ascertained from cancer genetics, surgical, and gastroenterological centres in Europe. Patients donated blood samples and information with full informed consent. Clinical status of affected individuals was confirmed histologically, from detailed medical notes or from cancer registries. Only families with at least two affected first-degree relatives were included. Previously reported families with BRCA2 mutations (20) were also excluded from linkage and haplotyping analysis (other than as controls representing families with no linkage to 4q32-34). Other excluded families included those with an unusually high incidence of other types of cancer, as such families might belong to other familial cancer syndromes; this included exclusion of families with p16 mutations associated with melanoma and pancreatic cancer.

Allelotyping

Allelotyping was carried out using nine mapping pairs between D4S413 and D4S415 spanning a 23-cM region of 4q32-34 (details of the primer sets are given in the Supplementary Material). PCR products were sized using the ABI377 DNA sequencer and/or the capillary-based 3100 ABI Genetic Analyser. The data were analyzed, and the alleles were assigned by Genotyper software (ABI). A marker of a defined size will incur an altered mobility depending on whether it was analyzed using a gel-based (ABI377) or capillary-based (ABI3100) system. Therefore, we analyzed all nine microsatellite markers of a known size from sequenced BAC clones on both systems and calculated and corrected the discrepancy between the two systems. Primers were obtained from the ABI prism linkage mapping set version 2 or were custom synthesized by MWG-Biotech (see Supplementary Table S4).

Defining Liability Classes

We have previously identified that penetrance for FPC families depends on age and ancestry (19). Individuals with one or more previous generations affected by pancreatic cancer are penetrant at a younger age than those with cases of pancreatic cancer only in their own generation (siblings or cousins), or subsequent generations (children, nieces, nephews, or grandchildren). Age-related penetrance cannot be accurately calculated due to inter- and intra-familial variation. However, an estimate can be made by regression analysis of the survival curve for affected individuals. As described previously (19), survival in the low-penetrance group declines linearly from the age of 50 with a gradient of 0.0255. For the high-penetrance group, a nonlinear regression of the survival curve was used (% survival = 1 − eln (Age) × 5.35 − 23). The chance of an individual being affected by a given age is a measure of the assumed penetrance for a carrier at this age. For example, if a carrier has a 25% chance of being affected by 55 years, then at 55 years, the mutation would be assumed to be 25% penetrant. Descriptions of other series of FPC families suggest that smoking habit dramatically influences age of onset and penetrance of pancreatic cancer; however, we previously reported that with our patients, there is no evidence of smoking influencing outcome in potential gene carriers. Whereas this is absence of evidence rather than proof of no effect, we clearly cannot justify modeling penetrance based on smoking habit (19).

Two-Point LOD Score Analysis

A two-point LOD score analysis was done for each of the nine markers using parametric two-point analyses with the MLINK program from the LINKAGE (version 5.1) software package (30) with three different models. The LOD scores were calculated based on the age-related penetrance and the probability of being a carrier based only on relationship with affected individuals. Models 1 and 2 assumed an autosomal-dominant mode of inheritance with a very low global rate for phenocopies (0.001) and an age-dependent penetrance (as above) to account for carriers who have not developed cancer. The age-dependent penetrance was implemented by using varying numbers of age-related liability classes: 20 classes were used for model 1, and six classes were used for model 2. Global penetrances were used for model 3 as only affected individuals were taken into account.

Candidate Gene Sequencing

The genes HMGB2, PPID and MORF4, and SPOCK3 were sequenced in affected individuals or obligate carriers in FPC families by direct sequencing. The five exons of the HMGB2 gene, 10 exons of PPID, the single exon of the MORF4 gene, and the 12 exons of SPOCK3 were PCR amplified with the primers shown in Supplementary Table S5. DNA sequencing was done using the 3100 ABI Genetic Analyser.

Haplotype Analysis

Haplotyping was possible in 191 individuals (40 affected) from 30 European Registry of Hereditary Pancreatitis and Familial Pancreatic Cancer families and 11 families from the German National Case Collection for Familial Pancreatic Cancer registry. None of these families had known BRCA2, but in eight of these families, BRCA2 had not been sequenced. As previously reported, there is an ∼15% incidence of BRCA2 mutations in apparently FPC families; thus, it is not unreasonable to suppose that at least one of these families carried a BRCA2 mutation (20). Details of the nationality, number of individuals in each family, and the proportion of cancer cases with histologic confirmation are given in the Supplementary Material. All families included at least three characterized generations with a median of four generations. It was possible to exclude linkage to 4q32-34 in 8 of 41 families, either by showing that two affected individuals did not share a haplotype, or that an affected individual shared the only possible linked haplotype with an unaffected individual. Unaffected individuals were defined using an age-related estimate of penetrance and assuming an autosomal-dominant mode of inheritance. Figure 1 shows an example of one such family. In G1 of the family in Fig. 1, an affected male (1-1) died from pancreatic cancer at age 79, and in G2, an affected female (2-1) died at age 62. Individual 2-1 has inherited the haplotype represented in blue from her affected father (1-1); this haplotype was also inherited by individual 2-3. The estimated age-related probability of a carrier in the high-risk generation not being affected by age 79 would be just 3.9%, and the chance of this patient being a carrier was 50% based on relationship (as she has first-degree relatives who developed cancer). Hence, the probability that individual 2-3 is a gene carrier and nevertheless still unaffected was <2%. The age-related probability of being affected was defined as described in Defining Liability Classes. Because two affected assumed gene carriers shared a haplotype with an assumed non-gene carrier we excluded linkage to the locus in this family.

