Updates on Imaging Assessment of Pancreatic Cancer for Determining Anatomic and Biologic Resectability

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is a highly lethal malignancy. Imaging plays a pivotal role in the diagnosis and management of patients with PDAC, with treatment strategies largely guided by the tumor stage and resectability at initial diagnosis. The pancreatic CT protocol is the preferred first-line imaging modality, while MRI and PET/CT serve as problem-solving tools. In non-metastatic PDAC, anatomical resectability is mainly determined by the extent of tumor–vessel contact and is categorized as resectable, borderline resectable, or locally advanced disease. Although treatment decisions have traditionally relied on anatomical resectability, a growing body of evidence has highlighted the limitations of this approach. Consequently, an expanded concept of resectability that incorporates anatomical, biological, and host-related conditional factors has emerged. Potential biomarkers that reflect tumor biology include carbohydrate antigen 19-9 levels, imaging tumor phenotype, lymph node status, and metabolic activity on PET/CT. Neoadjuvant therapy (NAT) is widely used to treat borderline resectable or locally advanced disease. However, radiological restaging after NAT remains challenging and tends to overestimate vascular invasion, probably because imaging cannot reliably differentiate viable tumors from post-treatment fibrosis. In this context, biological factors may provide incremental value for treatment-response assessment. This review provides an updated overview of imaging techniques, assessments of anatomical resectability, challenges in radiological restaging after NAT, and biological and conditional factors that may enable more refined risk stratification in patients with PDAC.

INTRODUCTION

Pancreatic ductal adenocarcinoma (PDAC) is a highly lethal malignancy with a reported five-year survival rate of approximately 10% [1]. Complete surgical resection is the only potentially curative treatment for PDAC. However, only 20% of newly diagnosed patients present with early-stage disease amenable to curative-intent resection [2]. Since the introduction of multiagent chemotherapy regimens, neoadjuvant therapy (NAT) has been increasingly adopted as the initial treatment approach for PDAC, with the expectation of improving surgical candidacy through conversion surgery and enhancing oncological outcomes.

Imaging evaluation plays a central role in the clinical decision-making process for patients with PDAC. The pancreatic CT protocol remains the cornerstone for clinical staging and assessing anatomical resectability, whereas MRI and fluorine-18-fluorodeoxyglucose (¹⁸F-FDG) PET/CT serve as valuable problem-solving tools [3,4]. Traditionally, treatment strategies for PDAC have been primarily guided by the tumor resectability status, with the resectability of non-metastatic PDAC determined largely by the extent of tumor–vessel contact. However, the high rate of recurrence after surgical resection, even among patients with anatomically resectable disease, highlights the limitations of treatment decision-making based on anatomical resectability alone [5]. Consequently, the concept of biological resectability has emerged, emphasizing the integration of imaging, laboratory, and biological biomarkers that may reflect tumor aggressiveness in therapeutic decision-making [5,6,7].

Patients initially treated with NAT should undergo reassessment to determine the feasibility of curative resection. However, radiological restaging is more challenging than initial staging because imaging cannot reliably differentiate viable tumors from post-treatment fibrosis. Various approaches have been proposed to address these challenges.

Accordingly, this review provides an updated overview of radiologic examinations, imaging-based assessments of tumor resectability, and biological and conditional factors that should be considered when determining optimal treatment strategies for patients with PDAC, as well as the challenges in radiologic restaging after NAT.

IMAGING TECHNIQUES

For accurate tumor staging and assessment of resectability in patients with PDAC, high-quality, dedicated imaging of the pancreas should be performed at initial presentation and at restaging after NAT. Most clinical practice guidelines recommend pancreatic protocol CT or MRI as the standard staging examination, with pancreatic protocol CT of the abdomen and pelvis being the preferred modality [3,8,9,10]. Some guidelines also recommend chest CT for complete staging [8,9].

