Abstract
Keywords
Pancreatic ductal adenocarcinoma (PDAC) is a leading cause of cancer-related mortality, and most cases are unresectable at diagnosis because of locally advanced or metastatic disease. 1 Despite advances in management, including non-operative treatment options, complete surgical resection is the main treatment option for curative intent. 2 Approximately half of individuals with PDAC have metastasis at presentation, while only 1 in 5 are candidates for surgical resection.1,2 Advances in imaging and treatment techniques have led to an increasing role of diagnostic imaging for preoperative assessment of PDAC. 2 The best imaging modality for preoperative staging has been debated, with some studies supporting the use of magnetic resonance imaging (MRI) over computed tomography (CT) because of increased detection of liver metastasis, which would preclude surgical resection. 1
Tumour size is also an important criterion for the resectability of PDAC, and also plays a role in predicting survival. 2 As such, determining the best imaging modality to assess PDAC size can play a significant role in preoperative planning. Prior studies have found diagnostic imaging tends to underestimate PDAC tumour size compared to pathological specimens.3,4 Meanwhile, interobserver variability in PDAC size measurement on CT and MRI has not been explored to a great extent. The study by Cocquempot et al sought to compare the diagnostic performance and interobserver variability of CT and MRI in predicting PDAC size compared to surgical specimens as the reference standard. 2
This was a retrospective observational cohort single-centre study which consecutively sampled all patients with surgically resected PDAC who underwent both CT and MRI preoperative imaging within 3 months of surgery. 2 The largest axial diameter was measured by 2 independent, blinded radiologists on each MRI sequence (8 in total), as well as on pancreatic parenchymal and portal venous phase CT images. 2 These measurements were compared to the maximum dimension obtained on fresh gross pathologic specimens before fixation. 2 Interobserver agreement for PDAC measurements on each MRI sequence and CT phase was reported. Furthermore, differences in PDAC size between pathological specimens and each MRI sequence/CT phase were assessed.
Overall, interobserver reproducibility ranged from good to excellent for all MRI sequences and CT phases, with the lowest interobserver agreement seen for the fat-saturated in-phase T1-weighted sequence, and the highest for the poral venous phase CT. 2 Pancreatic parenchymal phase CT images demonstrated the best accuracy for PDAC size compared to pathological specimens, with a median measurement different between imaging and pathology of 2 mm, which did not reach statistical significance (P = .051). 2 Meanwhile, portal venous phase CT and all MRI sequences demonstrated statistically significant underestimation of PDAC size (P < .05) compared to pathology specimens, with a median difference ranging from 1 to 6 mm. 2
Two prior studies compared PDAC measurement on CT and MRI to histopathological examination, and both reported that diagnostic imaging underestimates the size of PDAC.3,4 However, these studies were limited to select MRI and/or CT sequences/phases, or they did not specify the contrast phase during which the measurement was performed.3,4 In contrast, the study by Cocquempot et al had the added benefit of measurements for each individual MRI sequence and CT phase and found pancreatic parenchymal phase CT measurements were the most accurate compared to histopathological results. 2 The remaining CT phase and all MRI sequence measurement demonstrated similar results to the previously published literature.3,4
The study by Cocquempot et al demonstrated multiple strengths. First, all included patients underwent both CT and MRI within 3 months of surgical resection of the PDAC, with no significant statistical difference between the time from each imaging modality to surgery. 2 Furthermore, this study assessed measurement on 8 different MRI sequences and 2 CT phases independently, which has not been extensively explored in the previous literature. Finally, the study had 2 independent radiologists with different levels of experience as blinded readers in the study. 2
Multiple potential limitations were also noted in the study by Cocquempot et al. The study was limited by the overall sample size of 29 patients, although a total of 506 individual measurements on different MRI sequences and CT phases were actually performed. 2 Given the potential for 10 separate measurements to be made per patient, there was also a potential for recall bias; the readers were blinded, and the order of cases was randomized to limit the possibility. In addition, the potential delay between imaging and surgical resection of up to 3 months may have contributed to the underestimation of PDAC on imaging. However, in a prior study by Ma et al, patients underwent surgery within 14 days of imaging with similar results, suggesting that this time delay may have little impact on size underestimation. 3
This was a well-conducted study with a meaningful research question. The results point to a potential added value of CT for preoperative assessment of PDAC, in addition to MRI. These findings fuel the ongoing debate on the use of MRI vs CT in this context, with the current findings in the literature supporting the use of both modalities in PDAC. 2 Canada performs less MRI scans per 1 million population compared to OECD average due to limited number of available MRI scanners. 5 There is no difference in number of CT scans between Canada and the OECD average. 5 Therefore, findings of Cocquempot et al are relevant for Canadian audience. Further research is warranted with larger, prospective, rigorous trials assessing the roles of CT and MRI for preoperative imaging of PDAC.
