Abstract
Purpose:
To evaluate the associations between computed tomography (CT) imaging features extracted from the structured American Pancreatic Association (APA)/Society of Abdominal Radiology (SAR) template and overall survival in patients with resected pancreatic ductal adenocarcinoma (PDAC).
Methods:
This retrospective analysis included consecutive patients with PDAC who consented to genomic tumor testing and underwent preoperative imaging and curative intent surgical resection from December 2006 to July 2017. Two radiologists assessed preoperative CT imaging using the APA/SAR PDAC reporting template. Univariable associations between overall survival and imaging variables were evaluated using Cox proportional hazards regression.
Results:
The study included 168 patients (66 years ± 11; 91 women). 126/168 patients (75%) received upfront surgical resection whereas 42/168 (25%) received neoadjuvant therapy prior to surgical resection. In the entire cohort, features associated with decreased overall survival were tumor arterial contact of any kind (hazard ratio (HR) 1.89, 95% CI 1.13–3.14, p=0.020), tumor contact with the common hepatic artery (HR 2.33, 95% CI 1.35–4.04, p=0.009) and portal vein deformity (HR 3.22, 95% CI 1.63–6.37, p=0.003). In the upfront surgical group, larger tumor size was associated with decreased overall survival (HR 2.30, 95% CI 1.19–4.42, p=0.013). In the neoadjuvant therapy group, the presence of venous collaterals was the only feature associated with decreased overall survival (HR 2.28, 95% CI 1.04–4.99, p=0.042).
Conclusion:
The application of the APA/SAR pancreatic adenocarcinoma reporting template may identify predictors of survival that can aid in preoperative stratification of patients.
Keywords: Carcinoma, Pancreatic Ductal, Pancreatic Carcinoma, Neoadjuvant Therapy, Neoplasm Staging, Tomography, X-Ray Computed, Survival Analysis
Introduction
Pancreatic ductal adenocarcinoma (PDAC) will likely become the second leading cause of cancer-related death in the United States by 2030 [1]. Even with advances in treatment, patients with PDAC have a poor five-year survival rate of 9% [2]. Predicting survival in patients with PDAC can be challenging, and there is a lack of predictive and prognostic biomarkers. Thus, prediction of survival in the preoperative setting for PDAC remains a paramount topic of investigation.
Previously, various studies have assessed whether PDAC features on preoperative computed tomography (CT) are associated with patient survival, particularly with regards to tumor-vascular involvement [3–15]. Recently, in patients with resectable PDAC undergoing curative intent surgery, Pak et al. showed that when using a gestalt of preoperative CT and CA 19–9 levels, both surgeons and radiologists outperformed an accepted clinical nomogram in predicting two-year survival in these patients [16]. Previous studies have also assessed the utility of preoperative CT in predicting R0 (resection margin negative for microscopic tumor) and other histopathological tumor features [3–6,17–20]. Additionally, many studies have assessed the utility of preoperative CT to assess the local extent of tumor and treatment response post neoadjuvant therapy [21–27].
The National Comprehensive Cancer Network (NCCN) recommends staging PDAC with CT [28]. In addition, the NCCN also recommends the use of the American Pancreatic Association (APA) and Society of Abdominal Radiology (SAR)-endorsed PDAC structured reporting template [29] as it has been shown to provide superior evaluation of the images for surgeons and facilitate surgical planning [30]. However, comprehensive survival prediction in patients with PDAC undergoing curative intent surgery specifically using the imaging variables in the APA/SAR-endorsed template has not been attempted.
Therefore, the purpose of our study was to evaluate the associations between CT imaging features extracted from structured APA/SAR template variables and overall survival (OS) in patients with resected PDAC.
Materials and Methods
The Institutional Review Board approved and waived the need for written informed consent for this retrospective analysis. The study was compliant with the Health Insurance Portability and Accountability Act.
We retrospectively identified 285 consecutive patients with pathologically proven PDAC who had consented to genomic tumor testing and who underwent curative intent surgical resection from December 2006 to July 2017. For this study, curative intent surgical resection was either upfront surgical resection or surgical resection after neoadjuvant therapy. We excluded patients with imaging performed at an outside institution, no imaging performed, CT without multiplanar reconstructions, or MRI performed in lieu of CT (n = 87); patients with a CT study performed at our institution more than 60 days prior to surgical resection (n = 4), prior pancreatic surgery (n = 5), patients with PDAC arising in an intraductal papillary mucinous neoplasm (n = 11), and patients whose genomic tumor testing was performed using a biopsy remote to the surgery or on a specimen collected prior to neoadjuvant therapy (n = 5). Thus, of 285 eligible patients, 168 patients were included in the final study population, of which 126 had upfront surgical resection and 42 had surgical resection after neoadjuvant therapy. Figure 1 shows the patient inclusion flowchart.
Fig. 1.
Patient flowchart
Patient clinical and follow-up data was collected via retrospective medical record review. Date of last follow-up was defined as the date that patient was last known to be alive, either through a clinic visit, documentation of a phone call, or from an outside physician. Death was confirmed only if it had been recorded in the medical record. Pathological diagnosis of PDAC was confirmed on all surgical resection specimens. Genomic tumor testing results, although available, were not incorporated into this study.
Imaging Technique
All CT studies were performed at our institution after the administration of iodinated contrast (Omnipaque 300, GE Healthcare, New Jersey) on multidetector CT (Lightspeed 16 and VCT, GE Healthcare, Wisconsin). Varying intravenous contrast volumes/rates and scan parameters were used according to the various institutional protocols (Supplemental Table 1).
Imaging Evaluation
Preoperative CT studies were reviewed independently by two attending radiologists with 19 (S.G.) and 2 years (K.H.) of experience respectively after abdominal oncologic imaging fellowship at high volume pancreatic surgery centers. Both radiologists also routinely participated in the hepatopancreatobiliary tumor board. All CT studies were reviewed on Picture Archiving and Communication System (PACS) (GE, Centricity) workstations. Both radiologists recorded CT features using a standardized scoresheet based on the APA/SAR-endorsed PDAC reporting template [29]. Radiologists were free to view any contrast phase if the CT was performed as a multiphase study. Radiologists were aware of PDAC diagnosis and that the patients underwent subsequent resection; however, they were blinded to other clinical data. A training set of 5 patients from the cohort were reviewed by a third hepatopancreatobiliary radiologist with 9 years of experience (R.D.).