Figure 1.

Family in which 4q32-34 has been excluded. Individual 2-1 died of pancreatic cancer and is therefore assumed to be a carrier. Individual 2-3 has been classified as unaffected (a noncarrier). As 2-1 and 2-3 share both haplotypes, the family was excluded as having disease linkage to 4q32-34.

Figure 1.

Family in which 4q32-34 has been excluded. Individual 2-1 died of pancreatic cancer and is therefore assumed to be a carrier. Individual 2-3 has been classified as unaffected (a noncarrier). As 2-1 and 2-3 share both haplotypes, the family was excluded as having disease linkage to 4q32-34.

Close modal

This method of excluding families relies heavily on generalizations, clearly the definition of unaffected could include non-penetrant carriers, and it is possible that an affected individual is a phenocopy. Thus, there are four possible interpretations of the data:

  1. All families have disease linked to 4q32-34, and all exclusions are artifactual.

  2. No families have disease linked to 4q32-34, and haplotypic evidence of linkage is random.

  3. All excluded families have disease that is not linked to 4q32-34, but some families which cannot be excluded are linked to 4q32-34.

  4. Some excluded families have disease that is linked to 4q32-34, and others do not; some families which cannot be excluded are linked to 4q32-34.

The chance that an exclusion is artifactual will depend on the nature of the method used for exclusion. The possible methods are:

  1. Two affected individuals who share no haplotype (AA),

  2. An affected individual who shares both haplotypes with an unaffected individual (AU),

  3. A unique haplotype being shared by two affected individuals and an unaffected individual sharing the same haplotype (AAU),

  4. A unique haplotype being shared by two affected individuals and a third affected individual not sharing this haplotype (AAA),

  5. Two unaffected individuals having between them both possible haplotypes of an affected individual (AUU).

In most cases (25 of 41 families), exclusion of linkage to 4q32-34 was impossible, either because haplotypes were not informative, or because it proved impossible to classify a key individual as unaffected. However, in 16 families, there was sufficient haplotype data to examine exclusion; these are listed in Table 1. The probability that the available data would lead to exclusion of linkage is given in the table for each family.

Table 1.

Probability of exclusion by haplotype

Name of familyType of exclusionExluded (Y/N)Alternative exclusion criteriaProbability of exclusion
FamA AAU* — 0.25* 
FamB AA AU × 2, AAU AAU* 0.63* 
FamC AA* — 0.5* 
FamD AA AU × 2, AAU 0.56 
FamE AUU — 0.25 
FamF AAU — 0.5 
FamG AU AU, AUU, AAU, AAU 0.75 
FamH AU — 0.25 
FamJ — AA 0.25 
FamK — AU 0.25 
FamL — AA, AU, AAU, 0.75* 
   AA*, AU*  
FamN — AU, AU, AUU 0.56 
FamO — AA 0.25 
FamP — AU 0.25 
FamQ — AU 0.25 
FamR — AA 0.25 
Name of familyType of exclusionExluded (Y/N)Alternative exclusion criteriaProbability of exclusion
FamA AAU* — 0.25* 
FamB AA AU × 2, AAU AAU* 0.63* 
FamC AA* — 0.5* 
FamD AA AU × 2, AAU 0.56 
FamE AUU — 0.25 
FamF AAU — 0.5 
FamG AU AU, AUU, AAU, AAU 0.75 
FamH AU — 0.25 
FamJ — AA 0.25 
FamK — AU 0.25 
FamL — AA, AU, AAU, 0.75* 
   AA*, AU*  
FamN — AU, AU, AUU 0.56 
FamO — AA 0.25 
FamP — AU 0.25 
FamQ — AU 0.25 
FamR — AA 0.25 

NOTE: AA, two affected share no haplotype; AU, an affected shares both haplotypes with an unaffected; AUU, both haplotypes of an affected are carried by (different) unaffected individuals; AAA, only one haplotype is shared by two affected, which is not shared by a third; AAU, a single haplotype is shared by two affected and is also shared by an unaffected. The probabilities were calculated iteratively. For example, in FamD, there are two affected siblings with haplotypes (A1 and A2) and an unaffected sibling (U). Step 1 (exclusion by A1A2): 25% of families would be excluded. Step 2 (exclusion by A1U): 25% of families not already excluded would be expected to be excluded on this basis (56% of families would still not be excluded). Step 3 (exclusion by A2U): from steps 1 and 2, we know that the A2 shares at least one haplotype with A1, and that A1 does not have the same haplotypes as U. Nine possible combinations of haplotypes remain, only two of which will allow exclusion; therefore, 44% of families would still not be excluded (probability of exclusion = 0.56). Theoretically, there is a step 4 (A1A2U), whereby the unique haplotype shared by the affected is also shared by the unaffected; however, no family that would not have already been excluded in steps 1 to 3 could be excluded in this way.

*

Involves second-degree relationship.

Simple addition of these probabilities predicts that 6.5 of these families would be excluded by chance, assuming no linkage. This compares to eight families that were actually excluded. If any of these 16 families did have disease linked to 4q32-34, a bias against exclusion should be observed. No bias towards inclusion was observed.

