AIM: To analyze the benefits and harms of pancreatic cancer screening in familial high-risk individuals (HRIs).

METHODS: Studies were identified by searching PubMed, EBSCO, ClinicalTrials.gov and the Cochrane database from database inception to June 2014. We also obtained papers from the reference lists of pertinent studies and systematic reviews. English-language trials and observational studies were searched. The key words used as search terms were “screening” and “surveillance”. Cost-effectiveness, diagnostic rate, survival rate, mortality and adverse events were the outcomes of interest. Age, sex, lifestyle and other confounding factors were also considered. However, anticipating only a few of these studies, we also included observational studies with or without control groups. We also included studies concerning the anxiety associated with pancreatic cancer risk and other psychological changes in familial HRIs. We extracted details on study design, objectives, population characteristics, inclusion criteria, year of enrollment, method of screening, adjusted and unadjusted mortality, cost-effectiveness and adverse events from the included studies. Studies were assessed using the Reporting of Observational studies in Epidemiology (STROBE) checklist.

RESULTS: Sixteen studies on pancreatic cancer screening were included. Five studies included control groups, nine were observational studies without control groups, and the other two studies investigated the worry associated with pancreatic cancer risk. We found that pancreatic cancer screening resulted in a high curative resection rate (60% vs 25%, P = 0.011), longer median survival time (14.5 mo vs 4 mo, P < 0.001), and higher 3-year survival rate (20% vs 15.0%, P = 0.624). We also found that familial HRIs had a higher diagnostic rate of pancreatic tumors than controls (34% vs 7.2%, P < 0.001). In patients who underwent regular physical examinations, more stage I pancreatic cancers were observed (19% vs 2.6%, P = 0.001). In addition, endoscopic ultrasonography, which was the main means of detection, diagnosed 64.3% of pancreatic cancers. In comparison, endoscopic retrograde cannulation of the pancreas, magnetic resonance imaging, and computed tomography diagnosed 28.6%, 42.9%, and 21.4%, respectively. For mass lesions, instant surgery was recommended because of the beneficial effects of post-operative chemotherapy. However, in patients with intraductal papillary mucinous neoplasms, we did not find a significant difference in outcome between surgery and follow-up without treatment. Moreover, pancreatic cancer screening in familial HRIs had a greater perceived risk of pancreatic cancer (P < 0.0001), higher levels of anxiety regarding pancreatic cancer (P < 0.0001), and increased economic burden.

CONCLUSION: Pancreatic cancer screening in familial HRIs is associated with a higher detection rate and longer survival, although screening may influence psychological function and increase the economic burden.

Pancreatic cancer has become one of the most fatal diseases over the past years. The 5-year survival rate, estimated by the Surveillance Epidemiology and End Results program, is 4% and this figure is the lowest for all types of cancer[1]. There were 216400 new cases of pancreatic cancer and 213500 deaths worldwide from pancreatic cancer in 2000[2]. The reasons for the high mortality rate include a low early diagnostic rate, low eradication rate, and poor radiotherapy and chemotherapy response rates. Almost half of patients with early stage pancreatic cancer are asymptomatic. Once detected, the 4-year survival rate can increase to 78% following resection of pancreatic cancers < 2 cm[3]. The detection of pancreatic cancer in the early stage affects prognosis in the pancreatic cancer high-risk population, especially familial high-risk individuals (HRIs).

There is a familial aggregation of pancreatic cancer. Familial HRIs are defined as families with at least two first-degree relatives, suggesting an autosomal dominant penetrance[4]. Some researchers have suggested that individuals with three or more affected family members, one of whom must be a first-degree relative, are considered high-risk individuals and should undergo screening[5]. Persons who do not fit into this category can be classified as having sporadic pancreatic cancer.