Tumor size was measured on axial images on any available contrast phase. PDAC tumor size was grouped according to the American Joint Committee on Cancer 8th Edition Staging criteria (≤ 2, > 2 to ≤ 4 cm, and > 4 cm). The common bile and pancreatic ducts were considered dilated if a stent was present or if their diameters were > 1.0 cm and > 0.4 cm, respectively. Tumor attenuation was subjectively graded as hypoattenuating, isoattenuating, or hyperattenuating in reference to the normal background pancreas in the portal venous phase. Tumors considered unmeasurable/not identifiable were grouped with tumors ≤ 2 cm and considered isoattenuating. The location of the tumor in the head, neck, or body/tail was recorded according to anatomical definitions in the APA/SAR template. Additionally, arterial and venous tumor contact were graded according to the template: no contact, abutment (≤ 180° contact), or encasement (> 180° contact) [29]. Distinction between solid soft-tissue and increased hazy attenuation/stranding vascular contact was not required. Associated vascular deformity, defined as any focal vessel narrowing or contour irregularity was also recorded. Length of tumor-vessel contact was measured by the reader on the imaging plane of their choice. Presence of venous collaterals was subjectively assessed and grouped into anatomical regions according to the APA/SAR template [29]. Tumor involvement of variant hepatic, proper hepatic, and common hepatic artery anatomy was also recorded.
In the case of feature disagreement between the two reviewing radiologists, the following tie-breaking adjustments were made: the larger or greater of tumor size, duct diameters, and tumor-vascular contact readings were taken as the final read. Tumor location disagreement between head and neck was categorized as neck, while disagreement between neck and body/tail was categorized as body/tail. Disagreement in tumor attenuation was resolved to the most hypoattenuating category recorded. Tumor invasion of other organs and the presence of varices were considered present at final read if there was reader disagreement. Disagreement between tumor extension terminating above the first order jejunal branch of the superior mesenteric vein and extending to or below the branch was resolved to the latter.
For the purpose of this study, patients were not classified into resectability groups, as the definition of resectable, borderline resectable, and locally advanced PDAC varied over the 11-year period captured in this study. In the neoadjuvant therapy group, the preoperative CTs analyzed were performed post neoadjuvant therapy.
Statistical Analysis
OS was defined as the time from surgery until death or last follow-up, where death was considered an event and those lost to follow-up were censored. Univariable associations between OS and the final read of each imaging variable were evaluated using Cox proportional hazards regression. This analysis was performed on the entire cohort, and separately on subsets of the upfront surgical and neoadjuvant therapy treatment groups. A p-value of 0.05 was used for significance. All analyses were performed using R version 3.5.2.
Results
Patient Demographics
A total of 168 patients were included in the final study population with an average age of 66 years and a slight predominance of women (91/168, 54%). There were 126/168 patients (75%) who had upfront surgical resection whereas 42/168 (25%) received neoadjuvant therapy prior to surgical resection. Median follow-up time was 22.1 months (range: 2.0–138.0 months).
Patients underwent CT on average 16 days prior to surgery. The CTs were performed with various imaging protocols, with 139/168 (83%) performed as dedicated pancreatic protocol CTs (Supplemental Table 1.)
Presurgical CT Findings
Across the entire 168-patient cohort at final read, 22/168 patients (13%) had tumors that were unmeasurable or ≤ 2 cm in largest axial diameter, 125/168 (74%) had tumors > 2 cm to ≤ 4 cm, and 21/168 (12%) had tumors > 4 cm (Table 1). Tumors were present in the pancreatic head in 116/168 (69%), neck in 19/168 (11%), and body/tail in 33/168 patients (20%).
Table 1.
Patient demographics and summary of the presence of CT template variables
| Upfront surgery (n = 126) | Neoadjuvant therapy (n = 42) | Entire cohort (n = 168) | |
|---|---|---|---|
|
Demographics |
|||
|
Mean Age (years) |
69 ± 10 | 64 ± 11 | 66 ± 11 |
|
Sex |
|||
| Man | 55 (44%) | 22 (52%) | 77 (46%) |
|
Woman |
71 (56%) | 20 (48%) | 91 (54%) |
|
CT Template Variable |
|||
|
Maximum tumor axial dimension |
|||
| ≤ 2 cm | 18 (14%) | 4 (9.5%) | 22 (13%) |
| > 2 ≤ 4cm | 94 (75%) | 31 (74%) | 125 (74%) |
| > 4 cm | 14 (11%) | 7 (17%) | 21 (13%) |
|
Tumor location |
|||
| Head | 90 (71%) | 26 (62%) | 116 (69%) |
| Neck | 12 (9.5%) | 7 (17%) | 19 (11%) |
