1 Dipartimento di Oncologia, Biologia e Genetica (DOBiG), Università degli Studi di Genova, Genova; 2 Istituto Nazionale per la Ricerca sul Cancro (IST), Genova, Italy
Received 21 May 2003; revised 20 June 2003; accepted 12 August 2003
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Roughly 40% of germinal mutations in melanoma families (MF) affect p16INK4a and p14ARF. We investigated the association between INK4/ARF alterations and the occurrence of pancreatic cancer in MF and in sporadic pancreatic cancer (SPC) patients.
Patients and methods:
Forty-nine MF, 66 SPC cases and 54 controls were enrolled. The INK4/ARF locus was screened.
Results:
As compared with the general population, the risk of pancreatic cancer (PC) was increased 9.4-fold [95% confidence interval (CI) 2.733.4] and 2.2-fold (95% CI 0.85.7) in G101W-positive and -negative MF, respectively, while mean ages at onset were 61 and 77 years, respectively. A 1.7 (95% CI 1.062.79) increased risk of cancer at any site was observed among first-degree relatives of SPC cases as compared with controls. The G101W founder mutation was detected in 4% of SPC cases but the rate increased to 13% when tumor clustering in either branch of families was taken into account. One G101W-positive PC patient with a melanoma in a first-degree relative harbored a germline deletion of the second allele, including exon 1B.
Conclusions:
The presence of a deletion including exon 1B in two PC patients points to the involvement of p14ARF in the development of PC and may suggest that the increased risk of PC in MF is caused by impairment of both loci.
Key words: CDKN2A, familial melanoma, G101W mutation, increased risk, p14ARF, pancreatic cancer
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Pancreatic cancer has been reported to be associated with various environmental and lifestyle risk factors, occupational exposures and medical conditions; however, the only risk factors consistently reported are age and smoking status, and the etiology of the disease remains largely unknown [2].
A positive family history of pancreatic cancer [35] and a history of other cancers among first-degree relatives have been established as important risk factors for the disease [6, 7]. Moreover, the presence of multiple primary tumors in pancreatic cancer patients [8] or early age at onset [9] may suggest a genetic predisposition to the development of malignancies. Several genetic syndromes, such as ataxia telangiectasia, PeutzJeghers syndrome, hereditary nonpolyposis colorectal cancer and familial atypical mole-malignant melanoma syndrome (FAMMM), are also associated with a significantly increased risk of developing pancreatic cancer [10]. In addition, an excess of pancreatic cancer has been found to occur in breastovarian cancer families carrying BRCA1 and BRCA2 mutations [11].
Among the susceptibility genes involved in the syndromes mentioned above, CDKN2A/p16INK4a seems to play an important role. p16INK4a is a negative regulator of cell cycle progression at the G1S restriction point and is frequently altered in sporadic pancreatic adenocarcinomas. Virtually all pancreatic carcinomas show inactivation of the CDKN2A gene [12, 13]. An increased risk of developing pancreatic adenocarcinoma was found in melanoma and FAMMM kindreds carrying CDKN2A mutations; no excess of pancreatic adenocarcinoma occurred in kindreds with wild-type CDKN2A [1418].
We previously described a small number of familial melanoma kindreds from a very small area in Liguria, a northwestern region of Italy [19, 20], where the p16INK4a G101W germline founder mutation is prevalent [21]. Seven of those kindreds were found to carry the G101W mutation and displayed an increased risk of pancreatic cancer. An increased risk of pancreatic cancer in melanoma families has been reported in other founder populations: Borg et al. [22] showed that the 113insArg founder mutation was associated with an increased risk of multiple melanomas, breast cancer and pancreatic cancer; Vasen et al. [23] reported that putative mutation carriers belonging to Dutch FAMMM kindreds had a 17% cumulative risk of developing pancreatic cancer by 75 years of age.