The risk and incidence of pancreatic cancer is exceptionally high in familial pancreatic cancer kindred, where at least three first-degree relatives have already been diagnosed with pancreatic cancer[6]. There was a 57-fold increased risk of pancreatic cancer in these individuals compared with surveillance findings[6]. A meta-analysis that included nine studies indicated a significant increase in pancreatic cancer risk associated with having an affected relative, with an overall summary relative risk of 1.80[7]. Therefore, targeted screening of familial HRIs may detect potential pancreatic neoplasms or precursor lesions, including pancreatic intraepithelial neoplasia (PanIN), intraductal papillary mucinous neoplasms (IPMNs) and mucinous cystic neoplasms (MCNs)[8]. More and more professional societies recommend pancreatic cancer screening using imaging studies or other methods, primarily in patients at high risk for pancreatic cancer. However, the recommendations for pancreatic cancer screening include some questionable points. The cost-effectiveness, concern regarding pancreatic cancer risk, screening test accuracy, and attitude to cancer screening should be taken into consideration.

In this study, we conducted a systematic review to analyze the benefits and harms of pancreatic cancer screening in familial HRIs.

We included 16 studies in the final analysis. Five studies were case-control studies that included a comparison group[12-16], nine were observational studies without a comparison group[17-25], and the other two studies investigated anxiety associated with pancreatic cancer risk.

We reviewed and evaluated controlled and non-controlled studies systematically, and examined the benefits and harms of pancreatic cancer screening in familial HRIs. EUS combined with FNA testing was the most common screening method evaluated. Although only two studies[20,22] reported mortality, and many precursor lesions did well with curative or preventive therapy, there was also very-low-strength evidence to draw conclusions on the benefits and harms of pancreatic cancer screening in HRIs.

In this systematic review, we obtained useful information. In general, screening familial HRIs can identify early pancreatic tumors resulting in early intervention. One retrospective study[13] indicated subjecting individuals to regular examinations could result in the detection of early-stage pancreatic cancer leading to curative resection, and a greater median survival time. Nearly all the included studies reported precursor lesions and in some studies early treatment was carried out, which decreased the total mortality. The merged data showed a significantly higher diagnostic rate and more stage I pancreatic cancer in the experimental group. Although heterogeneity was present, these findings provide a reference point. Methodological flaws such as selection bias, lead-time bias, length-time bias and the lack of RCTs on pancreatic cancer screening were also observed. Although EUS is used widely, other methods such as MRI can also be useful in pancreatic cancer screening. Therefore, we were unsure whether survival rate would be affected by the detection of pancreatic cancer using this method. In addition, interval cancers should be taken into consideration. Interval cancers can be interpreted as those that present symptomatically between screening episodes and can skew the results of screening studies. Considering the latent period of early pancreatic tumors, when the tumor is detected will affect the survival rate.

Most studies were single-center and participants were from the same center, which avoided many biases. Only four studies[20,21,24,25] determined the smoking status of patients. Smoking is an established risk factor for pancreatic cancer[33]. Therefore, it must be taken into consideration. In addition, hypercholesterolemia and alcohol consumption represent significant and independent risk factors for pancreatic cancer[34]. Pannala et al[35] indicated that new-onset diabetes was a potential clue to the early diagnosis of pancreatic cancer. These factors may have influenced the final results. We did not know how much they affected the outcomes of these studies.

We chose familial HRIs as research subjects. In this population, there are also individuals with high risks for other diseases. Peutz-Jeghers syndrome with STK11/LKB1 gene mutation[36], hereditary pancreatitis with PRSS1 gene[37,38] or SPINK1 gene mutation[39], familial atypical multiple mole melanoma syndrome with CDKN2A or p16 gene mutations[40], hereditary breast ovarian cancer syndrome with BRCA2 and BRCA1 gene mutations[41,42], and Lynch syndrome with mismatch repair genes[43] are found in high risk populations, who should also receive attention. Nevertheless, these are not completely independent. Familial HRIs may also have gene mutations. The PRSS1 gene mutation is the main influencing factor in hereditary pancreatitis, an autosomal dominant disease. BRCA2 is one of the most common mutations, and was as high as 17% in one study[44]. Future studies on HRIs with gene mutations should be carried out to determine if they have a higher risk than HRIs without gene mutations. In addition, because of the complex nature of the pedigrees in pancreatic cancer, Wang et al[45] designed a tool known as PancPRO to identify familial HRIs, which was used for selecting and screening.

