Establishing a clinic-based pancreatic cancer and periampullary tumour research registry in Quebec

A.L. Smith bsc,*† C. Bascuñana msc,*† A. Hall msc,*† A. Salman msc,*‡ A.Z. Andrei bsc,*† A. Volenik msc,*†§ H. Rothenmund msc,*†§ D. Ferland ba bscn,*‡ D. Lamoussenery bscn,‡|| A.S. Kamath md,‡ R. Amre md mbbs,# D. Caglar md,# Z.H. Gao md phd,# D.G. Haegert md,# Y. Kanber md,# R.P. Michel bsc mdcm,# G. Omeroglu–Altinel md,# J. Asselah md,|| N. Bouganim md,|| P. Kavan md phd,|| G. Arena md,‡ J. Barkun md msc,‡ P. Chaudhury md msc,‡ S. Gallinger md msc,** W.D. Foulkes mbbs phd,*§ A. Omeroglu md,# P. Metrakos md,*‡ and G. Zogopoulos md phd*†‡§


INTRODUCTION
Pancreatic ductal adenocarcinoma (pdac) is a lethal malignancy with a 5-year survival rate of only 6% [1][2][3] .It represents the 4th leading cause of cancer-related death in Quebec, with an estimated 1290 new cases diagnosed and 1170 deaths in 2014 1 .These dismal To meet this research resource need in Quebec, we established a prospective clinic-based research registry, the Quebec Pancreas Cancer Study (qpcs).We used a clinic-based approach with a rapid ascertainment strategy because we hypothesized that those methods would facilitate high participation rates.The clinical and epidemiologic variables collected by the qpcs and the opcs overlap, with the goal of fostering research synergies between those two Canadian studies.
Although the primary objective of the qpcs is to establish a Quebec research resource for genetic susceptibility studies of pdac, the data and biospecimens collected will also allow for future epidemiologic, biomarker, and cancer biology studies of pdac.In addition, the qpcs includes cases of other periampullary tumours (pats) that are also rare, underrepresented in research studies, and often difficult to treat.Characterization of such cases could prove helpful in advancing the understanding of pdac and in contributing to more global studies of biliary and gastrointestinal precancerous and cancerous conditions 9,10 .
Here, we describe the methods used to establish the qpcs and the results obtained after the first 374 referrals, with an emphasis on pdac.

2.1.1
Inclusion Criteria Men and women more than 18 years of age with a diagnosis of pdac or pat were eligible to participate.Patients with disease confirmed by histology or cytology were included, as were patients diagnosed based on clinical and axial imaging when a biopsy was either unavailable or pending.Unaffected individuals with a family history of pdac or a diagnosis of a pdacassociated genetic syndrome (hereditary breast and ovarian cancer syndrome, hereditary pancreatitis, hereditary nonpolyposis colorectal cancer, Peutz-Jeghers syndrome, or familial atypical multiple-mole melanoma) 5 were also eligible to participate.
Although pdac is the primary focus of the qpcs, the registry was expanded in October 2012 to include pats, including pancreatic neuroendocrine tumours, solid pseudo-papillary epithelial neoplasms, premalignant lesions of the pancreas (for example, intraductal papillary mucinous neoplasms, mucinous cystic neoplasms), and preinvasive and malignant biliary, ampullary, gastric, and duodenal lesions.We defined "proband" as the first individual in a family to enrol in the qpcs, whether that individual was affected or unaffected.Institutional Research Board approval was obtained for the study.

2.1.2
Referrals and Enrolment Patients with a diagnosis of pdac or pat were informed of the study by the treating hepatopancreatobiliary surgeon or oncologist at the time of their initial consultation at the McGill University Health Centre (muhc).Patients interested in participating met with the qpcs study coordinator after their initial clinic visit or were contacted within 2 weeks for an inperson or telephone interview.Referrals from cancer genetics or oncology clinics outside of the muhc, patient self-referrals, and referrals of unaffected individuals with a family history or known genetic syndrome associated with the earlier-described conditions were also accepted.Self-referrals occurred primarily through the qpcs Web page (http://www.cancerpancreas.ca).If, after 2 attempts (made twice monthly), a study candidate could not be reached in person or by telephone, and no response ensued during the following 6 months, the participant was classified as a "nonresponder." Probands (or their next-of-kin) are contacted annually after enrolment to obtain updated demographics, vital status, medical history, and genetic testing, as well as any changes in family history.Written informed consent from all participants was obtained at the time of enrolment.