It would be predicted that if a subgroup of the 41 families had disease linked to 4q32-34, then they would have alleles in common. Clusters of alleles would also be expected between families. In six families, possible disease haplotypes were identified. These haplotypes are shown in Table 2. Also shown are the two possible disease haplotypes for a family where exclusion was possible but was not observed, and the allelotypes of a further family where neither haplotype of an unaffected individual could possibly be represented in an affected, but it was not possible to haplotype the affected.

Table 2.

Haplotypes not excluded from linkage to pancreatic cancer

MarkerSize allele frequency (% carrier, noncarrier, unknown)
FamJFamKFamOFamLFamNFamPFamQFamR
D4S413 216 (37, 43, 42) 242 (11, 4, 9) 216 (37, 43, 42) 226 (8, 12, 10) 226 (8, 12, 10) 214 (11, 10, 8) 228 (5, 2, 3) 216 (37, 43, 42) 214 (11, 10, 8)-240 (2, 0, 1) 
D4S393 121 (30, 30, 35) 125 (3, 6, 3) 131 (26, 15, 26) 121 (30, 30, 35) 123 (25, 26, 19) 121 (30, 30, 35) 121 (30, 30, 35) 123 (25, 26, 19) 123 (25, 26, 19)-131 (26, 15, 26) 
D4S1603 198 (13, 7, 11) 196 (33, 40, 34) 194 (14, 21, 16) 202 (8, 5, 6) 202 (8, 5, 6) 198 (13, 7, 11) 200 (8, 7, 9) 206 (4, 2, 4) 196 (33, 40, 34)-202 (8, 5, 6) 
D4S2952 182 (23, 16, 22) 186 (8, 5, 5) 190 (38, 45, 37) 190 (38, 45, 37) 204 (5, 4, 6) 186 (8, 5, 5) 182 (23, 16, 22) 190 (38, 45, 37) 184 (6, 16, 8)-194 (4, 2, 2) 
D4S1596 220 (33, 24, 36) 220 (33, 24, 36) 220 (33, 24, 36) 218 (56, 54, 52) 218 (56,54,52) 216 (9, 10, 6) 220 (33, 24, 36) 218 (56, 54, 52) 218 (56, 54, 52)-218 (56, 54, 52) 
D4S1597 290 (9, 13, 12) 290 (9, 13, 12) 276 (37, 50, 56) 292 (1, 5, 1) 282 (16, 3, 8) 274 (19, 13, 10) 276 (37, 50, 56) 284 (9, 5, 11) 282 (16, 3, 8)-284 (9, 5, 11) 
D4S1617 186 (53, 40, 51) 186 (53, 40, 51) 186 (53, 40, 51) 186 (53, 40, 51) 186 (53, 40, 51) 184 (29, 30, 24) 184 (29, 30, 24) 186 (53, 40, 51) 186 (53, 40, 51)-186 (53, 40, 51) 
D4S1539 206 (29, 29, 31) 206 (29, 29, 31) 208 (35, 25, 32) 208 (35, 25, 32) 208 (35, 25, 32) 208 (35, 25, 32) 208 (35, 25, 32) 212 (4, 2, 3) 206 (29, 29, 31) - 202 (2, 5, 4) 
D4S415 176 (32, 26, 32) 200 (7, 9, 6) 194 (12, 9, 12) 196 (11, 19, 18) 194 (12, 9, 12) 198 (13, 9, 10) 194 (12, 9, 12) 198 (13, 9, 10) 174 (5, 19, 7) - 174 (5, 19, 7) 
MarkerSize allele frequency (% carrier, noncarrier, unknown)
FamJFamKFamOFamLFamNFamPFamQFamR
D4S413 216 (37, 43, 42) 242 (11, 4, 9) 216 (37, 43, 42) 226 (8, 12, 10) 226 (8, 12, 10) 214 (11, 10, 8) 228 (5, 2, 3) 216 (37, 43, 42) 214 (11, 10, 8)-240 (2, 0, 1) 
D4S393 121 (30, 30, 35) 125 (3, 6, 3) 131 (26, 15, 26) 121 (30, 30, 35) 123 (25, 26, 19) 121 (30, 30, 35) 121 (30, 30, 35) 123 (25, 26, 19) 123 (25, 26, 19)-131 (26, 15, 26) 
D4S1603 198 (13, 7, 11) 196 (33, 40, 34) 194 (14, 21, 16) 202 (8, 5, 6) 202 (8, 5, 6) 198 (13, 7, 11) 200 (8, 7, 9) 206 (4, 2, 4) 196 (33, 40, 34)-202 (8, 5, 6) 
D4S2952 182 (23, 16, 22) 186 (8, 5, 5) 190 (38, 45, 37) 190 (38, 45, 37) 204 (5, 4, 6) 186 (8, 5, 5) 182 (23, 16, 22) 190 (38, 45, 37) 184 (6, 16, 8)-194 (4, 2, 2) 
D4S1596 220 (33, 24, 36) 220 (33, 24, 36) 220 (33, 24, 36) 218 (56, 54, 52) 218 (56,54,52) 216 (9, 10, 6) 220 (33, 24, 36) 218 (56, 54, 52) 218 (56, 54, 52)-218 (56, 54, 52) 
D4S1597 290 (9, 13, 12) 290 (9, 13, 12) 276 (37, 50, 56) 292 (1, 5, 1) 282 (16, 3, 8) 274 (19, 13, 10) 276 (37, 50, 56) 284 (9, 5, 11) 282 (16, 3, 8)-284 (9, 5, 11) 
D4S1617 186 (53, 40, 51) 186 (53, 40, 51) 186 (53, 40, 51) 186 (53, 40, 51) 186 (53, 40, 51) 184 (29, 30, 24) 184 (29, 30, 24) 186 (53, 40, 51) 186 (53, 40, 51)-186 (53, 40, 51) 
D4S1539 206 (29, 29, 31) 206 (29, 29, 31) 208 (35, 25, 32) 208 (35, 25, 32) 208 (35, 25, 32) 208 (35, 25, 32) 208 (35, 25, 32) 212 (4, 2, 3) 206 (29, 29, 31) - 202 (2, 5, 4) 
D4S415 176 (32, 26, 32) 200 (7, 9, 6) 194 (12, 9, 12) 196 (11, 19, 18) 194 (12, 9, 12) 198 (13, 9, 10) 194 (12, 9, 12) 198 (13, 9, 10) 174 (5, 19, 7) - 174 (5, 19, 7) 