| Body/tail | 24 (19%) | 9 (21%) | 33 (20%) |
|
Tumor attenuation |
|||
| Hypoattenuating | 98 (78%) | 39 (93%) | 137 (82%) |
| Isoattenuating | 28 (22%) | 3 (7.1%) | 31 (18%) |
|
Hyperattenuating |
0 | 0 | 0 |
|
PD dilatation |
106 (85%) | 31 (76%) | 137 (82%) |
|
CBD dilatation |
88 (70%) | 29 (69%) | 117 (70%) |
|
Invasion of any organ |
93 (74%) | 35 (83%) | 128 (76%) |
| Duodenum | 77 (61%) | 27 (64%) | 104 (62%) |
| Stomach | 24 (19%) | 9 (21%) | 33 (20%) |
| Spleen | 3 (2.4%) | 3 (7.1%) | 6 (4%) |
| Small bowel | 1 (0.8%) | 0 (0%) | 1 (< 1%) |
| Colon | 1 (0.8%) | 2 (4.8%) | 3 (2%) |
|
Adrenal |
1 (0.8%) | 3 (7.1%) | 4 (2%) |
|
Any tumor arterial contact |
27 (21%) | 20 (48%) | 47 (28%) |
|
Celiac axis |
|||
| No contact | 123 (98%) | 39 (93%) | 162 (96%) |
| Abutment | 3 (2.4%) | 0 (0%) | 3 (2%) |
| Encasement | 0 (0%) | 3 (7.1%) | 3 (2%) |
| Vessel deformity | 0 (0%) | 1 (2.4%) | 1 (< 1%) |
|
CHA |
|||
| No contact | 118 (94%) | 35 (83%) | 153 (91%) |
| Abutment | 7 (5.6%) | 5 (12%) | 12 (7%) |
| Encasement | 1 (0.8%) | 2 (4.8%) | 3 (2%) |
| Vessel deformity | 1 (0.8%) | 0 (0%) | 1 (< 1%) |
|
SMA |
|||
| No contact | 110 (87%) | 29 (69%) | 139 (83%) |
| Abutment | 16 (13%) | 10 (24%) | 26 (15%) |
| Encasement | 0 (0%) | 3 (7.1%) | 3 (2%) |
|
Vessel deformity |
0 (0%) | 2 (4.8%) | 2 (1%) |
|
Any tumor venous contact |
99 (79%) | 37 (88%) | 136 (81%) |
|
PV |
|||
| No contact | 106 (84%) | 30 (71%) | 136 (81%) |
| Abutment | 20 (16%) | 10 (24%) | 30 (18%) |
| Encasement | 0 (0%) | 2 (4.8%) | 2 (1%) |
| Vessel deformity | 8 (6.3%) | 5 (12%) | 13 (8%) |
|
SMV |
|||
| No contact | 49 (39%) | 16 (38%) | 65 (39%) |
| Abutment | 74 (59%) | 26 (62%) | 100 (60%) |
| Encasement | 3 (2.4%) | 0 (0%) | 3 (2%) |
| Vessel deformity | 51 (40%) | 15 (36%) | 66 (39%) |
|
Tumor vessel contact length (cm) |
0.95 (0.00, 1.58) | 1.10 (0.00, 1.60) | |
|
IVC |
1 (0.8%) | 0 (0%) | 1 (< 1%) |
| Deformity | 1 (0.8%) | 0 (0%) | 1 (< 1%) |
|
Tumor extension to first jejunal branch of SMV |
|||
| N/A | 49 (39%) | 16 (38%) | 65 (39%) |
| Above | 45 (36%) | 15 (36%) | 60 (36%) |
| At or below | 32 (25%) | 11 (26%) | 43 (26%) |
|
Venous collaterals present |
35 (28%) | 17 (40%) | 52 (31%) |
Abbreviations: PD = pancreatic duct, CBD = common bile duct, CHA = common hepatic artery, SMA = superior mesenteric artery, PV = portal vein, SMV = superior mesenteric vein, IVC = inferior vena cava
Tumor arterial contact was present at final read in 47/168 patients: 27/126 (21%) in the upfront surgical group and 20/42 (48%) in the neoadjuvant therapy group. Tumor venous contact was present in 136/168 patients: 99/126 (79%) in the upfront surgical group and 37/42 (88%) in the neoadjuvant therapy group.
Tumor Morphology
Increasing tumor size was associated with decreased OS in patients who underwent upfront surgical resection (Hazard Ratio (HR) 2.30, 95% Confidence Interval (CI) 1.19–4.42, p = 0.013). There was no association between tumor size and OS for the entire cohort (HR 1.59, 95% CI 0.97–2.59, p = 0.064) or in the neoadjuvant therapy group (HR 0.8, 95% CI 0.36–1.77, p = 0.6).
Tumor location in the pancreas, tumor attenuation, and the presence of bile or pancreatic ductal dilatation were not associated with OS in the entire cohort or either treatment group (see Table 2). 25% of tumors were classified as isoattenuating at the final read and no tumors were classified as hyperattenuating (Table 1).
Table 2.
Overall survival analysis for CT template variables
| Upfront Surgery (n = 126) |
Neoadjuvant Therapy (n = 42) |
Entire Cohort (n = 168) |
|||||||
|---|---|---|---|---|---|---|---|---|---|
| Template Variable | HR | 95% CI | P-value | HR | 95% CI | P-value | HR | 95% CI | P-value |
|
Tumor morphology |
|||||||||
| Maximum axial diameter | 2.3 | 1.19, 4.42 | 0.01 | 0.8 | 0.36, 1.77 | 0.6 | 1.59 | 0.97, 2.59 | 0.06 |
| Tumor area | 1.05 | 1.01, 1.10 | 0.03 | 1 | 0.90, 1.11 | > 0.9 | 1.04 | 1.00, 1.08 | 0.1 |
| Tumor location | 0.87 | 0.59, 1.27 | 0.5 | 0.69 | 0.39, 1.19 | 0.2 | 0.8 | 0.59, 1.08 | 0.13 |
| Tumor attenuation | 1.4 | 0.78, 2.53 | 0.3 | 0.62 | 0.14, 2.77 | 0.5 | 1.07 | 0.63, 1.81 | 0.8 |
|
Duct dilatation |
|||||||||
| Pancreatic duct | 0.77 | 0.39, 1.55 | 0.5 | 0.61 | 0.27, 1.39 | 0.3 | 0.69 | 0.41, 1.16 | 0.2 |
| Common bile duct | 0.94 | 0.51, 1.71 | 0.8 | 1.51 | 0.63, 3.61 | 0.3 | 1.09 | 0.67, 1.76 | 0.7 |
|
Invasion of adjacent organ |
|||||||||
| Any organ | 1.5 | 0.75, 3.00 | 0.2 | 1.86 | 0.62, 5.54 | 0.2 | 1.73 | 0.97, 3.11 | 0.051 |
| Invasion of duodenum | 1.21 | 0.67, 2.19 | 0.5 | 1.34 | 0.57, 3.13 | 0.5 | 1.33 | 0.83, 2.14 | 0.2 |
| Invasion of stomach | 0.86 | 0.40, 1.83 | 0.7 | 1.23 | 0.49, 3.10 | 0.7 | 1.05 | 0.60, 1.84 | 0.9 |
| Invasion of spleen | 1.73 | 0.23, 12.9 | 0.6 | 3.88 | 0.84, 17.9 | 0.13 | 2.83 | 0.87, 9.18 | 0.13 |