Work-Up of Families
With consent from the proband, relatives (both affected and unaffected at-risk blood relatives) were invited to enrol in the qpcs when the family history was consistent with an increased risk of pdac or related conditions.If an affected family member was deceased, consent for the individual was obtained from the appropriate next-of-kin.
Genetic counselling and familial risk assessment were provided to each proband and family after enrolment.Families with a history suggestive of a genetic syndrome were referred to clinical genetics.

Current OnCOlOgy-VOlume 22, number 2, April 2015
Copyright © 2015 Multimed Inc.Following publication in Current Oncology, the full text of each article is available immediately and archived in PubMed Central (PMC).

2.2.1
Family History At the time of enrolment, study participants were asked to provide a detailed family history either by interview with a genetic counsellor or by completing a family history questionnaire.In either case, a threegeneration pedigree capturing both the maternal and paternal branches of the family, with details about ethnicity and family history, was obtained.

2.2.2
Epidemiology Study participants were asked to complete a personal history questionnaire (phq), which included detailed questions about lifestyle and epidemiologic risk factors.Data collection included medical history, medication use, dietary patterns, physical activity, reproductive history, chemical exposures, and alcohol and tobacco consumption.If a participant was deceased at the time of enrolment, the next-of-kin was asked to complete the phq on the participant's behalf.

Clinical Data
Participants provided the qpcs with written consent to obtain medical records for confirmation of their diagnosis (by radiologic imaging or pathology report, or both) and for collection of data relating to clinical treatments and outcomes.For participants who presented with a prior diagnosis of a genetic syndrome, clinical genetics records were obtained to confirm mutation status.For deceased family members who had a diagnosis of cancer, consent to obtain medical records was obtained from the appropriate next-of-kin.We also obtained permission to access records collected by the Province of Quebec (Régie de l'assurance maladie du Québec), which allows the qpcs to perform a death clearance analysis.

2.2.4
Biospecimens At enrolment, probands and relatives were asked to provide blood (blood tubes: BD Biosciences, Mississauga, ON) or saliva samples (saliva collection kit: DNA Genotek, Ottawa, ON), or both, for biobanking.Blood samples were obtained either at a regular clinic appointment, by a research nurse at the time of enrolment, or preoperatively from participants who were undergoing surgical resection at the muhc.Otherwise, (for example, if a participant resided out of town), a blood kit was mailed to the participant so that blood could be drawn at a clinic closer to the participant's home.Blood samples were then shipped and received within 24-48 hours.Alternatively, participants had the option to provide a saliva specimen, which was returned to the study coordinator at the time of enrolment or by mail.If a deceased participant had received clinical genetic testing, archived dna samples from the testing laboratory were requested for biobanking.
Blood samples were processed within 3 hours (local blood draws) or within 24-48 hours (shipped samples) from the time of collection.Plasma samples (collected only in the case of local draws) were immediately placed on ice and processed within 3 hours of collection.White blood cells were separated using an ammonium chloride red blood cell lysis buffer, and lymphocytes were isolated using a gradient lymphocyte separation medium (Ficoll-Paque: GE Healthcare, Baie d'Urfe, QC; lymphocyte separation medium: Wisent Bioproducts, St-Bruno, QC).The plasma, white blood cell pellets, and lymphocyte pellets were stored in liquid nitrogen.Extracted dna was resuspended in Tris-edta buffer and stored at 4°C.
For participants with a diagnosis of pdac or a related pat, we aimed to obtain samples of the corresponding tissues for biobanking.With patient consent, we requested archived (formalin-fixed, paraffin-embedded) tissue samples from biopsies or resection specimens (or both) from the treating hospitals.We also obtained any available archived non-tumour tissue samples that could be used as surrogate germline tissues for deceased participants enrolled by a next-of-kin.
Consent was obtained from patients undergoing resection at the muhc for collection of fresh-frozen tissue samples for tissue biobanking.Samples were collected immediately after resection.A pathologist first examined each specimen macroscopically, confirmed the diagnosis, and determined if the tumour tissue was sufficient for biobanking without compromising the clinical pathology diagnosis.When possible, paired affected (that is, tumour) and adjacent unaffected (that is, "normal") tissues were obtained.Depending on availability and diagnosis, tumour samples collected from resection specimens were also used to establish patient-derived xenografts in immune-compromised mice.