NOTE: In six families, only one haplotype could be linked to pancreatic cancer. In family FamQ, two haplotypes are given (neither can be excluded). In FamR, no haplotype could be confidently assigned, but any haplotype produced from the alleles in the table could be linked to pancreatic cancer in this family. The alleles were compared with those from 77 families on the registry: 53 carriers, 30 noncarriers, and 181 unknowns (allele frequencies in brackets). Alleles in bold are discussed in the text.

In total, allelotyping was done on 264 individuals from 77 families; this included 30 noncarriers (off-kindred family members and individuals from families with a known BRCA2 mutation), 53 carriers (affected individuals or obligate carriers), and 181 unknowns (family members who are not affected but might be carriers). The frequency for each allele in carriers, noncarriers, and unknowns is given beneath the alleles in Table 2. Alleles associated with a subpopulation of families with 4q32-34 linkage would be expected to be more common in carriers than in noncarriers with an intermediate representation in individuals who might be carriers. As noted above, BRCA2 sequencing was not carried out in all families; in 35 families, no individual had complete BRCA2 sequencing; thus, potentially five or more of these families harbored a BRCA2 mutation.

Patterns of alleles are discernible, including relative enrichment of particular alleles, but the significance of these patterns is difficult to interpret. To better analyze these data, subregions must first be classified according to the likelihood that they are linked to pancreatic cancer in the studied families.

Two-Point LOD Score Analysis

For two-point LOD score analysis, there is no limitation to families where haplotyping was possible; thus, allelotypes of 196 on-kindred individuals from 70 families were used. Families with known BRCA2 mutations and two families where the only allelotypes were from participants with no first-degree relative with pancreatic cancer were excluded. Two-point LOD score analysis of all 70 families permitted exclusion mapping for each marker. The results of LOD score analyses are shown in Fig. 2. LOD scores for each marker are plotted against cM distance (Marshfield map; ref. 31) of the locus as a function of the recombination frequency. LOD scores were plotted to a maximum distance of 60 cM from D4S413 due to the fact that the end of chromosome 4 was ∼53 cM telomeric of D4S413.

Figure 2.

Two-point LOD scores. Two-point LOD scores (Y-axis) for each marker plotted against cM distance along the locus (X-axis). A. Model 3 used a global penetrance function as only affected individuals are taken in to account. B. Model 2 assuming an age-dependent penetrance with two risk categories (low and high) and six liability classes. C. Model 1 assuming an age-dependent penetrance with two risk categories (low and high) and 20 liability classes (numerical values are given in the Supplementary Table S6).

Figure 2.

Two-point LOD scores. Two-point LOD scores (Y-axis) for each marker plotted against cM distance along the locus (X-axis). A. Model 3 used a global penetrance function as only affected individuals are taken in to account. B. Model 2 assuming an age-dependent penetrance with two risk categories (low and high) and six liability classes. C. Model 1 assuming an age-dependent penetrance with two risk categories (low and high) and 20 liability classes (numerical values are given in the Supplementary Table S6).

Close modal

Two-point LOD score analyses using model 3 (see Materials and Methods) are shown in Fig. 2A (this is given in numerical form in Supplementary Table S6). This model assumed that only affected individuals were gene carriers and did not allow for genetic anticipation. This model did not allow us to either include or exclude any part of the region; this represents a lack of power and would also be true in any nonparametric analysis. However, using an exclusion threshold of −2 for model 2 (with six liability classes) linkage was excluded for most of the region (Fig. 2B). Only the shaded regions adjacent to D4S1539 were not excluded. Similarly, most of the region can be excluded using model 1 (Fig. 2C) except for the region telomeric to D4S1617.

Allele Clustering in Candidate Haplotypes

From the two-point LOD scores, it would seem that the least unlikely region to contain a possible disease gene lies in the region of marker D4S1597. From the haplotypes that could be linked to pancreatic cancer in Table 2, it can be seen that the only allele of D4S1597 that seems to be overrepresented in carriers is allele 282. This appears in 16% of carriers but only 3% of noncarriers, with 8% of unknowns carrying the allele. In family FamN, allele D4S1597:282 is sandwiched between alleles D4S1596:218 and D4S1617:186. This same combination could also be possible in FamR. In total, the combination was seen in 31 individuals from 10 families. However, the haplotype can be excluded as a disease haplotype in families FamE (present in an unaffected individual), FamJ (absent in an affected), FamL (absent in two affected individuals), FamY (present in an unaffected), and FamZ (absent in one affected individual). The remaining families (FamN, FamR, FamT, FamU, and FamV) include one Italian family and four U.K. families.