| Invasion of adrenal | - | - | - | 0.97 | 0.22, 4.27 | > 0.9 | 1.12 | 0.27, 4.60 | 0.9 |
| Invasion of colon | - | - | - | - | - | - | 1.91 | 0.26, 14.0 | 0.6 |
| Invasion of small bowel | - | - | - | - | - | - | - | - | - |
|
Tumor arterial contact |
|||||||||
| Any vessel | 1.82 | 0.93, 3.59 | 0.5 | 1.32 | 0.57, 3.08 | 0.5 | 1.89 | 1.13, 3.14 | 0.02 |
| Tumor celiac axis contact | 1.54 | 0.82, 2.88 | 0.2 | 1.99 | 1.07, 3.71 | 0.07 | |||
| CHA contact | 4.29 | 1.79, 10.3 | 0.8 | 1.3 | 0.61, 2.77 | 0.5 | 2.33 | 1.35, 4.04 | 0.009 |
| CHA deformity | 9.53 | 1.24, 73.3 | 0.008 | 5.96 | 0.81, 44.0 | 0.14 | |||
| SMA contact | 1.42 | 0.60, 3.33 | 0.4 | 1.44 | 0.72, 2.87 | 0.3 | 1.75 | 1.03, 2.97 | 0.2 |
| SMA deformity | - | - | - | 2.65 | 0.34, 20.4 | 0.4 | 5.3 | 0.72, 39.0 | 0.2 |
|
Tumor venous contact |
|||||||||
| Any major vein | 0.91 | 0.48, 1.71 | 0.8 | 0.9 | 0.30, 2.71 | 0.9 | 1 | 0.58, 1.71 | > 0.9 |
| PV | 1.26 | 0.64, 2.49 | 0.5 | 0.73 | 0.32, 1.68 | 0.4 | 1.08 | 0.65, 1.80 | 0.8 |
| PV deformity | 1.69 | 0.49, 5.77 | 0.4 | 3.22 | 1.63, 6.37 | 0.003 | |||
| SMV or PV confluence | 1.12 | 0.67, 1.87 | 0.7 | 1.27 | 0.54, 2.97 | 0.6 | 1.21 | 0.80, 1.84 | 0.4 |
| SMV or PV confluence deformity | 0.94 | 0.54, 1.63 | 0.8 | 1.47 | 0.65, 3.32 | 0.4 | 1.09 | 0.70, 1.71 | 0.7 |
| SMV or PV confluence tumor contact length | 1.11 | 0.81, 1.53 | 0.5 | 1.11 | 0.69, 1.79 | 0.7 | 1.16 | 0.90, 1.50 | 0.3 |
| Tumor extension to or below first jejunal branch | 0.99 | 0.71, 1.38 | > 0.9 | 1.13 | 0.69, 1.86 | 0.6 | 1.07 | 0.82, 1.40 | 0.6 |
| Venous collateral present | 0.95 | 0.52, 1.73 | 0.9 | 2.28 | 1.04, 4.99 | 0.04 | 1.33 | 0.84, 2.12 | 0.2 |
Abbreviations: CHA = common hepatic artery, SMA = superior mesenteric artery, PV = portal vein, SMV = superior mesenteric vein
Tumor-Vascular Involvement
Arterial involvement
Tumor arterial contact of any kind was associated with decreased OS in the entire cohort (HR 1.89, 95% CI 1.13–3.14, p = 0.020) (Figure 2). Additionally, in the entire cohort, tumor contact specifically with the common hepatic artery was associated with decreased OS (HR 2.33, 95% CI 1.35–4.04, p = 0.009). In the upfront surgery group, only contact with the common hepatic artery was associated with decreased OS (HR 4.29, 95% CI 1.79–10.3, p=0.008). Tumor arterial contact in the neoadjuvant therapy group was not associated with decreased OS (HR 1.32, 95% CI 0.57–3.08, p = 0.5) (Table 2). Figure 3 is an example patient with both tumor arterial and venous involvement.
Fig. 2.
Kaplan–Meier overall survival curve for tumor arterial contact (entire cohort)
Fig. 3.
72-year-old female with pancreatic neck adenocarcinoma; (A) axial pancreatic arterial phase CT and (B) coronal portal venous phase reconstructions. The tumor (red dotted outline) encases the splenic artery (short blue arrow), abuts the common hepatic artery (long blue arrow) and abuts the portosplenic confluence (dotted blue outline). The patient underwent distal pancreatectomy with splenectomy and en bloc resection of the celiac axis (Appleby procedure). The patient’s overall survival was 7.5 months
Venous involvement
Portal vein deformity was associated with decreased OS across the entire cohort (HR 3.22, 95% CI 1.63–6.37, p = 0.003). Otherwise, the presence tumor venous involvement (abutment or encasement) and the length of tumor venous contact were not associated with OS in the entire cohort or in either treatment group (Table 2). Additionally, tumor involvement of the superior mesenteric vein extending to and/or below the first order jejunal branch was not associated with OS (HR 1.07, 95% CI 0.82–1.40, p = 0.6). The presence of venous collaterals was associated with decreased OS in the neoadjuvant therapy group only (HR 2.28, 95% CI 1.04–4.99, p = 0.042). Figure 4 is an example patient with tumor venous involvement.
Fig 4.
70-year-old female with pancreatic head adenocarcinoma; (A) and (B) axial portal venous phase and (C) coronal portal venous phase. The uncinate process tumor (red dotted outline) abuts the portal vein (red arrow), superior mesenteric vein (short blue arrow) at the level of the first jejunal tributary (blue dotted line) and second jejunal tributary (long blue arrow). The patient underwent pancreaticoduodenectomy with venous resection near the first jejunal tributary. The patient’s overall survival was 3 years and 11 months
Inter-observer agreement
Template variable inter-observer agreement kappa and percent agreement are detailed in Supplemental Table 2. Overall there was moderate inter-observer agreement in the entire cohort and both treatment groups, with no difference between the treatment groups (p = 0.728). Variables where the kappa was not calculated due to insufficient occurrences in either of the treatment group were excluded from this calculation (8/33).