2.2.5
Data Storage Family history and pedigree data were stored and manipulated using a genetic data management system (Progeny Clinical, version 9: Progeny Software LLC, Delray Beach, FL, U.S.A.; http://www.progenygenetics.com/).All study participants were assigned a unique qpcs identification number to ensure anonymity and confidentiality.Biospecimen data were stored separately using the Canadian BioSample Repository software package (http:// www.biosample.ca).Each biospecimen was assigned a database biobank repository number unique from the qpcs identification number.The qpcs repository is registered with the Canadian Tumour Repository Network.rolment is ongoing, 28 referrals (7.5%) were pending enrolment at July 31, 2014.Of the completed patient referrals (that is, excluding the pending enrolments), 306 probands (88.4%) had been enrolled, 20 patients (5.8%) were nonresponders, 10 patients (2.9%) had declined to participate, and 10 patients (2.9%) were palliative or deceased at the time of attempted contact, translating into participation rates of 88.4% and 89.2% for all referrals and for pdac referrals respectively.

Between
Among the 306 probands enrolled, 277 provided family history data (90.5%), and 172 (56.2%) completed the phq.In addition to probands enrolled, 56 relatives of probands from 25 families were also enrolled.Considering all subjects enrolled [n = 362 (that is, 306 probands and 56 relatives of probands)], the qpcs now has 198 participants with a pdac diagnosis, 182 (91.9%) of whom have provided family history data, and 108 of whom (54.5%) have also made epidemiologic and clinical data available.Notably, most pdac-affected probands (69.5%) were enrolled within 3 months of their diagnosis.
Of the 190 probands with a pdac diagnosis, 52 (27.4%) had tumours that were resectable, 58 (30.5%) had locally advanced disease, and 80 (42.1%) had metastatic disease at the time of enrolment.The patients with benign surgical pathologies were enrolled based on premalignant or malignant preoperative clinical diagnoses.The preoperative diagnosis for the case with the low-grade sarcoma was pdac.Multiple synchronous primary tumours were diagnosed in 2 probands.Additionally, 23 probands (7.5%) were unaffected, but were enrolled because of an increased risk of pdac: that is, because of a significant family history, a mutation carrier with a known genetic syndrome (for example, Peutz-Jeghers), or in one case, a  chronically elevated serum level (>500 U/mL) of the CA19-9 tumour biomarker that was found incidentally without radiologic evidence of a lesion.
The study has ascertained 668 tissue samples from 333 of the 362 total enrolled subjects.Whole blood (plasma and buffy layer) was collected from 237 subjects (71.2%) and buffy layer alone from 20 subjects (6.0%).A blood sample could not be collected from 69 subjects (20.7%); however, unaffected tissue samples in the form of saliva, dna, or non-tumour tissue specimens (formalin-fixed, paraffin-embedded or fresh-frozen) were obtained.Affected tissue biospecimens (formalin-fixed, paraffin-embedded or fresh-frozen, or both) were collected from 189 subjects (56.8%).Notably, both affected and unaffected tissue specimens were collected from 182 subjects (54.7%).
Table iii describes the characteristics of the 198 enrolled subjects with a pdac diagnosis (that is, pdac-affected probands and pdac-affected relatives).Demographic and epidemiologic data were curated using available family history and phq data from the pdac subjects or their next-of-kin.Mean age at diagnosis was 66.1 ± 10.5 years, and the ratio of men to women was 1.4:1.Most subjects with pdac were white (79.2% maternal, 80.1% paternal), with an enrichment of cases (37.4%) having French-Canadian ancestry (at least 1 parental origin, Table iii).In addition, 56.8% of pdac-affected subjects had an ancestry (that is, French-Canadian, Ashkenazi Jewish, Greek, German, Polish, or Latvian) known to harbour recurrent germline ("founder") mutations in the BRCA1, BRCA2, and PALB2 pdac predisposing genes (Table iii) [11][12][13][14][15][16][17][18][19] .Table iii also shows data describing education, environmental exposures, weight loss, and history of type 2 diabetes and pancreatitis for enrolled patients with pdac.
Table iv summarizes the distribution of enrolled families with a high probability of hereditary pdac.In families with an available family history and at least 1 case of pdac or otherwise at risk of pdac (n = 211), tumours in 152 patients (72.0%) were classified as sporadic because the family had no history of pdac and did not meet criteria for a genetic syndrome associated with hereditary pdac.Another 59 patients (28.0%) had an increased likelihood of hereditary pdac either because of multiple pdac diagnoses in the family [that is, familial pdac (fpdac), 18.5%], because of diagnosis of a genetic syndrome whose tumour spectrum includes pdac (3.8%), or because of young age of onset (diagnosed with pdac at 50 years of age or younger, 5.7%).Those observations are consistent with the distribution of pancpro scores for all pdac probands with an available pedigree (n = 175, Figure 1).pancpro is risk prediction tool (courtesy of BayesMendel Lab, Harvard University, Cambridge, MA, U.S.A.) designed to estimate the probability that an individual carries a deleterious mutation in a pdac susceptibility gene 20 .