A cluster of alleles around D4S1617 (D4S1597:290, D4S1617:186, and D4S1539:206) were found in possible disease haplotypes of FamJ and FamK. This same cluster is also found in seven other families. However, in two of these (FamE and FamQ), the cluster is present in individuals classified as unaffected, and in one family, it has come from an off-kindred individual (FamG). Therefore, there are four families (one Swedish and three British) that could be associated according to this cluster (FamJ, FamK, FamW, and FamX).

These two groups of families have been designated D4S1617-186 and D4S1597-282. If these groups represent part of a small subpopulation of families where 4q32-34 is associated with pancreatic cancer then it would be assumed that they would share common characteristics that might distinguish them from other families, particularly those families where 4q32-34 has been excluded.

Characteristics of the two groups of families are shown in Table 3. The first group is characterized by more male cancer, earlier age of onset, and presence of other forms of cancer. The second group has more female cases, later age of onset, and fewer other cancer types. For comparison, the gender distribution and median age of cancer death are given in Table 4 for patients from families excluded from having 4q32-34 linkage, including five families with known BRCA2 mutations. Values are also given in Table 4 for the remaining pancreatic cancer families on the combined European Registry of Hereditary Pancreatitis and Familial Pancreatic Cancer/German National Case Collection for Familial Pancreatic Cancer registry where 4q32-34 linkage is not excluded (described as “unknown”). The apparent earlier age of onset for D4S1597-282 is supported in Fig. 3A by curves of cumulative incidence of cancer death for the affected individual in the four groups of families described in Table 3. Figure 3B shows survival analysis of all potential gene carriers on the combined registry; this suggests a 50% lifetime risk to age 85 for the majority of families; as approximately half the family members will not be mutation carriers, this suggests a very high penetrance. However, the D4S1597-282 group shows a trend to a lower lifetime risk (although this does not reach significance).

Table 3.

Grouping of families according to haplotype

FamilyNo. affectedGender of affected (male/female)Average age of cancer deathOther /Cancers
D4S1596:218, D4S1597:282, D4S1617:186 (D4S1597 282)     
    FamN 2:0 42.5 None 
    FamR 2:0 51.5 1× Gastric 
    FamT 2:0 45 1× Pancreatic acinar cell 
    FamU 3:0 51 Leukemia 
    FamV 1:1 58.5 1× breast, 1× lung 
D4S1597:290, D4S1617:186, D4S1539:206 (D4S1617 186)     
    FamJ 0:4 77.3 None 
    FamK 0:2 64.5 None 
    FamW 1:2 59.3 None 
    FamX 0:3 73 1× Gastric 
FamilyNo. affectedGender of affected (male/female)Average age of cancer deathOther /Cancers
D4S1596:218, D4S1597:282, D4S1617:186 (D4S1597 282)     
    FamN 2:0 42.5 None 
    FamR 2:0 51.5 1× Gastric 
    FamT 2:0 45 1× Pancreatic acinar cell 
    FamU 3:0 51 Leukemia 
    FamV 1:1 58.5 1× breast, 1× lung 
D4S1597:290, D4S1617:186, D4S1539:206 (D4S1617 186)     
    FamJ 0:4 77.3 None 
    FamK 0:2 64.5 None 
    FamW 1:2 59.3 None 
    FamX 0:3 73 1× Gastric 
Table 4.

Characteristics of family groups

Type of familyNo. familiesGender of affected (male/female)Median age of cancer death (interquartile range)
D4S1597-282 10:1 47 (43-54) 
D4S1617-186 1:11 71 (61-78) 
Not linked 13 15:19 63 (59-71) 
Unknown 142 185:176 62 (54-69) 
Type of familyNo. familiesGender of affected (male/female)Median age of cancer death (interquartile range)
D4S1597-282 10:1 47 (43-54) 
D4S1617-186 1:11 71 (61-78) 
Not linked 13 15:19 63 (59-71) 
Unknown 142 185:176 62 (54-69) 
Figure 3.

Survival curves for family groupings. Two groups of families were identified with clusters of alleles that could be associated with pancreatic cancer (D4S1597:282 and D4S1617:186). Kaplan-Meier analysis with event time from birth to age of death from pancreatic cancer was used to compare these families with families excluded as having linkage to 4q32-34 and the bulk of the families where exclusion was impossible. A. Affected individuals in D4S1597-282 have a significantly earlier age of onset than affected individuals in other families (with 95% confidence intervals). B. All potential carriers.

Figure 3.

Survival curves for family groupings. Two groups of families were identified with clusters of alleles that could be associated with pancreatic cancer (D4S1597:282 and D4S1617:186). Kaplan-Meier analysis with event time from birth to age of death from pancreatic cancer was used to compare these families with families excluded as having linkage to 4q32-34 and the bulk of the families where exclusion was impossible. A. Affected individuals in D4S1597-282 have a significantly earlier age of onset than affected individuals in other families (with 95% confidence intervals). B. All potential carriers.