Discussion
Our results demonstrate that several imaging features in the APA/SAR PDAC reporting template are associated with OS in patients who were treated with curative intent surgery. Notably, in our entire study cohort, tumor arterial contact of any kind, contact with the common hepatic artery, and portal vein deformity were associated with decreased OS. In contradistinction, other template features such as the presence, length, and circumferential degree of tumor venous contact; tumor attenuation; bile/pancreatic duct dilatation; and invasion of adjacent organs were not associated with OS in our study cohort.
Our finding that tumor arterial contact, and specifically tumor contact with the common hepatic artery, was associated with OS is partially in line with the prior literature. A recent study by Toesca et al. [6] (n = 294) demonstrated that the increasing degree of combined tumor contact with the superior mesenteric artery, celiac axis, and common hepatic artery were associated with OS and metastatic-free survival; however, this study included patients with unresectable PDAC in contrast to resectable PDAC in our study. Similarly, Hayasaki et al. [4] (n = 285) found that celiac axis, common hepatic artery, and superior mesenteric artery encasement was associated with worse OS; however, the study population all received neoadjuvant therapy and included patients with inoperable disease after neoadjuvant therapy and 26 patients with metastatic disease found at laparotomy. Yamada et al. [11] (n = 382) found that superior mesenteric artery/celiac axis abutment and encasement was associated with decreased survival compared with no tumor artery contact; however, this study excluded patients with prior neoadjuvant therapy. A study by Kozak et al. [12] included exclusively patients with borderline resectable and locally advanced PDAC (n = 69), of whom only 22% went onto resection; in their study, common hepatic artery abutment or encasement was associated with decreased OS.
There are limited prior studies assessing the prognostic value of local organ invasion. Chang et al. [15] showed that tumor invasion of the duodenum assessed on preoperative CT in 76 patients undergoing pancreaticoduodenectomy (4% with neoadjuvant therapy) was associated with decreased progression free survival (PFS) and OS. Our study found no association between local organ invasion, including specifically duodenal invasion, and OS. Our study included a larger study population, including a larger proportion that received neoadjuvant therapy before surgical resection. The discordant finding is possibly compounded by the different classification methods of duodenal invasion; Chang et al.’s [15] classification was by consensus, compared to our tie-breaking method, which is probably more sensitive and less specific. Chang et al. [15] also demonstrated that extrapancreatic perineural invasion on preoperative CT was associated with decreased survival; however, this was not assessed in our study as it was not part of the APA/SAR template.
The rate of isoattenuating tumors (22% of 126 undergoing upfront resection) was higher than that reported by Kim et al. [10] (11.9% of 253 patients undergoing curative resection). This is possibly due to our less stringent classification requirement that tumor only be isoattenuating in the portal venous phase, whereas Kim et al. required both arterial and portal venous isoattenuation for classification. The different classification rate is possibly why our finding of no association between tumor attenuation and OS in the whole cohort or in either treatment group was discordant with Kim et al., who found that isoattenuating PDACs had longer median survival compared to other PDAC appearances [10].
Prior literature has demonstrated that the severity of tumor venous involvement on preoperative imaging is associated with OS [3–6,11]; however, our study showed that only portal vein deformity in the entire cohort demonstrated an association with decreased OS. This includes no significance of tumor involvement extending to or below the first order jejunal branch draining into the superior mesenteric vein. The discordance is probably due to differences in study design including different patient selection criteria, CT assessment at different stages of treatment, assessment of vascular involvement by intraoperative imaging, and inclusion of patients with different initial resectability classifications. Local practice patterns and surgical expertise may also affect OS in these patients as the subjectivity of unresectable disease and the willingness to perform venous resection will vary between institutions.
CT assessment after neoadjuvant therapy has been shown to be limited for staging and assessing tumor response [21–27]. Our findings echo these studies in that tumor size, the presence and severity of tumor arterial involvement, and adjacent organ invasion are not predictive of OS in the neoadjuvant therapy group. This is further supported by literature that surgical exploration should be considered in all patients who have not progressed on neoadjuvant therapy, whether or not the primary tumor has demonstrated a significant response to treatment [21,31–33]. However, radical surgical approaches involving arterial resections must be confined to highly selected patients, given that meta-analyses have found they are associated with higher rates of postoperative morbidity [34] and worse short and long-term outcomes [35]. Evaluation of treatment response in the neoadjuvant therapy group will be separately assessed as part of a multistage project.
Previous studies have also demonstrated reduced inter-observer agreement in the setting of neoadjuvant therapy [26,31,32], a trend that was not observed in our study. This is likely due to differences in study population and design; for example, Beleu et al. [32] included a general radiologist as 1 of 3 readers. The APA/SAR encourages the use of the template by general radiologists and outcomes must be reproducible amongst both expert and generalist readers. The high inter-observer agreements demonstrated in other studies in relation to PDAC staging were likely due to these studies classifying tumors into broad resection categories, rather than single feature assessment [36,37].
Although the majority of scans used for assessment were dedicated pancreatic protocol CTs (83%), the inclusion of non pancreatic protocol CTs is a limitation of this study, a consequence of the retrospective nature of this study. Due to this limitation the standardized scoresheet graded tumor attenuation in the portal venous phase, however the APA/SAR PDAC reporting template recommends specification of tumor attenuation in the pancreatic parenchymal phase. Some imaging protocols will have also changed over the near 11-year period of study inclusion. Further, the results may have been influenced by the readers’ awareness that all the patients included ultimately underwent curative intent surgery. Additionally, including only patients with genetic tumor analysis and non-stratification for demographic or clinical factors limits the study. The future application of these results is limited to patients who are deemed surgically resectable at our institution; their prognostic utility is unknown in patients with resectable disease who are precluded from curative resection due to functional status or other factors. Further, the group who received neoadjuvant therapy in our study was heterogeneous as the specific chemotherapy regimen and the use of radiation was not standardized between patients over the time period included. As most of the patients did not undergo radiation therapy the standardized scoresheet did not require distinction between solid soft-tissue and increased hazy attenuation/stranding vascular contact, however this distinction is recommended in the APA/SAR PDAC reporting template. Univariable analysis was undertaken as the number of patients and limited number of events was insufficient to build a multivariable model. The data collected in this study will form part of future multivariable analysis including demographic, histologic, serologic, and genomic parameters.