DISCUSSION
Research into pdac has been hampered by a paucity of data and biospecimens.Perhaps the most significant challenge in conducting pdac studies is recruitment of cases, because of rapid progression and death after diagnosis.The qpcs was designed as a clinic-based study with the goal of rapid case ascertainment after a pdac or pat diagnosis.Using that approach, we achieved participation rates of 88.4% for all referrals and 89.2% for referrals with a pdac diagnosis.Moreover, just 3.8% of pdac cases were palliative or deceased at the time of attempted contact.Previous population-based pdac registries have reported much lower participation rates of 35%-56%, with a significant proportion of pdac patients being deceased at the time of contact (28%-44%) 7,21 .In contrast to such low participation rates, the Mayo Clinic's clinic-based Pancreatic Cancer Genetic Study has reported enrolment rates (approximately 80%) similar to those achieved by the qpcs 22 .A comparison of study results supports our hypothesis that a clinic-based recruitment approach with an emphasis on rapid ascertainment facilitates high participation rates in pdac studies.
Although the qpcs achieved high participation rates, the study has limitations associated with a clinicbased approach.The low proportion of palliative or deceased pdac cases at attempted contact could reflect a selection bias because our patients were referred primarily from surgery and oncology clinics, likely favouring patients with a higher performance status.The qpcs is also affected by the pattern of referrals to the muhc.In 2003, the Quebec Ministry of Health and Social Services created the Réseau universitaire intégré de santé, assigning a portion of the province's territory to each of the four provincial medical schools.
McGill's territory includes Nunavik, the Cree Territory, Nord du Québec, Abitibi-Témiscamingue, Outaouais, western Montéregie, and western Montreal.As such, the muhc's referrals come largely from those regions of Quebec.Despite those probable biases, the demographics of the qpcs case series are consistent with previous reports in North America 1,2,23 , suggesting that the qpcs is an accurate representation of the patient population with pdac.The inclusion, in addition to pdac, of related pats is a unique research resource.Further, the qpcs results demonstrate that research registries can be successfully integrated into high-volume tertiary care surgery and oncology clinics.Moreover, the enrolment of cases through surgery and oncology clinics provides unique opportunities for studies of disease heredity in the prospectively collected cases, unselected for genetic susceptibility.
A number of lifestyle and environmental risk factors are associated with pdac.Although our cohort is currently small for epidemiologic analyses, we observed trends in risk factors associated with pdac that are consistent with those previously reported 5,24 .We collected level of education as a surrogate for socioeconomic status, because low socioeconomic status has been associated with higher pdac mortality rates 25 .Of participants who returned the phq (n = 109), 44.0% reported obtaining a high school diploma or less, and 56.0%indicated having received at least some college or university education.Smoking is the most common risk factor associated with pdac.Consistent with earlier reports 5 , more than half the phq questionnaire responders (51.4%) reported a history of cigarette smoking.Notably, 78.5% were exposed either directly or indirectly to cigarette smoke in their lifetime (Table iii).Compared with abstention and occasional drinking, heavy alcohol consumption (≥6 drinks daily) has been associated with an increased risk of pdac 26 .Of our phq responders, 70.1% indicated consumption of more than 1 alcoholic beverage weekly (mean: 10.0 ± 17.9 alcoholic beverages consumed weekly).
Long-standing type 2 diabetes (>10 years) has also been shown to increase risk of pdac by a factor of 1.51, and new-onset diabetes (<3 years) can be an early symptom of the disease 27,28 .Furthermore, individuals with a history of chronic pancreatitis (>2 years before the pdac diagnosis) have a risk of pdac that is increased by a factor of 2.71, and new-onset pancreatitis could be a symptom of pdac-associated ductal obstruction 29 .In our study, 30 participants (28.9%) reported a diagnosis of type 2 diabetes before their pdac diagnosis, with a mean of 8.9 ± 7.9 years between that diagnosis and the pdac diagnosis (range: 0-25 years).A diagnosis of pancreatitis was reported by 6 participants (5.8%) at a mean of 8.0 ± 10.0 years before their pdac diagnosis (range: 0-26 years).Although our cohort is small, those observations are consistent with the results of a pooled case-control analysis showing that 6.2% of pdac patients have a history of pancreatitis 29 .Younger patients (<65 years of age) with a history of pancreatitis could be at increased risk for a genetic predisposition to pdac, including hereditary pancreatitis 30 .