Close modal

Candidate Gene Sequencing

DNA from 10 affected individuals from 10 different families was used for sequencing of candidate genes within the 4q32-34 locus. In addition, these genes were sequenced in pancreatic cancer cell lines Capan-2, Panc-1, and FAMPAC, the latter cell line being isolated from a patient with a familial predisposition to pancreatic cancer (32).

The entire D4S413-D4S315 region was scrutinized to identify candidate FPC genes. Within this region, several genes were identified that may be associated with cancer, including growth regulators, such as vascular endothelial growth factor C (33) and annexin A10, which has been shown to be down-regulated in hepatocellular carcinoma (34). However, these were considered to be unlikely candidate FPC genes as a mutation would aid tumor progression rather than have a causative effect.

Four genes within the locus were chosen as being of particular interest. Cyclophilin D/Cyclophilin 40 (PPID) is a component of the mitochondrial permeability transition pore, and the gene is located between markers D4S413 and D4S393. Inhibition of cyclophilin D hyperpolarizes the mitochondrial membrane potential and inhibits apoptosis; thus, a mutation might similarly inhibit apoptosis and so be tumorigenic (35). PPID is also associated with the cancer-related stress protein Hsp 90 and involved in targeting Hsp 90 to inactive glucocorticoid receptor (36) and dynein (37). Mortality factor 4 (MORF4), thought to transcriptionally modulate genes important for senescence pathways or cell growth control (38), and Tetican 3 (SPOCK3), a metalloendopeptidase inhibitor with calcium binding activity that interferes with tumor invasion in brain (39), also map to this region. One gene involved in DNA repair was identified within the locus [high mobility group box 2 (HMGB2)], which maps in close proximity to D4S1617. HMGB2 acts in an alternative DNA damage detection system to MutS homologues (40).

No mutations or polymorphisms were identified in the HMGB2, PPID, or MORF4 genes in the 10 individuals and three cell lines used. Full-length sequences of SPOCK3 were only obtained for five patients, including representatives from FamJ of the D4S1617-186 group and FamU of the D4S1597-282 group. No mutations were identified, although several known single nucleotide polymorphisms in both coding and noncoding regions were observed.

Haplotype analysis was possible in 41 FPC families; these were classified according to a set of working criteria for exclusion of families from disease linked to 4q32-34. Although exclusion based on these criteria is no guarantee that a family does not have a disease mutation in this region, there should have been a bias against exclusion if there was linkage in a significant number of families. Exclusion was observed in 8 of 16 families where a conclusion could be made (slightly more than would have been expected by chance). However, it is impossible to quantify how many 4q32-34 families there could be without giving bias against exclusion. There was some evidence of overrepresentation of particular combinations of microsatellite alleles in those families where exclusion was impossible. Two-point LOD score analysis did not support disease association with 4q32-34 but did allow relative levels of improbability to be defined. The least unlikely region corresponded to a region of clustered alleles in haplotypes that could not be excluded from linkage. These combinations of alleles enabled small groups of families to be defined, which might represent part of subpopulations with 4q32-34 linkage. Consistent with this, these subgroups did have distinct characteristics, but this could be explained by grouping of particular populations simply based on non-uniform distribution of particular alleles, although the families had no obvious racial connection. Candidate genes were sequenced in the families where linkage was most likely, but no mutations were identified. Nevertheless, mutations could have been missed, and we still cannot eliminate the possibility of linkage in a subset of families. It is unlikely that a majority of families have their disease linked to 4q32-34 (from the two-point LOD scores and the lack of any apparent bias against exclusion of families based on haplotypes). The subgroup classified as D4S1597-282 would seem the most likely grouping to have a disease gene within 4q32-34, as the low age of cancer death seems to mark this group out as significantly different from other pancreatic cancer families (consistent with the majority of families not being from this group). However, none of the individuals from the families within D4S1597-282 report diabetes, except as a direct consequence of their tumors. This contrasts strongly with the original 4q32-34 family described by Eberle et al. (29), where diabetes was common. In fact, none of the families on the combined European Registry of Hereditary Pancreatitis and Familial Pancreatic Cancer/German National Case Collection for Familial Pancreatic Cancer registry show an unusual incidence of diabetes, other than the diabetes associated with the onset of cancer.

In conclusion, it is impossible to rule out 4q32-34 in some of these families, but any gene mutated in this locus is unlikely to represent a predominant gene responsible for FPC in Europe.

Grant support: Cancer Research UK, Royal Liverpool University Hospital, Augustus Newman Foundation, the North West Cancer Research Fund, North West NHS Biomed Research Committee, Medical Research Council Gastroenterology and Pancreas Research Cooperative Grant, the Deutsche Krebshilfe 70-3085-Ba4 Germany.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Note: Supplementary data for this article are available at Cancer Epidemiology Biomakers and Prevention Online (http://cebp.aacrjournals.org/).