Patients with precursor lesions or early-stage pancreatic cancer are mainly asymptomatic. Therefore, the detection method is very important. Biomarkers and imaging tests are both optional. In view of the existence of gene mutations, genetic testing is also a feasible method. Through this method, early intervention can avoid the occurrence of pancreatic cancer. However, because of the cost and difficulty of implementation, some methods may only apply to certain populations. CA199 is the most widely used biomarker. However, it has a very low positive predictive value[46]; therefore, it is mainly used in the follow-up and surveillance of pancreatic cancer. Abdominal ultrasound is also rarely used for pancreatic cancer screening because of its low sensitivity. One study using abdominal ultrasound as the initial detection method reported a diagnostic rate of 7.5%-15% in stage I pancreatic cancer[13]; however, this method was unable to confirm the findings in patients with negative results using other methods. As the main screening tool, the accuracy of EUS can reach 87.1%-92%[47,48], and it also showed a high diagnostic rate of precursor lesions (PanINs and IPMN) (Table 2) in our included studies. Thus, EUS is an ideal detection method. However, EUS is an invasive procedure that is easily affected by the operator, has a comparatively high cost and cannot be used for screening the general population. However, EUS is necessary in HRIs. MRI as a screening tool also has excellent results[20,25], and is a non-invasive procedure without exposure to radiation. However, the lack of relevant papers ,meant that we could not conclude that EUS is superior to MRI. CT and ERCP are rarely used because of radiation exposure and the risk of acute pancreatitis.

The natural history of early-stage pancreatic cancer or precursor lesions has not been well-characterized. Therefore, long-term surveillance or preventative surgery is controversial. Preventative surgery may be risky, as mentioned above. The current viewpoint is that IPMNs have the potential for malignant transformation[49] and may result in early-stage pancreatic cancer. Unfortunately, we do not know the frequency and rate of progression of early-stage pancreatic cancer or precursor lesions. Tanno et al[50] performed a prospective study that included 82 patients with branch duct IPMN. Five patients (6.1%) developed adenoma, one (1.2%) had carcinoma in situ, one (1.2%) was borderline, six (7.3%) had cystic dilatation, and 69 (84.1%) had no change at a median follow-up of 59 mo. Tanno et al[50] confirmed the possibility of canceration of IPMNs during 59 mo. Therefore, IPMNs or other precursor lesions detected in the included studies had the potential to develop into adenomas. Although the IPMNs were stable during the follow-up period, it was possible that they could change to adenomas beyond the follow-up period. The results of the remaining patients with precursor lesions, which were recovery or death from cancer, would influence the final mortality and cost-effectiveness measures.

Some limitations were present in our systematic review. Firstly, we excluded non-English-language studies, but mitigated the risk by searching more databases. One study indicated that language restrictions did not appear to bias estimates of the effectiveness of a conventional intervention[51], which also supported our evidence. Secondly, the lack of RCTs, and the existence of selection bias, lead-time bias and length-time bias were limitations in our included studies. These reduced the credibility of the review. Thirdly, although we identified some information regarding the harms of screening, we did not determine the cost-effectiveness of screening. Some studies referred to the cost or cost-effectiveness[16,31]; however, a large RCT would be the most definitive way to estimate the cost-effectiveness of pancreatic cancer screening.

In conclusion, the current results showed that pancreatic cancer screening in familial HRIs is associated with a higher detection rate and longer survival. For precursor lesions, follow-up or surgery is controversial. However, screening may influence psychological function and increase economic burden. Thus, in developed countries, pancreatic cancer screening should be promoted. In addition, further RCTs on pancreatic cancer screening in familial HRIs should be performed to provide more conclusive evidence.