Several studies have shown that an increased body mass index (bmi) is associated with pdac risk 31 .the present study, 74.8% of responders with pdac were overweight (bmi > 25) and 14.5% were obese (bmi > 35) 1 year before enrolment.Interestingly, the average bmi for participants diagnosed with type 2 diabetes was higher than the average bmi for participants without diabetes, both at the time of enrolment (26.3 ± 4.20 vs.  25.6 ± 4.46) and at 1 year before enrolment (30.9 ± 5.29 vs. 29.0 ± 5.61).Because of discrepancies in the time of enrolment relative to the time of pdac diagnosis, the latter observations might be underestimated, given that weight loss is a common symptom of pdac.To account for that possibility, we also determined the mean bmi at enrolment and 1 year before enrolment (24.0 ± 4.66 and 27.2 ± 4.31 respectively) only for responders with pdac who were enrolled within 3 months of diagnosis (n = 64).Because the assessment of lifestyle and environmental exposures was retrospective and self-reported, the possibility of recall bias cannot be excluded.
Research registries for pdac have made important contributions to investigations of the genetic causes of pdac, including the discovery of novel pdac susceptibility genes 6,32,33 .Estimates suggest that 10% of pdac cases are attributable to Mendelian inheritance 34 .Although a small fraction of such hereditary cases are explained by germline mutations in known pdac susceptibility genes (for example, BRCA1, BRCA2, PALB2, CDKN2A) 34 , one of the most important questions in field remains the identification of the genetic causes of fpdac in which known genes are not implicated.The qpcs was designed with the goal of collecting high-quality data and biospecimens for genetic studies of pdac.Importantly, we aimed to collect epidemiologic and clinical variables similar to those collected by the opcs 7 , which will allow for research collaboration by the two Canadian studies.In addition, the qpcs design has allowed us to contribute data and biospecimens to the multicentre Pancreatic Cancer Genetic Epidemiology Consortium 35 , as part of a larger collaborative effort to elucidate the genetics underlying fpdac.
Through our integrated genetic counselling program, the qpcs has obtained detailed pedigrees and family history data from 90.5% of enrolled probands, including 174 probands with pdac (91.6%).High-risk pdac families have been extensively characterized, including acquisition of biospecimens from affected and unaffected blood relatives.Enrolled families are followed prospectively by annual contact to monitor for new diagnoses in the family.As has been the experience of other pdac registries, the qpcs expects that its prospective approach will likely identify new (incident) cases of pdac in family members who were unaffected at the time of enrolment, whether their kindred had multiple or only 1 pdac-affected relative at the time of qpcs enrolment 6 .Moreover, the ascertainment of both unaffected (that is, surrogate germline) and affected tissues in 113 cases of pdac (57.1% of enrolled cases) is particularly valuable for genetic predisposition studies 36 .Finally, the concomitant collection of lifestyle and environmental exposure data for the participants will allow for rationalization of non-genetic contributors to pdac causation.
Notably, more than half the enrolled patients in the qpcs affected with pdac (56.8%) had an ancestry (at least 1 parental origin) known to harbour founder BRCA1 and BRCA2 mutations [11][12][13][14][15][16][17][18][19] .The founder populations represent a genetically enriched subgroup ideal for gene discovery studies 8,[37][38][39] .The enrichment of qpcs with participants of French-Canadian descent is particularly valuable because fpdac is prevalent in that population 40 .In addition, French-Canadian founder mutations in the BRCA1, BRCA2, and PALB2 genes have already been described [11][12][13] , providing the qpcs with a unique opportunity to study the prevalence of those founder mutations among French-Canadian participants with pdac.BRCA1-and BRCA2-associated pdac is of particular interest because those tumours can have beneficial treatment responses to dna cross-linking agents (that is, platinum salts) and parp inhibitors [41][42][43][44][45][46] The research directions of the qpcs include characterizing the hereditary contribution to pdac.In particular, as a collaborative effort with the opcs, we are using next-generation sequencing to search for novel pdac susceptibility genes in fpdac kindreds.Moreover, we will take advantage of the French-Canadian patient population collected by the qpcs to look for recurrent genetic mutations in known pdac susceptibility genes, because those results could have important implications for clinical genetic screening in that population.