We thank the work undertaken by R. Mountford, J. Leslie, and N. Howes from Liverpool; Margarete Schneider (German National Case Collection for Familial Pancreatic Cancer study office); and the following who, in addition to the coauthors, have provided families and ongoing clinical information: M. Delhaye (Belgium); D. Kruger, L. Sunde (Denmark); L. Estévéz-Schwarz, W. von Bernstorff, M. Colombo-Benkmann, W. Böck, K. Breitschaft, S. Dülsner, T. Eberl, S. Eisold, E. Endlicher, M. Ernst, L. Estévéz-Schwarz, B. Gerdes, B. Ghadimi, T. Gress, R. Grützmann, J.W. Heise, O. Horstmann, L. Jochimsen, C. Jung, H. Messmann, R. Metzner, T. Mundel, K. Prenzel, O. Pridöhl, J. Rudolph, K.M. Schulte, C. Schleicher, J. Schmidt, K. Schulmann, H. Vogelsang, H. Witzigmann, N. Zügel (Germany); A. Oláh, V. Ruszinko (Hungary); D. Campra, G. Uomo, S. Pedrazzoli (Italy); A. Staka (Latvia); Å. Andrén-Sandberg (Norway); J. Jansen (The Netherlands); A. Brady, C. Brewer, J. Bennett, J. Booth, L. Botham, J. Cahill, B. Carmichael, C. Chapman, O. Claber, W. Crisp, T. Cole, J. Cook, L. Cowley, H. Cupples, B.R. Davidson, G. Davies, M. Deakin, H. Dorkins, D. Eccles, R. Eeles, I. Ellis, F. Elmslie, G. Evans, S. Fairgrieve, C. Faulkner, J. Foster, A. Howick, S. Hodgson, C. Imrie, L. Irvine, L. Izatt, M. James, C. Johnson, B. Kerr, D. Kumar, A. Laucassen, F. Lalloo, S. Laws, M. Lombard, D. Longdon, R. Loke, D. McBride, J. MacKay, E. Maher, M. Mehta, C. Mitchell, G. Mitchell, P. Morrison, L. O'Dair, K. Pape, J. Raeburn, L. Rae, C. Russell, E. Sheridan, J. Slavin, C. Smith, L. Snadden, G. Sobala, R. Sutton, J. Thomson, M. Tischkowitz, S. Tomkins, L. Walker, K. Wedgwood, P. Zack (United Kingdom and Ireland); E. Bjorck, E. Svarthol, J. Permert, I. Ihse (Sweden). Also Jan Schmidt for providing the FAMPAC cell line.

