Our study revealed that the NRR did not decrease during the last 2 decades among patients with PDAC entering the operating room for an intention-to-cure pancreatectomy. The observed overall NRR of 24.5% was consistent with those of previously reported series among patients diagnosed with PDAC. 
There is no doubt that the technical performances of CT and DW-MRI have improved over the last 2 decades. As a surrogate of this enhancement, the sizes of the extra-pancreatic disease which contraindicate the resection significantly decreased in the second period. Moreover, vascular contact is certainly better assessed nowadays via imaging eventually associated to EUS , but these improvements were not correlated with a reduction in NRR, which was surprising. It is now well-known that a short (<30 days) delay between CT scan and surgery is crucial in order not to increase the NNR . In our series, the majority of patients underwent a 3-phase CT scan with an “optimal” (18 days) delay [15-17] prior to surgery whatever the period of inclusion. This insured a relevant comparison between the two periods and the absence of NRR reduction could not be attributed to a higher delay between imaging and surgery in period #1 that counterbalanced the enhancement of imaging.
Routine laparoscopy was not performed, and this could be a drawback of our study; we suppose that any included patients could have been spared exploration if laparoscopy was performed. However, laparoscopy is not routinely performed worldwide and a recent large series reported that about 10% of patients underwent this procedure . Similarly, its relevance remains under debate. Nevertheless, we wanted to highlight that 12% of patients in our series who underwent an exploratory laparoscopy were finally found to have a contraindication to resection intraoperatively (with a "short" delay from laparoscopy to exploration of 18 days). This suggested that negative exploratory laparoscopy did not systematically imply a resection, mainly because it is difficult to explore major vasculature involvement in this way. As we strongly believe that positive PALNs are a contraindication to resection in patients with PDAC , we started a prospective evaluation of routine laparoscopy at diagnosis with PALN resection this year. This strategy will help assess the laparoscopic relevance at diagnosis, and we will report our results after the first 100 cases have been assessed.
Finally, the CA 19-9 serum level was independently associated with a higher NRR, reminding us the importance of the biological dimension in patients with PDAC [20, 21].
Not surprisingly, liver DW-MRI appeared to be a crucial tool for staging as already reported [4,22]. However, it could not be considered in isolation. Indeed, liver DW-MRI is a “focal” exam that only screens a specific zone (the liver and pancreas), has blind spots (subcapsular small liver metastasis and interference of duct dilation in case of bile duct obstruction) and consequently could not replace CT scan. To reinforce this, we showed that a CT/DW-MRI combination significantly reduced the NRR compared with patients who only underwent a CT scan (RR, 1.8). However, we did not observe a decrease in NRR between the two periods, despite a significantly higher liver MRI rate during period #2 among patients who did not have locally advanced disease at diagnosis. This could be a contradictory observation; however, two major changes in the patients’ strategies dramatically impacted period #2 at our institution and could explain the lack of reduction in NRR. First, we integrated the intraoperative PALNs assessment in our decision-making strategies for patients with PDAC scheduled for pancreatectomy.  Indeed, among cases with positive frozen section results, we did not perform tumor resection . Thus, not surprisingly, we observed a significantly higher NRR rate due to PALNs invasion during period #2. However, this relevant difference was not sufficient to increase the overall NRR probably because very few patients presented with PALNs invasion (3%). Second, since 2010, pancreatic surgery was performed by a dedicated surgical team . This increased both the number of cases and the complexity of the procedures performed. Consequently, we will have decreased the risk of achieving non-resection if our “local” criteria of resection had remained constant between the two periods. However, we pushed over the limits of resectability with venous and arterial resection  and this increased the risk of non-resection resulting in the absence of significant impact on the NRR.
Finally, we did not observe a reduction in NRR when considering patients with resectable disease at diagnosis. However, as the difference was not significant (due to an insufficient sample size), we noted a trend of NRR reduction between the two periods in this sub-population (13% vs 18%; P=.07). By suppressing the potential bias due to local invasion, we could argue that CT/DW-MRI helped to better identify liver metastasis or carcinomatosis at staging among patients with resectable disease.
Our study was not designed to assess oncology outcomes. However, we observed encouraging changes in survival between the two periods, probably due to improvements in surgery and perioperative treatment.
Our study supported the notion that patients diagnosed with pancreatic PDAC must benefit from CT and liver DW-MRI. Inclusion of the CA 19-9 level and laparoscopy might also help to reduce the NRR and consequently spare the patient from useless exploratory surgery. In the future, imaging [23-25] development in association with EUS [26,27], and assistance during laparoscopy  could improve the relevance of tumor staging. However, the aim of staging is to detect existing distant metastasis or vascular invasion that precludes resection. This concept will probably be challenged in the near future by new biomarkers such as circulating tumor cell number  or genomic assessment of the tumor [30,31] that could predict poor outcomes. If such staging became obvious and relevant, the pancreatic surgeon will then face a complex ethical situation: is a patient without any contraindication to resection based on the “classic” imaging staging to be spared resection because these new tools predict rapid disease progression?