The majority of mutations which have been associated with an increased risk of pancreatic cancer are located on exon 2 [1420] and are thus predicted to affect the amino acid sequence of p14ARF (due to alternative splicing of exon 1B with CDKN2A exon 2) [24]. Additionally, although few families have been reported to harbor mutations in p14ARF which predispose to melanoma [25, 26], a germline deletion affecting p14 but not CDKN2A in a melanoma-neural system tumor syndrome family has recently been identified [27]. Therefore, the involvement of the INK4/ARF locus in pancreatic cancer development deserves further study. To this purpose, and more specifically to evaluate the role of the G101W founder mutation in the development of pancreatic cancer among melanoma-prone kindreds, we extended our previous analysis to a greater number of families and to a pool of unselected pancreatic cancer patients.
![]() |
Patients and methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Features of melanoma families. The 49 melanoma families from Liguria were comprised of 967 relatives, including 118 melanoma cases. Twelve of 118 melanoma cases were excluded because information on the age of diagnosis of melanoma was not recorded (n = 4) or diagnosis was earlier than 1958 (n = 8) when no written documentation was available prior to that date. Thus, 106 familial melanoma patients diagnosed from January 1959 to December 2000 were studied from a total of 955 relatives, 366 of which belonged to the unaffected branches and were reclassified as controls (at date of analysis, 207 alive and 159 deceased); of the 589 members of the affected branches, 409 were alive and 180 deceased.
The 18 G101W families and 31 non-mutated families comprised 219 and 370 members, respectively. A total of 132 members were screened for the mutation (52 from the G101W families and 80 from non-mutated families). Thirty-four members of the 18 G101W-positive kindreds were found to harbor the mutation. The mean ages at diagnosis of melanoma in the G101W families and mutation-negative kindreds were 46.5 (median, 42; range, 2092) and 49.4 years (median 49, range 1685), respectively.
Verification of diagnosis
Information on melanoma patients was collected from histopathological and clinical records from different hospitals in Liguria. Once an affected family member was identified, access to the medical file and/or autopsy report was requested. If medical files and/or autopsy reports were not available, information on malignancy was obtained from the Ligurian Cancer Registry [29], which holds incidence figures from 1986 to 1996. Information on family history of pancreatic cancer was reported by patients during genetic counseling. Four out of 12 of those reports were confirmed. For two cases diagnosed before 1986 or after 1996 (two of four patients) confirmation was provided by declarations of a general physician (one patient) or an oncologist (one patient). The other two were confirmed by the local cancer registry. The remaining eight cases were not confirmed either because they were diagnosed before 1986 (five patients), or because the patients were diagnosed in another region and no documentation was available (three patients).
Hereditary susceptibility to cancer in unselected pancreatic cancer patients from Liguria
The study was designed as a hospital-based casecontrol study at the National Cancer Institute (NCI) in Genoa, the largest town in Liguria, accounting for roughly 50% of the regions total population.
Cases. Eligible patients were those with a diagnosis of pancreatic cancer 18 years of age who were referred to the Gastrointestinal Unit of the National Institute for Cancer Research (NCI) of Genoa for endoscopic retrograde cholangio-pancreatography (ERCP). In all, 66 patients (36, male; 30, female) were interviewed. A histological diagnosis of exocrine pancreatic cancer was available for 23 patients (34.8%) and a cytological diagnosis was obtained for six patients (9.1%). In the remaining cases, the diagnosis was made on the basis of a diagnostic laparotomy [18 cases (27.2%)] or imaging [19 cases (28.8%)].
Two patients (one male, one female) were excluded from the analysis, as the diagnosis of pancreatic cancer was not confirmed, one male was excluded because the pathological examination showed an endocrine carcinoma and two cases (both males) were excluded because they were unable to report their family history of cancer, leaving 61 cases available for analysis.
Controls. Controls were identified at the San Martino General Hospital, Genoa, which is located close to the NCI. Subjects were eligible as controls if they had been admitted for diseases other than cancer and unrelated to digestive tract disease or to known risk factors for pancreatic cancer. Controls were matched to cases by gender, age (±5 years) and region of residence. In all, 54 eligible controls were identified and interviewed. The main diagnostic categories for hospital admission were acute surgical (n = 19) and musculoskeletal diseases (n = 35).