1
Jemal A, Murray T, Samuels A, et al. Cancer statistics, 2003.
CA Cancer J Clin
2003
;
53
:
5
–26.
2
Parkin DM, Bray FI, Devesa SS. Cancer burden in the year 2000. The global picture.
Eur J Cancer
2001
;
37
Suppl 8:
4
–66.
3
Bramhall SR, Allum WH, Jones AG, et al. Treatment and survival in 13,560 patients with pancreatic cancer, and incidence of the disease, in the West Midlands: an epidemiological study.
Br J Surg
1995
;
82
:
111
–5.
4
Fernandez E, La Vecchia C, D'Avanzo B, Negri E, Franceschi S. Family history and the risk of liver, gallbladder, and pancreatic cancer.
Cancer Epidemiol Biomarkers Prev
1994
;
3
:
209
–12.
5
Bartsch DK, Kress R, Sina-Frey M, et al. Prevalence of familial pancreatic cancer in Germany.
Int J Cancer
2004
;
110
:
902
–6.
6
Hemminki K, Li X. Familial and second primary pancreatic cancers: a nationwide epidemiologic study from Sweden.
Int J Cancer
2003
;
103
:
525
–30.
7
Wong T, Howes N, Threadgold J, et al. Molecular diagnosis of early pancreatic ductal adenocarcinoma in high-risk patients.
Pancreatology
2001
;
1
:
486
–509.
8
Silverman DT, Schiffman M, Everhart J, et al. Diabetes mellitus, other medical conditions and familial history of cancer as risk factors for pancreatic cancer.
Br J Cancer
1999
;
80
:
1830
–7.
9
Lynch HT, Brand RE, Lynch JF, Fusaro RM, Kern SE. Hereditary factors in pancreatic cancer.
J Hepatobiliary Pancreat Surg
2002
;
9
:
12
–31.
10
Giardiello FM, Brensinger JD, Tersmette AC, et al. Very high risk of cancer in familial Peutz-Jeghers syndrome.
Gastroenterology
2000
;
119
:
1447
–53.
11
Park JG, Park YJ, Wijnen JT, Vasen HF. Gene-environment interaction in hereditary nonpolyposis colorectal cancer with implications for diagnosis and genetic testing.
Int J Cancer
1999
;
82
:
516
–9.
12
The Breast Cancer Linkage Consortium. Cancer risks in BRCA2 mutation carriers.
J Natl Cancer Inst
1999
;
91
:
1310
–6.
13
Giardiello FM, Offerhaus GJ, Lee DH, et al. Increased risk of thyroid and pancreatic carcinoma in familial adenomatous polyposis.
Gut
1993
;
34
:
1394
–6.
14
Yamamoto H, Itoh F, Nakamura H, et al. Genetic and clinical features of human pancreatic ductal adenocarcinomas with widespread microsatellite instability.
Cancer Res
2001
;
61
:
3139
–44.
15
Goldstein AM, Struewing JP, Chidambaram A, Fraser MC, Tucker MA. Genotype-phenotype relationships in U S. melanoma-prone families with CDKN2A and CDK4 mutations.
J Natl Cancer Inst
2000
;
92
:
1006
–10.
16
Su GH, Hruban RH, Bansal RK, et al. Germline and somatic mutations of the STK11/LKB1 Peutz-Jeghers gene in pancreatic and biliary cancers.
Am J Pathol
1999
;
154
:
1835
–40.
17
Rogers CD, van der Heijden MS, Brune K, et al. The genetics of FANCC and FANCG in familial pancreatic cancer.
Cancer Biol Ther
2004
;
3
:
167
–9.
18
Petersen GM, de Andrade M, Goggins M, et al. Pancreatic cancer genetic epidemiology consortium.
Cancer Epidemiol Biomarkers Prev
2006
;
15
:
704
–10.
19
Greenhalf W, McFaul C, Earl J, et al. Anticipation in familial pancreatic cancer.
Gut
2006
;
55
:
252
–8.
20
Hahn SA, Greenhalf B, Ellis I, et al. BRCA2 germline mutations in familial pancreatic carcinoma.
J Natl Cancer Inst
2003
;
95
:
214
–21.
21
Murphy KM, Brune KA, Griffin C, et al. Evaluation of candidate genes MAP2K4, MADH4, ACVR1B, BRCA2 in familial pancreatic cancer: deleterious BRCA2 mutations in 17%.
Cancer Res
2002
;
62
:
3789
–93.
22
Grutzmann R, McFaul C, Bartsch DK, et al. No evidence for germline mutations of the LKB1/STK11 gene in familial pancreatic carcinoma.
Cancer Lett
2004
;
214
:
63
–8.
23
Bartsch DK, Sina-Frey M, Lang S, et al. CDKN2A germline mutations in familial pancreatic cancer.
Ann Surg
2002
;
236
:
730
–7.
24
Rulyak SJ, Lowenfels AB, Maisonneuve P, Brentnall TA. Risk factors for the development of pancreatic cancer in familial pancreatic cancer kindreds.
Gastroenterology
2003
;
124
:
1292
–9.
25
Rieder H, Sina-Frey M, Ziegler A, et al. German national case collection of familial pancreatic cancer: clinical-genetic analysis of the first 21 families.
Onkologie
2002
;
25
:
262
–6.
26
Horwitz M. The genetics of familial leukemia.
Leukemia
1997
;
11
:
1347
–59.
27
van der Heijden MS, Yeo CJ, Hruban RH, Kern SE. Fanconi anemia gene mutations in young-onset pancreatic cancer.
Cancer Res
2003
;
63
:
2585
–8.
28
Klein AP, Beaty TH, Bailey-Wilson JE, et al. Evidence for a major gene influencing risk of pancreatic cancer.
Genet Epidemiol
2002
;
23
:
133
–49.
29
Eberle MA, Pfutzer R, Pogue-Geile KL, et al. A new susceptibility locus for autosomal dominant pancreatic cancer maps to chromosome 4q32–34.
Am J Hum Genet
2002
;
70
:
1044
–8.
30
Lathrop GM, Lalouel JM. Easy calculations of lod scores and genetic risks on small computers.
Am J Hum Genet
1984
;
36
:
460
–5.
31
Broman KW, Murray JC, Sheffield VC, White RL, Weber JL. Comprehensive human genetic maps: individual and sex-specific variation in recombination.
Am J Hum Genet
1998
;
63
:
861
–9.
32
Eisold S, Ryschich E, Linnebacher M, et al. Characterization of FAMPAC, a newly identified human pancreatic carcinoma cell line with a hereditary background.
Cancer
2004
;
100
:
1978
–86.
33
Wynendaele W, Derua R, Hoylaerts MF, et al. Vascular endothelial growth factor measured in platelet poor plasma allows optimal separation between cancer patients and volunteers: a key to study an angiogenic marker in vivo?
Ann Oncol
1999
;
10
:
965
–71.
34
Liu SH, Lin CY, Peng SY, et al. Down-regulation of annexin A10 in hepatocellular carcinoma is associated with vascular invasion, early recurrence, and poor prognosis in synergy with p53 mutation.
Am J Pathol
2002
;
160
:
1831
–7.
35
Doyle V, Virji S, Crompton M. Evidence that cyclophilin-A protects cells against oxidative stress.
Biochem J
1999
;
341
:
127
–32.
36
Riggs DL, Roberts PJ, Chirillo SC, et al. The Hsp90-binding peptidylprolyl isomerase FKBP52 potentiates glucocorticoid signaling in vivo.
EMBO J
2003
;
22
:
1158
–67.
37
Galigniana MD, Harrell JM, Murphy PJ, et al. Binding of hsp90-associated immunophilins to cytoplasmic dynein: direct binding and in vivo evidence that the peptidylprolyl isomerase domain is a dynein interaction domain.
Biochemistry
2002
;
41
:
13602
–10.
38
Tominaga K, Olgun A, Smith JR, Pereira-Smith OM. Genetics of cellular senescence.
Mech Ageing Dev
2002
;
123
:
927
–36.
39
Kokawa A, Kondo H, Gotoda T, et al. Increased expression of cyclooxygenase-2 in human pancreatic neoplasms and potential for chemoprevention by cyclooxygenase inhibitors.
Cancer
2001
;
91
:
333
–8.
40
Krynetski EY, Krynetskaia NF, Bianchi ME, Evans WE. A nuclear protein complex containing high mobility group proteins B1 and B2, heat shock cognate protein 70, ERp60, and glyceraldehyde-3-phosphate dehydrogenase is involved in the cytotoxic response to DNA modified by incorporation of anticancer nucleoside analogues.
Cancer Res
2003
;
63
:
100
–6.

Supplementary data