Short-term Results of a Magnetic Resonance Imaging–Based Swedish Screening Program for Individuals at Risk for Pancreatic Cancer

Importance  Pancreatic cancer is the fourth leading cause of cancer-related death in Western countries. In approximately 10% of all patients with pancreatic cancer, it is possible to define a positive family history for pancreatic cancer or for one of the other related genetic syndromes. A screening program for individuals at risk is recommended; however, surveillance modalities have not been defined yet.

Objective  To analyze the short-term results of a prospective clinical surveillance program for individuals at risk for pancreatic cancer using a noninvasive magnetic resonance imaging (MRI)–based screening protocol.

Design, Setting and Participants  A prospective observational study of all patients with a genetic risk for developing pancreatic cancer who were referred to Karolinska University Hospital between January 1, 2010, and January 31, 2013, using an MRI-based surveillance program. All patients were investigated for the most common genetic mutations associated with pancreatic cancer.

Exposure  A noninvasive MRI-based screening protocol.

Main Outcomes and Measures  The ability of MRI to identify potential precancerous or early cancers in individuals at risk for pancreatic cancer.

Results  Forty patients (24 women and 16 men) were enrolled. The mean age was 49.9 years. The mean length of follow-up was 12.9 months. The numbers of relatives affected by pancreatic cancer were 5 in 2 patients (5%), 4 in 5 patients (12.5%), 3 in 17 patients (42.5%), 2 in 14 patients (35%), and 1 in 2 patients (5%). In 4 patients (10%), a p16 mutation was found; in 3, a BRCA2 mutation (7.5%); and in 1, a BRCA1 mutation (2.5%). In 16 patients (40%), MRI revealed a pancreatic lesion: intraductal papillary mucinous neoplasia (14 patients, 35%) and pancreatic ductal adenocarcinoma (2 patients, 5%). One patient had a synchronous intraductal papillary mucinous neoplasia and pancreatic ductal adenocarcinoma. Five patients (12.5%) required surgery (3 for pancreatic ductal adenocarcinoma and 2 for intraductal papillary mucinous neoplasia), while the remaining 35 are under continued surveillance.

Conclusions and Relevance  During a median follow-up of approximately 1 year, pancreatic lesions were detected in 40% of the patients, of whom 5 underwent surgery. Although the study time was relatively short, the surveillance program in individuals at risk seems to be effective.

Pancreatic cancer is the fourth leading cause of cancer-related death in the United States and many Western countries.¹ Because the incidence and mortality rates are almost identical,² pancreatic cancer can be considered a global lethal disease. Although treatment has improved, the resection rate in patients with ductal adenocarcinoma remains around 30%, and the 5-year survival rate is less than 20%.³ Because of the low incidence of pancreatic cancer in the general population, population-based screening is not considered cost-effective. In the past 2 decades, data from the literature have demonstrated that pancreatic cancer can be the phenotypic expression of some of the known genetic syndromes,^4-9 and they have led to the identification of a familial risk factor for the development of pancreatic cancer.¹⁰ In the latter condition, called familial pancreatic cancer (FPC), the increased risk is associated with the number of affected family members.¹¹Quiz Ref ID According to prospective epidemiological studies, a positive family history of pancreatic cancer is present in approximately 10% of all consecutive probands.¹² The identification of a population at risk; the suggestion that early surgical treatment of pancreatic cancer can improve the prognosis¹³; and, in particular, the identification of precursor lesions, such as pancreatic intraepithelial neoplasia (PanIN) and intraductal papillary mucinous neoplasia (IPMN),¹⁴ which are associated with the natural history of FPC, contributed to the development of national and international guidelines for the surveillance of individuals at increased risk.^15-17 However, to the present day, consensus has not been reached regarding inclusion criteria for a clinical surveillance program, screening modalities, and target lesions. Traditionally, individuals with a 10-fold relative risk for developing pancreatic cancer were considered eligible for entering a screening program.^15,16 However, the International Cancer of the Pancreas Screening Consortium suggested inclusion of individuals with only a 5-fold relative risk in a surveillance program.¹⁷ Early pancreatic cancer, IPMN lesions, and PanIN lesions are considered the target lesions of clinical screening, although the actual detection of PanIN lesions during a surveillance program remains debatable and uncertain.¹⁷ The imaging modality techniques used in the existing programs are, alone or in combination, magnetic resonance imaging (MRI), computed tomography, endoscopic ultrasonography (EUS), and endoscopic retrograde cholangiopancreatography. Previous recommendations demonstrated a tendency to use low-aggressive modalities such as MRI and EUS.^16,17 The results of clinical studies that have been conducted so far are inconclusive and difficult to compare owing to the use of different screening modalities and inclusion criteria. The yield of FPC screening programs reported in the literature ranges from 1.3% to 50%.^14,18-27 To our knowledge, the cost-effectiveness of a surveillance program for FPC has not been demonstrated yet.