A trained interviewer administered a standardized questionnaire to cases and controls. They were asked about their own disease, cancer occurrence in the family (type and site of primary tumor), age at diagnosis among first-, second- and third-degree blood-relatives. Cancer site among relatives was recorded as reported by cases and controls and no attempt was made to verify the diagnoses. An in-person interview was conducted with 48 of 61 (78.7%) cases and with 52 of 54 (96.3%) controls, for the others a family member was interviewed.
Cancer clustering in the family was hypothesized when the index case (or control) reported a history of invasive cancer in at least two blood relatives (first and/or second degree) belonging to the same branch of the family.
Cases and controls received detailed information about the purpose of the study. Written informed consent concerning participation in the study, potential risks and benefits, use of hospital information and DNA analyses was obtained. All subjects who agreed to participate in the study were asked to provide a 20 ml sample of peripheral blood to perform molecular analyses. Cases and controls were interviewed while still at the hospital. The study was approved by the Ethics Committee of the NCI of Genoa.
Molecular analysis
INK4/ARF mutation analysis in melanoma families and pancreatic cancer patients. DNA was extracted from peripheral blood leukocytes using standard procedures. Specific PCR products of each of the four exons at the locus and at the intronexon junctions were screened for mutations by single strand conformational polymorphism (SSCP) analysis, followed by direct sequencing of the variants. Primer sequences and screening conditions for exons 1, 2 and 3 of CDKN2A have been previously described [28]. Thirty-one probands from CDKN2A-negative families were screened for alternative exon 1ß (encoding p14ARF) mutations and for the recently described IVS2-105A/G deep intronic mutation [30] by PCR and SSCP. Primers used for exon 1ß amplification (447 bp) were p14F, 5'-TCCCAGTCTGCAGTTAAGGGG-3' and p14R, 5'-GTCTAAGTCGTTGTAACCC-3; the deep intronic variant was investigated using primers 105F, 5'-GGGTACATGCACGTGAAGC-3' and 105R, 5'-GTTTCTCAGAGCCTCTCTGGT (designed to amplify a 182 bp fragment). Positive controls (kind gift of Dr S. Puig, Barcelona, Spain) for exon 1ß 60ins16 mutation and for the IVS2-105A/G variant (kind gift of Dr M. Harland, Leeds, UK) were included for each experiment. Briefly, 60 ng DNA were amplified in a final volume of 15 µl containing 5 pmol primers, 200 µM dNTPs and 1.5 mM MgCl2, 0.5 U TaqGold (Applied Biosystem, Foster City, CA) in a PE2400 mastercycler with an amplification profile of 95°C for 12 min, followed by 35 cycles at 95°C for 50 s, 60°C for 30 s, 72°C for 1 min and a final extension at 72°C for 7 min.
DNA from 48 pancreatic cancer patients was screened for both CDKN2A (exons 1 and 2) and exon 1B mutations as described above. Thirty-five controls were also screened for CDKN2A mutations.
Microsatellite analysis at the INK4/ARF locus. Microsatellite analysis at the INK4/ARF locus was performed using markers (IFNA, D9S736, D9S1749, D9S974, D9S942, D9S1748, D9S1604, D9S171, all primer oligonucleotide sequences are available at http://www.gdb.org). For each primer pair, one oligonucleotide was labeled with a 5' fluorescent amidite (6' FAM or HEX). Amplification was carried out using TaqGold (Applied Biosystems) and PCR products were visualized using an ABI 310 automated DNA sequencer (Applied Biosystems).
Statistical analysis
Melanoma families. The incidence of pancreatic cancers was compared with the expected incidence in the general population and among spouses of those affected (control group). The number of person-years of observation was calculated for each subject from birth until death, diagnosis of cancer, or the last known time at which the subject was alive and cancer-free. Person-years at risk were stratified according to gender and age (5-year intervals) and p16 status (mutated or wild type). The number of expected cases, by age (5-year intervals) and gender, was calculated from Ligurian Cancer Registry data (19931996) [29]. The expected number of pancreatic cancer cases was calculated in each stratum by multiplying the stratum-specific incidence rate of pancreatic cancer in the Ligurian Cancer Registry by the number of person-years of observation in the study population (mutation carriers, wild type, controls). We calculated the crude incidence ratio (ratio of observed to expected number of cases); the expected number of cases was summarized to yield the total number of cases expected on the basis of the Cancer Registry. The 95% confidence intervals (CI) were calculated assuming a Poisson distribution for the observed number of melanoma or pancreatic cancer cases in the study population.