In conclusion, the APA/SAR PDAC reporting template offers a number of preoperative CT features that may serve as a prognostic tool in pancreatic adenocarcinoma patients selected for surgical resection. The presence of arterial involvement was associated with decreased OS in our cohort of patients. Aside from portal vein deformity, the presence and extent of tumor venous involvement were not associated with OS, lending support to the current surgical approach in patients undergoing curative intent surgery with tumors categorized as borderline/locally advanced due to tumor venous involvement. In the group of patients who received neoadjuvant therapy prior to surgical resection, however, nearly all template features failed to show associations with OS, reflecting the difficulty in assessing tumor extent in this setting. Nevertheless, our study shows the potential advantage of using a pancreas ductal adenocarcinoma reporting template to not only improve communication with the surgical team, but to build databases that facilitate the investigation of associations between specific imaging findings and patients or surgical outcomes. Routine use of the template across institutions may thus create opportunities to obtain larger scale data in multi-center studies.
Supplementary Material
Supplemental table 1. CT protocol specifications.
Supplemental table 2. Interobserver agreement and percentage agreement for CT template features.
Funding:
This study was funded in part through the NIH/NCI Cancer Center Support Grant P30 CA008748.
Footnotes
Publisher's Disclaimer: This Author Accepted Manuscript is a PDF file of an unedited peer-reviewed manuscript that has been accepted for publication but has not been copyedited or corrected. The official version of record that is published in the journal is kept up to date and so may therefore differ from this version.
Conflicts of interest/Competing interests: The authors declare that they have no conflict of interest or competing interests.
Ethics approval: Institutional Review Board approved. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. For this type of study formal consent is not required. This article does not contain any studies with animals performed by any of the authors.
Consent to participate: The Institutional Review Board waived the need for written informed consent for this retrospective analysis. The study was compliant with the Health Insurance Portability and Accountability Act.
Consent for publication: All authors consent to publication. This manuscript has not been previously published, nor is under consideration for publication elsewhere.
Availability of data and material: Data can be made available for review.
Code availability: Not applicable
Specific Remark: We thank Joanne Chin for editorial assistance.
References
- 1.Rahib L, Smith BD, Aizenberg R, Rosenzweig AB, Fleshman JM, Matrisian LM (2014) Projecting Cancer Incidence and Deaths to 2030: The Unexpected Burden of Thyroid, Liver, and Pancreas Cancers in the United States. 10.1158/0008-5472.CAN-14-0155 [DOI] [PubMed] [Google Scholar]
- 2.Howlader N, Noone AM, Krapcho M, Miller D, Bishop K, Altekruse SF, Kosary CL, Yu M, Ruhl J, Tatalovich Z, Mariotto A, Lewis DR, Chen HS, Feuer EJCK (2015) Cancer Statistics Review, 1975–2013 - SEER Statistics. SEER Cancer Stat. Rev. 1975–2013, Natl. Cancer Institute. Bethesda, MD. [Google Scholar]
- 3.Nakao A, Kanzaki A, Fujii T, Kodera Y, Yamada S, Sugimoto H, Nomoto S, Nakamura S, Morita S, Takeda S (2012) Correlation Between Radiographic Classification and Pathological Grade of Portal Vein Wall Invasion in Pancreatic Head Cancer. Ann Surg 255:103–108. 10.1097/SLA.0b013e318237872e [DOI] [PubMed] [Google Scholar]
- 4.Hayasaki A, Isaji S, Kishiwada M, Fujii T, Iizawa Y, Kato H, Tanemura A, Murata Y, Azumi Y, Kuriyama N, Mizuno S, Usui M, Sakurai H (2018) Survival analysis in patients with pancreatic ductal adenocarcinoma undergoing chemoradiotherapy followed by surgery according to the international consensus on the 2017 definition of borderline resectable cancer. Cancers (Basel) 10:. 10.3390/cancers10030065 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Tran Cao HS, Balachandran A, Wang H, Nogueras-González GM, Bailey CE, Lee JE, Pisters PWT, Evans DB, Varadhachary G, Crane CH, Aloia TA, Vauthey J-N, Fleming JB, Katz MHG (2014) Radiographic tumor-vein interface as a predictor of intraoperative, pathologic, and oncologic outcomes in resectable and borderline resectable pancreatic cancer. J Gastrointest Surg 18:269–278. 10.1007/s11605-013-2374-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Toesca DAS, Jeffrey RB, von Eyben R, Pollom EL, Poullos PD, Poultsides GA, Fisher GA, Visser BC, Koong AC, Chang DT (2019) Predicting Pancreatic Cancer Resectability and Outcomes Based on an Objective Quantitative Scoring System. Pancreas 48:622–628. 10.1097/MPA.0000000000001314 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Cassinotto C, Chong J, Zogopoulos G, Reinhold C, Chiche L, Lafourcade J-P, Cuggia A, Terrebonne E, Dohan A, Gallix B (2017) Resectable pancreatic adenocarcinoma: Role of CT quantitative imaging biomarkers for predicting pathology and patient outcomes. Eur J Radiol 90:152–158. 10.1016/j.ejrad.2017.02.033 [DOI] [PubMed] [Google Scholar]