The aim of this study was to analyze the short-term results of a prospective clinical surveillance program for individuals at risk for pancreatic cancer using a noninvasive MRI-based screening protocol.

From January 1, 2010, to January 31, 2013, 40 patients with an increased genetic risk for developing pancreatic cancer were enrolled in an MRI-based surveillance program at Karolinska University Hospital. The present study was approved by the local ethical committee in Stockholm, EPN, and participants provided written informed consent.

Individuals were enrolled in the study through 3 different routes: (1) relatives of patients treated for pancreatic cancer at Karolinska University Hospital who had a positive family history of pancreatic cancer or a positive history of a genetic syndrome associated with pancreatic cancer; (2) individuals with a genetically increased risk who were referred from other Swedish centers; or (3) those at genetically increased risk who were referred by general practitioners.

All individuals with a 10-fold higher risk for developing pancreatic cancer than the general population were included in the study.^15,16 The inclusion criteria are outlined in the Box. In case a patient was suspected to have 1 of the known genetic syndromes, the individual was enrolled only if a corresponding gene mutation was detected. All patients who fulfilled the criteria listed in the Box and were aged 45 years or 10 years younger than the youngest affected family member were considered, as recommended by previous guidelines.¹⁶

Box Section Ref ID

All patients fulfilling the inclusion criteria were enrolled in the clinical study. The screening protocol, based on MRI, is described in Figure 1. For each patient, a full personal and familial medical history (including a pedigree) was obtained and a clinical examination was performed. All patients underwent MRI/magnetic resonance cholangiopancreatography with secretin.¹⁷ If the result of the MRI was negative (no abnormalities detected), rescreening after 1 year, using the same modalities, was recommended. Only patients with some abnormality on MRI underwent EUS with or without fine-needle aspiration and/or computed tomographic scan. In case of unspecific findings or of IPMN without indication for surgery,²⁸ a 6-month follow-up with MRI was recommended. Every patient underwent genetic testing for the most common gene mutations associated with FPC (BRCA1, BRCA2, and p16).

Solid nodules and suspected IPMN lesions were considered the target lesions of the screening program.¹⁷ Owing to the low specificity and sensitivity of the available techniques for the detection of PanIN lesions and/or associated lobulocentric atrophy,¹⁷ PanIN lesions were not considered a screening target.

Every patient with a positive finding on screening was discussed at the pancreatic multidisciplinary conference. Patients with suspected cancer were treated with a radical surgical procedure. Patients with a suspected premalignant lesion (IPMN) underwent a radical or parenchyma-sparing surgical resection according to guidelines.¹⁷ By principle, no patient underwent total pancreatectomy as a purely preventive measure.

Comparison of continuous variables was performed using a t test. Comparison of categorical variables was done by χ² analysis using GraphPad Prism Software.

Sixteen men (40%) and 24 women (60%) were included in the study. The mean age was 49.9 years (range, 23-76 years). In 38 of the study individuals, the increased risk was based on a history of FPC (95%), while in 2 more patients (5%) with a single first-degree relative with pancreatic ductal adenocarcinoma (PDAC), a BRCA2 syndrome had been genetically confirmed. Genetic testing revealed a p16 mutation in 4 patients (10%), 3 patients (7.5%) were found to have a BRCA2 mutation, and 1 patient (2.5%) had a BRCA1 mutation. In the FPC group, there were 14 individuals (35%) with 2 affected relatives, 17 (42.5%) with 3 affected relatives, 5 (12.5%) with 4 affected relatives, and 2 (5%) with 5 affected relatives. The mean follow-up was 12.9 months (range, 0-36 months).