The chi square test was used to test for differences among proportions and comparison of age between groups was performed using the MannWhitney test.
The age-specific risk of developing pancreatic cancer was calculated according to the KaplanMeier method for survival analysis [31]. The end point was the diagnosis of pancreatic cancer, death or end of follow-up (December 2000) or the last date when the patient was known to be alive.
Casecontrol study. The relative risk of cancer among relatives of cases of pancreatic cancer and of hospital controls was estimated using the Cox proportional hazard method [32]. When appropriate, gender was introduced in the model as a covariate.
For all statistical tests a significance level of 0.05 was used, and all reported P values are two-sided. All analyses were performed using the statistical package SPSS for Windows, release 5 (SPSS, Inc., Chicago, IL).
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
In the three groups, other cancers were reported in 38 (10.4%), 39 (10.5%) and 19 subjects (8.7%), respectively. After melanoma, pancreatic adenocarcinoma was the second most frequent cancer in G101W-positive families.
The risk of developing pancreatic cancer among our melanoma kindreds is shown in Table 1. A statistically significant 4-fold increased risk of pancreatic cancer was observed among subjects belonging to the melanoma families (95% CI 2.27.5). However, when the mutational status of the family was considered, the risk of pancreatic cancer in the wild-type families fell to 2.2 (95% CI 0.85.7), which was not statistically significant, while among the mutated families the risk rose to 9.4 (95% CI 2.733.4). The Poisson regression analysis provided a 2.1-fold increased risk of pancreatic cancer (95% CI 0.63.2) for subjects belonging to a melanoma family which was not statistically significant. The observed risk associated with the presence of the G101W mutation in the family was 4.4 (95% CI 1.215.5).
|
|
Hereditary susceptibility to cancer in unselected pancreatic cancer patients
A total of 61 pancreatic cancer patients (case probands) and 54 controls (control probands) were available for analysis. Cases comprised 36 males and 25 females; the mean age at diagnosis was 64.6 years [standard deviation (SD) 10.9; range 4183]; 67.2% were born in northern Italy. Among case probands, two patients (case nos 101 and 111) reported that they had had a histologically confirmed cutaneous malignant melanoma removed 4 and 10 years before the diagnosis of pancreatic cancer, respectively. Controls comprised 32 males and 22 females; the mean age at the time of interview was 64.4 years (SD 10.4; range 4279); 85.2% were born in northern Italy. In all, cases and controls reported 1521 relatives, whose features are summarized in Table 2.
|
The estimated RR of cancer at any site was 1.72 (95% CI 1.062.79; P = 0.028) for first-degree relatives and 1.06 (95% CI 0.671.67; P = 0.81) for second-degree relatives of cases as compared with those of controls. Concerning the most common cancer sites, first-degree relatives of cases showed a non statistically significant 2-fold increased risk of cancer of the large bowel (10 cases and four controls, RR = 2.3; 95% CI 0.77.4), lung (five cases and two controls, RR = 2.3; 95% CI 0.411.9), uterus (unspecified) (five cases and two controls, RR = 2.1; 95% CI 0.411.0) and aerodigestive tract (four cases and two controls, RR = 2.3; 95% CI 0.411.7) as compared with relatives of controls. As for rare cancers, two cases of pancreatic cancer, one case of melanoma and one of rhabdomyosarcoma were reported among first-degree relatives of cases. None of these three tumor types were recorded among first-degree relatives of controls.
We further evaluated the individual pedigree of cases and controls in order to identify cancer clustering. Separately, we analyzed the paternal and maternal lineage in the attempt to understand whether the clusters were consistent with a genetic predisposition to the development of tumors. A clustering was identified in 19 branches out of 61 case families (31.1%), and in five of 54 (9.3%) control families (P = 0.004).