CONCLUSIONS
We have demonstrated that a rapid ascertainment protocol in a clinic-based pdac and pat research study can achieve high participation rates of 88.4% for all referrals and 89.2% for pdac referrals.After the opcs 7 , the qpcs is the second pdac research registry to be reported in Canada, and it provides a valuable resource available to the scientific community at large for studies of pdac and related pats.The registry will also facilitate the identification in Quebec of eligible participants for clinical screening protocols as new imaging technologies and biomarkers emerge.

CONFLICT OF INTEREST DISCLOSURES
We have read and understood Current Oncology's policy on disclosing conflicts of interest, and we declare that we have none.

figure 1
figure 1 Ages at diagnosis (in years) and pancpro scores for all enrolled probands with a diagnosis of pancreatic ductal adenocarcinoma and an available pedigree (n = 175).Dashed lines indicate the mean age and the mean PancPRO score (ranges given in parentheses).pancpro model courtesy of BayesMendel Lab, Harvard University, Cambridge, MA, U.S.A.
you goes to the patients and families participating in the qpcs.GZ is a clinical research scholar of the Fonds de recherche du Québec-Santé.ALS is a recipient of the Cedars Cancer Institute Fellowship award and a McGill Integrated Cancer Research Training Program award.Establishment of the qpcs has been funded by the Research Institute of the muhc, the foundations of the Royal Victoria and Montreal General hospitals, and the Department of Oncology of the muhc (to GZ).This work is dedicated to the memory of Rosalind Goodman.

Current OnCOlOgy-VOlume 22, number 2, April 2015
April 1, 2012, and July 31, 2014, the qpcs received 374 patient referrals, of which 230 had pdac (Table i).Most referrals (88.5%) came from the hepatopancreatobiliary surgery and oncology clinics; the remainder came from genetics clinics or were self-referrals.Because qpcs en- Copyright © 2015 Multimed Inc.Following publication in Current Oncology, the full text of each article is available immediately and archived in PubMed Central (PMC).

table i
Quebec Pancreas Cancer Study referrals and enrolment between April 1, 2012, andJuly 31, 2014 a Percentages calculated on completed referrals.b Percentages calculated on enrolled patients.pdac = pancreatic ductal adenocarcinoma; hpb = hepatopancreatobiliary.

table iii
Demographics and epidemiology of 198 enrolled patients with pancreatic ductal adenocarcinoma (pdac) a For whom data are available.Pts = patients; bmi = body mass index.Current OnCOlOgy-VOlume 22, number 2, April 2015Copyright © 2015 Multimed Inc.Following publication in Current Oncology, the full text of each article is available immediately and archived in PubMed Central (PMC).

table iv
Predisposition to pancreatic ductal adenocarcinoma (pdac) in 211 enrolled families a with available family histories With 1 or more cases of pdac (n = 209) or otherwise at risk of pdac (n = 2).b Defined as 2 or more pdac-affected relatives in a family, without diagnosis of a known genetic syndrome. a

VOlume 22, number 2, April 2015
Copyright © 2015 Multimed Inc.Following publication in Current Oncology, the full text of each article is available immediately and archived in PubMed Central (PMC).