Quiz Ref IDOverall, 16 individuals (40%) had a positive finding at some point during the screening program. The characteristics of these patients are detailed in Table 1. Individuals with a positive finding were older than those with negative screening results (58.6 years vs 44.2 years; P < .001). The 2 groups did not differ significantly regarding sex, smoking habit, number of affected relatives (first or second degree), and mean age of affected relatives (Table 2). In the group of patients with positive screening results, the median interval between the start of the screening and the detection of a pancreatic lesion was 6.2 months (range, 0-36 months). In 12 individuals (30%), the lesion was detected at point zero. The radiology of the lesions is described in Figure 2. Quiz Ref IDThe clinical diagnosis of these lesions was branch duct (BD) IPMN in 9 patients (22.5%), mixed-type IPMN in 3 (7.5%), main duct IPMN in 2 (5%), and PDAC in 2 (5%). One patient with mixed-type IPMN lesions also had simultaneous PDAC. The mean diameter of the cystic lesions detected was 10.6 mm (range, 5-23 mm). A total of 5 patients (12.5%) underwent surgery during the screening program, and their characteristics are outlined in Table 3. Three of the patients who underwent resection were affected by IPMN lesions. Intermediate-grade dysplasia was found in 2 patients, one of whom had undergone resection, while the other patient was treated with multiple enucleations (Figure 3). A third patient had been followed up for a 1-cm BD-IPMN in the pancreatic tail by MRI and EUS in 6-month intervals, as per the study protocol (Figure 1). After 2 years of surveillance, a 1.5-cm solid lesion was detected in the pancreatic head, the cystic lesion in the tail had progressed (increased cyst size up to 2 cm), and the main pancreatic duct had become dilated. Consequently, the patient underwent total pancreatectomy, and histology confirmed the presence of a noninvasive mixed-type IPMN lesion with high-grade dysplasia in the tail region and synchronous PDAC (T3N0M0) in the pancreatic head. Two additional patients were treated for a solid pancreatic lesion, which was histologically confirmed to be PDAC. The first patient was treated for an early cancer (T1N0M0). The second patient had missed 1 of the yearly MRI examinations and developed unspecific abdominal pain a few months before the next control was due. The MRI showed an advanced pancreatic cancer in the pancreatic head (T4N1M0) (Figure 4). Quiz Ref IDAll individuals who had a positive screening result but did not undergo surgical resection (n = 11) were screened with MRI at 6-month intervals. Of these, 7 (63.6%) had a stable finding over a median follow-up time of 9.4 months (range, 0-36 months). Four more individuals (36.4%) showed limited, nonsignificant progression of the lesion during a median follow-up of 19 months (range, 12-26 months).

While several national and international societies and consortia recommend a surveillance program for individuals at risk for developing pancreatic cancer,^15-17 to our knowledge, the actual efficacy of such preventive programs has not been demonstrated so far.

In principle, an efficient surveillance program should allow the detection of early cancer or premalignant lesions. Furthermore, it should be cost-effective and based on noninvasive investigations.^16,17 The screening program proposed in this study is based on MRI only as a first investigational modality because it is noninvasive and low cost.

Interpretation of the results of this study requires careful consideration. The high detection rate of lesions in this study series (40%) is in line with results from previous studies and confirms that high-risk individuals have a higher tendency to develop premalignant precursor lesions.^14,29 The observation that patients with a positive finding on screening were older than those with negative screening results confirms that in high-risk individuals, the risk for developing pancreatic lesions is age related. It therewith validates the International Cancer of the Pancreas Screening Consortium guidelines¹⁷ regarding the suggested age at which to commence screening. Most lesions were suspected IPMNs, as has been reported previously by others.^14,30 However, the real benefit to the individual of early detection of these lesions remains controversial, in particular because the natural history of IPMNs is not well known and may be shorter than in the general population.¹⁴