In nine of 19 branches, we observed additional primary tumors which are not inconsistent with a partial spectrum of family cancer syndromes (melanoma-pancreas, HNPCC, breast-ovarian), or pancreatic cancer in patients with a previous diagnosis of cutaneous malignant melanoma (in two patients) or familiality for pancreatic cancer (one patient with a first-degree relative affected by pancreatic cancer) (Table 3) (Figure 2). Pancreatic cancer was diagnosed at a younger age among patients belonging to branches with clustering (mean age at diagnosis, 60.4 years; range 4182 years) than among those belonging to the branches with no clustering (mean age at diagnosis, 66.6 years; range 4783) (Z = 1.987; P = 0.047).
|
|
Out of 48 pancreatic cancer patients screened for CDKN2A and exon 1B mutations, two were found to harbor the CDKN2A G101W mutation. No mutation was found in the 35 controls tested. The two pancreatic cancer patients carrying the p16 mutation showed an association of pancreatic cancer with other tumors in first and second-degree relatives (prostate, colon, larynx and melanoma) (Figure 2B). A cumulative 4% mutation frequency was found among all pancreatic cancer cases analyzed, but the frequency increased to 13% (2 of 15 cluster families tested) when we observed pancreatic cancer patients from cluster families alone.
Analysis of markers flanking the CDKN2A locus showed that genotypes of the two pancreatic cancer patients carrying the G101W mutation were consistent with the previously reported G101W core haplotype [21], although the phase could not be determined.
One of the two patients harboring the mutation revealed a region of hemizygosity involving four markers (D9S942 to D9S171), thus suggesting a germline deletion of the wild-type allele containing exon 1B in this patient. The two patients affected by melanoma and pancreatic cancer, and found to be negative at CDKN2A, exon 1B and deep intronic variant screening, were also typed for microsatellite analysis. One of them carried a deletion involving markers D9S942 to D9S1604 (Table 4).
|
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The Ligurian melanoma families described here showed a 9-fold increase in the risk of developing pancreatic cancer as compared with the general Ligurian population. These results are comparable with those reported by Vasen et al. [23]: the mean age at diagnosis of pancreatic cancer was 57 years in our study and 58 years in the Leiden study, and the cumulative risk of pancreatic cancer among non-mutated families did not differ from that in the general population. The cumulative risk among the Dutch families was higher than that observed here (17% versus 6.7%). This may be due to the impact of geneenvironment or genegene interactions on mutations which in both cases are predicted to affect the role of p14ARF.
As it is almost impossible to test relatives affected by pancreatic cancer in a retrospective studydue to the very poor prognosis of these patientsour analysis may suffer from a lack of genotypic information on pancreatic cancer cases. Hence, we chose to investigate family history and INK4/ARF mutations in unselected pancreatic cancer cases from the same geographic area.
All pancreatic cancer cases included in the casecontrol study were selected and reviewed by the same gastroenterologist (V.P.) who carried out the ERCP analysis. A histological or cytological diagnosis of pancreatic cancer was obtained for only 45.9% (28 of 61) of the cases enrolled in the study. So, in spite of a careful review of the clinical records, we cannot exclude misclassification of some cases. In addition, 13 patients underwent palliative surgery because of inoperable disease, and in the surgical report, a pancreatic mass involving the contiguous organs was recorded. The diagnosis was based on imaging alone in 19/61 patients (31.1%). Nonetheless, the rate of confirmation of pancreatic cancer in epidemiological studies rarely exceeds 70% [6, 33], and in a subset of an international casecontrol study of the International Agency for Research on Cancer (IARC) [34], the rate was as low as 43%.