- 8.Ishikawa O, Ohigashi H, Imaoka S, Furukawa H, Sasaki Y, Fujita M, Kuroda C, Iwanaga T (1992) Preoperative indications for extended pancreatectomy for locally advanced pancreas cancer involving the portal vein. Ann Surg 215:231–6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Zhu L, Shi X, Xue H, Wu H, Chen G, Sun H, He Y, Jin Z, Liang Z, Zhang Z (2016) CT Imaging Biomarkers Predict Clinical Outcomes After Pancreatic Cancer Surgery. Medicine (Baltimore) 95:e2664. 10.1097/MD.0000000000002664 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Jin Hee Kim M, Seong Ho Park M, Eun Sil Yu M, Myung-Hwan Kim M, Jihun Kim M, Jae Ho Byun M, Seung Soo Lee M, Hye Jeon Hwang M, Jae-Yeon Hwang M, Sang Soo Lee M, Moon-Gyu Lee M (2010) Visually Isoattenuating Pancreatic Adenocarcinoma at Dynamic-Enhanced CT: Frequency, Clinical and Pathologic Characteristics, and Diagnosis at Imaging Examinations. Radiology 257:87–96. 10.1148/radiol.10100015/-/DC1 [DOI] [PubMed] [Google Scholar]
- 11.Yamada S, Fujii T, Takami H, Hayashi M, Iwata N, Kanda M, Tanaka C, Sugimoto H, Nakayama G, Koike M, Fujiwara M, Kodera Y (2017) Evaluation and proposal of novel resectability criteria for pancreatic cancer established by the Japan Pancreas Society. Surg (United States) 162:784–791. 10.1016/j.surg.2017.04.023 [DOI] [PubMed] [Google Scholar]
- 12.Kozak GM, Epstein JD, Deshmukh SP, Scott BB, Keith SW, Lavu H, Yeo CJ, Winter JM (2018) Common Hepatic Artery Abutment or Encasement Is an Adverse Prognostic Factor in Patients with Borderline and Unresectable Pancreatic Cancer. J Gastrointest Surg 22:288–294. 10.1007/s11605-017-3595-7 [DOI] [PubMed] [Google Scholar]
- 13.Park MY, Shin SH, Song KB, Hwang D, Lee JH, Lee Y-J, Kim SC (2019) Validation of the eighth edition of the American Joint Committee on Cancer staging system and proposal of an improved staging system for pancreatic ductal adenocarcinoma. Ann Hepato-Biliary-Pancreatic Surg 23:46. 10.14701/ahbps.2019.23.1.46 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Shrestha A, Baskota A, Li FY, Ma W-J, Yang Q, Jie H-H, Bijukchhe SM (2018) Impact of tumor size on survival outcome of pancreatic carcinoma following pancreatic resection: A systematic review and meta-analysis. Asian J Med Sci 9:1–10. 10.3126/ajms.v9i1.18714 [DOI] [Google Scholar]
- 15.Chang ST, Jeffrey RB, Patel BN, DiMaio MA, Rosenberg J, Willmann JK, Olcott EW (2016) Preoperative Multidetector CT Diagnosis of Extrapancreatic Perineural or Duodenal Invasion Is Associated with Reduced Postoperative Survival after Pancreaticoduodenectomy for Pancreatic Adenocarcinoma: Preliminary Experience and Implications for Patient C. Radiology 281:816–825. 10.1148/radiol.2016152790 [DOI] [PubMed] [Google Scholar]
- 16.Pak LM, Gonen M, Seier K, Balachandran VP, D’Angelica MI, Jarnagin WR, Kingham TP, Allen PJ, Do RKG, Simpson AL (2018) Can physician gestalt predict survival in patients with resectable pancreatic adenocarcinoma? Abdom Radiol 43:2113–2118. 10.1007/s00261-017-1407-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Arvold ND, Niemierko A, Mamon HJ, Fernandez-Del Castillo C, Hong TS (2011) Pancreatic cancer tumor size on CT scan versus pathologic specimen: Implications for radiation treatment planning. Int J Radiat Oncol Biol Phys 80:1383–1390. 10.1016/j.ijrobp.2010.04.058 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Hong SMBM, Lee SS, Kim JH, Kim HJ, Byun JH, Hong SMBM, Song K-B, Kim SC (2018) Pancreatic Cancer CT: Prediction of Resectability according to NCCN Criteria. Radiology 289:710–718. 10.1148/radiol.2018180628 [DOI] [PubMed] [Google Scholar]
- 19.Cassinotto C, Dohan A, Zogopoulos G, Chiche L, Laurent C, Sa-Cunha A, Cuggia A, Reinhold C, Gallix B (2017) Pancreatic adenocarcinoma: A simple CT score for predicting margin-positive resection in patients with resectable disease. Eur J Radiol 95:33–38. 10.1016/j.ejrad.2017.06.028 [DOI] [PubMed] [Google Scholar]
- 20.Shen YN, Guo CX, Pan Y, Chen YW, Tang TY, Li YW, Lu JH, Jin G, Qin RY, Yao WY, Liang TB, Bai XL (2018) Preoperative prediction of peripancreatic vein invasion by pancreatic head cancer. Cancer Imaging 18:. 10.1186/s40644-018-0179-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Katz MHG, Fleming JB, Bhosale P, Varadhachary G, Lee JE, Wolff R, Wang H, Abbruzzese J, Pisters PWT, Vauthey JN, Charnsangavej C, Tamm E, Crane CH, Balachandran A (2012) Response of borderline resectable pancreatic cancer to neoadjuvant therapy is not reflected by radiographic indicators. Cancer 118:5749–5756. 10.1002/cncr.27636 [DOI] [PubMed] [Google Scholar]
- 22.Cassinotto C, Cortade J, Belleannée G, Lapuyade B, Terrebonne E, Vendrely V, Laurent C, Sa-Cunha A (2013) An evaluation of the accuracy of CT when determining resectability of pancreatic head adenocarcinoma after neoadjuvant treatment. Eur J Radiol 82:589–593. 10.1016/j.ejrad.2012.12.002 [DOI] [PubMed] [Google Scholar]
- 23.Wagner M, Antunes C, Pietrasz D, Cassinotto C, Zappa M, Sa Cunha A, Lucidarme O, Bachet J-BB (2017) CT evaluation after neoadjuvant FOLFIRINOX chemotherapy for borderline and locally advanced pancreatic adenocarcinoma. Eur Radiol 27:3104–3116. 10.1007/s00330-016-4632-8 [DOI] [PubMed] [Google Scholar]