In the 5 patients who underwent surgery in this study, the preoperative diagnosis was confirmed histologically. In 2 of the patients, surgery had been undertaken for noninvasive IPMNs of mixed type and multifocal BD type, with intermediate-grade dysplasia. In 3 additional patients, PDAC had been the indication for surgical resection. In 1 of these patients, screening had detected a small BD-IPMN lesion in the tail of the pancreas, which after 2 years of 6-month-interval follow-ups showed progression to mixed-type IPMNs and development of a synchronous cancer in the head of the pancreas. In a second patient, screening had allowed resection of an early cancer (T1N0M0). However, in a third patient who had missed the second planned yearly surveillance investigation, cancer was detected at an already advanced stage. Quiz Ref IDTherefore, in only 3 of the 5 patients who underwent resection, or in 7.5% of the entire patient series, surgery was a prophylactic measure or early cancer treatment. Furthermore, because in 2 of these patients the resected lesion was a noninvasive IPMN lesion with only intermediate-grade dysplasia, the real benefit to patients is not really known. One patient who underwent surgery had developed an interval cancer, which screening had failed to detect at an early stage and which must have progressed rapidly during the 6-month intervals between 2 successive MRI investigations. Finally, the last patient had not been compliant with the study protocol and was diagnosed as having operable but locally advanced disease. In how far the inclusion in this screening protocol of EUS as a baseline investigation would allow earlier cancer detection is controversial. Several studies in the literature have shown that MRI can be useful for the detection not only of small cystic lesions, but even of early solid tumors of the pancreas.^25,31 Furthermore, data in the current study demonstrate that detection of early pancreatic cancer (T1N0M0) was possible in 1 of the patients when using MRI only. Conversely, even in a patient who was under surveillance for IPMNs with MRI and EUS at 6-month intervals, PDAC was detected only at stage T3N0M0. Further, it has to be acknowledged that EUS is significantly less frequently used in Europe than in the United States. Hence, MRI-based screening is likely to find more acceptance in European pancreatic cancer centers. Moreover, MRI has the potential benefit of being less dependent on investigator expertise than EUS. In this study, PanIN lesions were excluded as a screening target. Although PanINs are well established as precursor lesions, the macroscopic correlate of these lesions (ie, lobulocentric atrophy) is nonspecific. Moreover, even in the International Cancer of the Pancreas Screening Consortium guidelines,¹⁷ agreement regarding the detection and treatment of PanINs has not been reached.

An MRI-based protocol for the surveillance of individuals at risk for developing pancreatic cancer seems to detect cancer or premalignant lesions with good accuracy. The exclusive use of MRI can reduce costs, increase availability, and guarantee the safety of the individuals under surveillance compared with protocols that are based on more aggressive methods. However, because of the small number of patients and the divergent results, this study did not allow evaluation of the efficacy of MRI as a single screening modality. One principal obstacle to effective surveillance, encountered in the current study as in previous studies, is the lack of knowledge about the natural history of premalignant lesions of the pancreas, as well as the lack of criteria for reliable prediction of progression and outcome of these lesions in individual high-risk patients.

Corresponding Author: Marco Del Chiaro, MD, PhD, Pancreatic Surgery Unit, Division of Surgery, Department of Clinical Science, Intervention, and Technology (CLINTEC), Karolinska Institute, SE-171 77 Stockholm, Sweden (marco.del-chiaro@karolinska.se).

Accepted for Publication: October 16, 2014.

Published Online: April 8, 2015. doi:10.1001/jamasurg.2014.3852.

Author Contributions: Dr Löhr had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Del Chiaro, Gustafsson, Hansson, Segersvärd, Andren-Sandberg, Löhr.

Acquisition, analysis, or interpretation of data: Verbeke, Kartalis, Pozzi Mucelli, Hansson, Haas, Löhr.

Drafting of the manuscript: Del Chiaro, Verbeke, Andren-Sandberg, Löhr.

Critical revision of the manuscript for important intellectual content: Verbeke, Kartalis, Pozzi Mucelli, Gustafsson, Hansson, Haas, Segersvärd, Andren-Sandberg, Löhr.

Statistical analysis: Del Chiaro.

Obtained funding: Löhr.

Administrative, technical, or material support: Verbeke, Gustafsson, Haas, Segersvärd, Löhr.

Study supervision: Del Chiaro, Andren-Sandberg, Löhr.

Conflict of Interest Disclosures: None reported.