Large casecontrol studies carried out in Italy estimated that roughly 3% of newly diagnosed pancreatic cancers are consistent with the existence of a genetic component in the familial aggregation of pancreatic cancer [35]. In our study population, the observed frequency of first-degree family history of pancreatic cancer among cases was 3.3%, which was slightly lower than that reported by other epidemiological studies [3, 7]. None of the controls reported family history of pancreatic cancer; this may be due to the low expected frequency (0.61.5%) of pancreatic cancer among relatives of healthy subjects [3, 7]. It should be noted that information on the family history of cancer was collected during the interview and no attempt was made to verify the reported information. Until two decades ago, the diagnosis of pancreatic cancer was difficult because of a lack of technical devices, in addition to the massive invasion of the contiguous organs at diagnosis very frequently precluding histological diagnosis. Therefore, we cannot exclude the fact that relatives with pancreatic cancer may have been reported to have died of some other abdominal cancer (most frequently, liver cancer). However, we can rule out the recall bias among cases, as most were interviewed shortly after their ERCP examination when the diagnosis of pancreatic cancer had not yet been confirmed. In our casecontrol study, we observed that first-degree relatives of patients with pancreatic cancer had a statistically significant 1.7-fold increased risk of developing cancer at any site as compared with controls. The cumulative incidence of cancer (all sites) among first-degree relatives of patients with pancreatic cancer was 35% and among relatives of controls it was 18.8%. When we considered the most frequent sites of neoplastic disease, we found that first-degree relatives of patients with pancreatic cancer had a 2-fold increased risk of developing cancer located in the large bowel, lung, liver, uterus and aerodigestive tract. None of these risks were statistically significant, but the small size of our study population affects the power of the study as well as the precision of the risk estimates.
A significantly higher frequency of tumor aggregations involving both first- and second-degree relatives was observed among cases as compared with controls (19 versus seven). In these families, all the affected relatives belonged to the same familial branch, and in some cases the observed aggregation partially resembled well-known cancer syndromes, although not all the established criteria for identification were met (Table 3). Although we are currently unable to establish the precise significance of this preliminary finding, it may suggest that other gene mutations (BRCA1 and -2, p53, mismatch repair genes, etc.) are possibly involved and that such kindreds could therefore be usefully tested in order to establish the proportion of pancreatic cancers attributable to different cancer syndromes.
On the basis of the distribution of age at diagnosis of pancreatic cancer in our G101W families (57.5 years) and in the present subset of apparently sporadic pancreatic cancer (60.5 years) who show a tumor clustering in the family, this feature could be reasonably hypothesized to predict positive mutation status and the melanomapancreas syndrome, as already reported [14, 23, 36].
p16 mutations have been found in patients with both pancreatic cancer and multiple primaries, including melanoma, as well in pancreatic cancer patients from families with pancreatic cancer and melanoma [7, 37, 38]. The two patients in our series who developed both melanoma and pancreatic cancer appeared to be carrying no CDKN2A mutations; however, the presence of a region of hemizygosity (D9S942 to D9S1604) including exon 1B (but not CDKN2A), in one of two patients points to the involvement of p14ARF alterations in susceptibility to pancreatic cancer. Indeed, susceptibility to multiple cancers may be associated to different, contiguous tumor suppressor gene deletions, as reported for proneness to tumors of the nervous system in melanoma kindreds [27, 39].
The p16 G101W mutation recurrent in Liguria was detected in two pancreatic cancer patients out of 15 cluster families tested (13%). The presence of the conserved microsatellite core (markers D9S974 to D9S1604) confirmed the same ancestral origin previously described for this mutation in melanoma families [21]. Interestingly, the younger (48 years) of the two showed the same region of hemizygosity described above (Figure 2B and Table 4). This could point to the presence in this patient of a single G101W allele, the exon 2 mutation also affecting p14ARF, and consequently suggest that absence of any wild-type p14ARF may impact on pancreatic tumorigenesis. Furthermore, impairment of p14ARF by G101W mutation translates into Arg116Cys, which in turn leads to changes in basic arginine residues within the nucleolar potential localization domain and hdm2 interaction [25].
Further characterization of the described deletion together with expression studies may further clarify whether this deletion is responsible for pancreatic cancer, or whether susceptibility to multiple cancers arises from different deletions involving tumor suppressor genes.