- 24.Ferrone CR, Marchegiani G, Hong TS, Ryan DP, Deshpande V, McDonnell EI, Sabbatino F, Santos DD, Allen JN, Blaszkowsky LS, Clark JW, Faris JE, Goyal L, Kwak EL, Murphy JE, Ting DT, Wo JY, Zhu AX, Warshaw AL, Lillemoe KD, Castillo CF Del (2015) Radiological and Surgical Implications of Neoadjuvant Treatment With FOLFIRINOX for Locally Advanced and Borderline Resectable Pancreatic Cancer. Ann Surg 261:12–17. 10.1097/SLA.0000000000000867 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Kim YE, Park MS, Hong HS, Kang CM, Choi JY, Lim JS, Lee WJ, Kim MJ, Kim KW (2009) Effects of neoadjuvant combined chemotherapy and radiation therapy on the CT evaluation respectability and staging in patients with pancreatic head cancer. Radiology 250:758–765. 10.1148/radiol.2502080501 [DOI] [PubMed] [Google Scholar]
- 26.Morgan DE, Waggoner CN, Canon CL, Lockhart ME, Fineberg NS, Posey Iii JA, Vickers SM, Posey JA, Vickers SM (2010) Resectability of pancreatic adenocarcinoma in patients with locally advanced disease downstaged by preoperative therapy: A challenge for MDCT. Am J Roentgenol 194:615–622. 10.2214/AJR.08.1022 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Zins M, Matos C, Cassinotto C (2018) Pancreatic Adenocarcinoma Staging in the Era of Preoperative Chemotherapy and Radiation Therapy. Radiology 287:374–390. 10.1148/radiol.2018171670 [DOI] [PubMed] [Google Scholar]
- 28.Tempero MA, Malafa MP, Al-Hawary M, Asbun H, Bain A, Behrman SW, Benson AB, Binder E, Cardin DB, Cha C, Chiorean EG, Chung V, Czito B, Dillhoff M, Dotan E, Ferrone CR, Hardacre J, Hawkins WG, Herman J, Ko AH, Komanduri S, Koong A, LoConte N, Lowy AM, Moravek C, Nakakura EK, O’Reilly EM, Obando J, Reddy S, Scaife C, Thayer S, Weekes CD, Wolff RA, Wolpin BM, Burns J, Darlow S (2017) Pancreatic Adenocarcinoma, Version 2.2017, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc Netw 15:1028–1061. 10.6004/jnccn.2017.0131 [DOI] [PubMed] [Google Scholar]
- 29.Al-Hawary MM, Francis IR, Chari ST, Fishman EK, Hough DM, Lu DS, Macari M, Megibow AJ, Miller FH, Mortele KJ, Merchant NB, Minter RM, Tamm EP, Sahani DV, Simeone DM (2014) Pancreatic ductal adenocarcinoma radiology reporting template: Consensus statement of the society of abdominal radiology and the American pancreatic association. Gastroenterology 146:248–260. 10.1053/j.gastro.2013.11.004 [DOI] [PubMed] [Google Scholar]
- 30.Brook OR, Brook A, Vollmer CM, Kent TS, Sanchez N, Pedrosa I (2015) Structured Reporting of Multiphasic CT for Pancreatic Cancer: Potential Effect on Staging and Surgical Planning. Radiology 274:464–472. 10.1148/radiol.14140206 [DOI] [PubMed] [Google Scholar]
- 31.Cassinotto C, Mouries A, Lafourcade J-P, Terrebonne E, Belleannée G, Blanc J-F, Lapuyade B, Vendrely V, Laurent C, Chiche L, Wagner T, Sa-Cunha A, Gaye D, Trillaud H, Laurent F, Montaudon M (2014) Locally Advanced Pancreatic Adenocarcinoma: Reassessment of Response with CT after Neoadjuvant Chemotherapy and Radiation Therapy. Radiology 273:108–116. 10.1148/radiol.14132914 [DOI] [PubMed] [Google Scholar]
- 32.Beleù A, Calabrese A, Rizzo G, Capelli P, Bellini N, Caloggero S, Calbi R, Martini PT, De Robertis R, Carbognin G, Marchegiani G, Scarpa A, Salvia R, Bassi C, D’onofrio M (2019) Preoperative Imaging Evaluation after Downstaging of Pancreatic Ductal Adenocarcinoma: A Multi-Center Study. Cancers (Basel) 11:267. 10.3390/cancers11020267 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Michelakos T, Pergolini I, Castillo CF Del, Honselmann KC, Cai L, Deshpande V, Wo JY, Ryan DP, Allen JN, Blaszkowsky LS, Clark JW, Murphy JE, Nipp RD, Parikh A, Qadan M, Warshaw AL, Hong TS, Lillemoe KD, Ferrone CR (2019) Predictors of Resectability and Survival in Patients With Borderline and Locally Advanced Pancreatic Cancer who Underwent Neoadjuvant Treatment With FOLFIRINOX. Ann Surg 269:733–740. 10.1097/SLA.0000000000002600 [DOI] [PubMed] [Google Scholar]
- 34.Gong H, Ma R, Gong J, Cai C, Song Z, Xu B (2016) Distal pancreatectomy with en bloc celiac axis resection for locally advanced pancreatic cancer: A systematic review and meta-analysis. Med (United States) 95:. 10.1097/MD.0000000000003061 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Mollberg N, Rahbari NN, Koch M, Hartwig W, Hoeger Y, Büchler MW, Weitz J (2011) Arterial resection during pancreatectomy for pancreatic cancer: A systematic review and meta-analysis. Ann. Surg 254:882–893 [DOI] [PubMed] [Google Scholar]
- 36.Loizou L, Duran CV, Axelsson E, Andersson M, Keussen I, Strinnholm J, Bartholomä W, Del Chiaro M, Segersvärd R, Lundell L, Kartalis N (2018) Radiological assessment of local resectability status in patients with pancreatic cancer: Interreader agreement and reader performance in two different classification systems. Eur J Radiol 106:69–76. 10.1016/j.ejrad.2018.06.014 [DOI] [PubMed] [Google Scholar]
- 37.Loizou L, Albiin N, Ansorge C, Andersson M, Segersvärd R, Leidner B, Sundin A, Lundell L, Kartalis N (2013) Computed tomography staging of pancreatic cancer: A validation study addressing interobserver agreement. Pancreatology 13:570–575. 10.1016/j.pan.2013.09.004 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplemental table 1. CT protocol specifications.
Supplemental table 2. Interobserver agreement and percentage agreement for CT template features.