![]() |
Acknowledgements |
---|
![]() |
FOOTNOTES |
---|
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
2. Ghadirian P, Lynch HT, Krewski D. Epidemiology of pancreatic cancer: an overview. Cancer Detect Prev 2003; 27: 8793.[CrossRef][ISI][Medline]
3. Lynch HT, Brand RE, Deters CA et al. Hereditary pancreatic cancer. Pancreatology 2001; 1: 466471.[CrossRef][ISI][Medline]
4. Ghadirian P, Liu G, Gallinger S et al. Risk of pancreatic cancer among individuals with a family history of cancer of the pancreas. Int J Cancer 2002; 97: 807810.[CrossRef][ISI][Medline]
5. Klein AP, Hruban RH, Brune KA et al. Familial pancreatic cancer. Cancer J 2001; 7: 266273.[ISI][Medline]
6. 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: 18301837.[CrossRef][ISI][Medline]
7. Ghadirian P, Boyle P, Simard A et al. Reported family aggregation of pancreatic cancer within a population-based casecontrol study in the Francophone community in Montreal, Canada. Int J Pancreatol 1991; 10: 183196.[ISI][Medline]
8. Gerdes B, Bartsch DK, Ramaswamy A et al. Multiple primary tumors as an indicator for p16INK4a germline mutations in pancreatic cancer patients? Pancreas 2000; 21: 369375.[CrossRef][ISI][Medline]
9. Schenk M, Schwartz AG, ONeal E et al. Familial risk of pancreatic cancer. J Natl Cancer Inst 2001; 93: 640644.
10. Hruban RH, Petersen GM, Ha PK, Kern SE. Genetics of pancreatic cancer. From genes to families. Surg Oncol Clin N Am 1998; 7: 123.[Medline]
11. Gayther SA, Mangion J, Russell P et al. Variation of risks of breast and ovarian cancer associated with different germline mutations of the BRCA2 gene. Nat Genet 1997; 15: 103105.[ISI][Medline]
12. Caldas C, Hahn SA, da Costa LT et al. Frequent somatic mutations and homozygous deletions of the p16 (MTS1) gene in pancreatic adenocarcinoma. Nat Genet 1994; 8: 2732.[ISI][Medline]
13. Schutte M, Hruban RH, Geradts J et al. Abrogation of the Rb/p16 tumor-suppressive pathway in virtually all pancreatic carcinomas. Cancer Res 1997; 57: 31263130.[Abstract]
14. Goldstein AM, Fraser MC, Struewing JP et al. Increased risk of pancreatic cancer in melanoma-prone kindreds with p16INK4 mutations. N Engl J Med 1995; 333: 970974.
15. Bergman W, Watson P, de Jong J et al. Systemic cancer and the FAMMM syndrome. Br J Cancer 1990; 61: 932936.[ISI][Medline]
16. Lynch HT, Brand RE, Deters CA et al. Update on familial pancreatic cancer. Curr Gastroenterol Rep 2001; 3: 121128.[Medline]
17. Whelan AJ, Bartsch D, Goodfellow PJ. Brief report, a familial syndrome of pancreatic cancer and melanoma with a mutation in the CDKN2 tumor-suppressor gene. N Engl J Med 1995; 333: 975977.
18. Goldstein AM, Struewing JP, Chidambaram A et al. Genotypephenotype relationships in U.S. melanoma-prone families with CDKN2A and CDK4 mutations. J Natl Cancer Inst 2000; 92: 10061010.
19. Ciotti P, Strigini P, Bianchi-Scarra G. Familial melanoma and pancreatic cancer. Ligurian Skin Tumor Study Group. N Engl J Med 1996; 334: 469470.
20. Ghiorzo P, Ciotti P, Mantelli M et al. Characterization of Ligurian melanoma families and risk of occurrence of other neoplasia. Int J Cancer 1999; 83: 441448.[CrossRef][ISI][Medline]
21. Ciotti P, Struewing JP, Mantelli M et al. A single genetic origin for the G101W CDKN2A mutation in 20 melanoma-prone families. Am J Hum Genet 2000; 67: 311319.[CrossRef][Medline]
22. Borg A, Sandberg T, Nilsson K et al. High frequency of multiple melanomas and breast and pancreas carcinomas in CDKN2A mutation-positive melanoma families. J Natl Cancer Inst 2000; 92: 12601266.
23. Vasen HF, Gruis NA, Frants RR et al. Risk of developing pancreatic cancer in families with familial atypical multiple mole melanoma associated with a specific 19 deletion of p16 (p16-Leiden). Int J Cancer 2000; 87: 809811.[CrossRef][ISI][Medline]
24. Chin L, Pomerantz J, DePinho RA. The INK4a/ARF tumor suppressor: one genetwo productstwo pathways. Trends Biochem Sci 1998; 23: 291296.[CrossRef][ISI][Medline]
25. Rizos H, Darmanian AP, Holland EA et al. Mutations in the INK4a/ARF melanoma susceptibility locus functionally impair p14ARF. J Biol Chem 2001; 276: 4142441434.
26. Hewitt C, Lee Wu C, Evans G et al. Germline mutation of ARF in a melanoma kindred. Hum Mol Genet 2002; 11: 12731279.
27. Randerson-Moor JA, Harland M, Williams S et al. A germline deletion of p14(ARF) but not CDKN2A in a melanoma-neural system tumour syndrome family. Hum Mol Genet 2001; 10: 5562.
28. Mantelli M, Barile M, Ciotti P et al. High prevalence of the G101W germline mutation in the CDKN2A (P16Ink4a) gene in 62 Italian malignant melanoma families. Am J Med Genet 2002; 107: 214221.[CrossRef][ISI][Medline]
29. Zanetti R, Gafà L, Pannelli F et al. (eds). Il cancro in Italia. I dati di incidenza dei Registri Tumori 19931998. Roma, Italy, Il Pensiero Scientifico 2002; 249259.
30. Harland M, Mistry S, Bishop DT, Bishop JA. A deep intronic mutation in CDKN2A is associated with disease in a subset of melanoma pedigrees. Hum Mol Genet 2001; 10: 26792686.
31. Kaplan EL, Meyer P. Non parametric estimation for incomplete observations. J Am Stat Assoc 1958; 53: 457474.[ISI]
32. Cox DR. Regression model and life-tables. J R Stat Soc B 1972; 34: 187220.[ISI]
33. Gullo L, Pezzilli R, Morselli-Labate AM. Diabetes and the risk of pancreatic cancer. The Italian Pancreatic Cancer Study Group. N Engl J Med 1994; 331: 8184.
34. Boyle P, Maisonneuve P, Bueno de Mesquita B et al. Cigarette smoking and pancreas cancer: a casecontrol study of the search programme of the IARC. Int J Cancer 1996; 67: 6371.[CrossRef][ISI][Medline]
35. Fernandez E, La Vecchia C, Decarli A. Attributable risks for pancreatic cancer in northern Italy. Cancer Epidemiol Biomarkers Prev 1996; 5: 2327.[Abstract]
36. Lynch HT, Brand RE, Hogg D et al. Phenotypic variation in eight extended CDKN2A germline mutation familial atypical multiple mole melanoma-pancreatic carcinoma-prone families: the familial atypical mole melanoma-pancreatic carcinoma syndrome. Cancer 2002; 94: 8496.[CrossRef][ISI][Medline]
37. Lal G, Liu L, Hogg D et al. Patients with both pancreatic adenocarcinoma and melanoma may harbor germline CDKN2A mutations. Genes Chromosom Cancer 2000; 27: 358361.[CrossRef][ISI][Medline]
38. Bartsch DK, Sina-Frey M, Lang S et al. CDKN2A germline mutations in familial pancreatic cancer. Ann Surg 2002; 236: 730737.[CrossRef][ISI][Medline]
39. Bahuau M, Vidaud D, Jenkins RB et al. Germ-line deletion involving the IKNK4 locus in familial proneness to melanoma and nervous system tumors. Cancer Res 1998; 58: 22982303.[Abstract]