Pancreatic CT Protocol

The dedicated pancreatic CT protocol requires technical optimization to ensure accurate evaluation of the tumor extent and vascular involvement. Essential components include multidetector CT scanners (preferably ≥64 detector rows), thin detector collimation, rapid intravenous contrast injection (3–5 mL/s), and dual-phase imaging consisting of pancreatic and portal venous phases [3,11]. The pancreatic phase provides maximal contrast between the hypovascular tumor and the enhanced normal pancreatic parenchyma, facilitating tumor detection and assessment of tumor contact with the major arteries. The portal venous phase is optimal for evaluating venous tumor involvement and adjacent organ invasion and for detection of liver metastases. In some institutions, delayed-phase imaging is performed in addition to biphasic imaging, since it may improve the detection of PDAC, particularly by revealing tumors that appear isoattenuating on biphasic imaging [12]. According to the National Comprehensive Cancer Network (NCCN) guidelines, the recommended scan times for the pancreatic and portal venous phases are 40–50 seconds and 65–70 seconds after intravenous contrast injection, respectively [3]. In practice, however, many institutions employ bolus-tracking techniques to optimize the timing of the pancreatic phase. The image should be reconstructed using a thin section (preferably 2–3 mm), and multiplanar reconstruction is strongly recommended. At a minimum, coronal images should be routinely reviewed to accurately assess tumor–vessel contact, involvement of adjacent organs, and anatomic variations, all of which are critical for surgical planning.

Magnetic Resonance Imaging

MRI is commonly used as a problem-solving tool, particularly for identifying suspicious PDACs that are not clearly visible on CT and for detecting hepatic metastasis [3,13]. The MRI protocol for PDAC assessment typically includes non-enhanced axial T1- and T2-weighted imaging, diffusion-weighted imaging, T2-weighted MR cholangiopancreatography, and contrast-enhanced T1-weighted imaging during the pancreatic, portal venous, and delayed phases with thin sections (preferably 2–3 mm) [3]. The accuracy of MRI for local tumor staging has been reported to be comparable to that of CT [14,15,16]. However, MRI offers a distinct advantage in detecting hepatic metastases, particularly small lesions that may appear indeterminate or occult on CT [16,17,18]. Meta-analyses have demonstrated higher sensitivity of MRI than that of CT for identifying liver metastases from PDAC while maintaining similar specificity [17]. Hepatobiliary phase imaging and diffusion-weighted imaging further improve lesion conspicuity and are particularly valuable for detecting minute metastasis [16,17,18].

PET/CT

¹⁸F-FDG PET/CT is not routinely recommended for PDAC staging. However, this method may also be useful for detection of occult distant metastasis [19]. Accordingly, PET/CT may be considered in patients with non-metastatic disease on routine CT who exhibit high-risk features, such as markedly elevated carbohydrate antigen (CA) 19-9 levels, large primary tumors, or suspicious lymph node metastasis [3]. Recently, ¹⁸F-FDG PET/CT has gained attention as a tool for predicting tumor aggressiveness by providing information on tumor metabolic activity [7,20]. In addition, ¹⁸F-FDG PET/CT may be useful to assess treatment response in patients who receive NAT.

IMAGING ASSESSMENT OF PANCREATIC CANCER AT INITIAL DIAGNOSIS

In patients with suspected or known PDAC, a comprehensive imaging assessment is required to evaluate distant metastasis, tumor resectability, lymph node status, extra-pancreatic organ involvement, anatomic variations, and other findings that may affect patient management. In 2014, the Society of Abdominal Radiology and the American Pancreatic Association published a consensus statement on radiology reporting for PDAC [4]. This statement standardized the terminology and outlined the essential components for imaging assessment, thereby facilitating systematic radiologic evaluation of patients with PDAC and multidisciplinary communication.

Assessment of Tumor–Vessel Contact

For non-metastatic PDAC, tumor resectability is primarily determined by the extent of tumor involvement of major vessels. The major vessels included the celiac artery (CA), hepatic artery (HA), superior mesenteric artery (SMA), variant arteries, portal vein (PV), superior mesenteric vein (SMV), and inferior vena cava (IVC). The degree of tumor–vessel contact is classified into three categories [3,4]: no contact, defined as a preserved fat plane between the tumor and the vessel; abutment, defined as solid tumor contact involving ≤180° of vessel circumference; and encasement, defined as solid tumor contact involving >180° of vessel circumference. Tumor abutments are further divided into abutments without vessel deformity and abutments with deformity, based on the presence of vessel contour irregularities (Fig. 1). In addition, the presence of vessel stenosis, obstruction, or thrombosis should be explicitly reported. Tumor extension to the first branch of the SMA and SMV should also be documented, although these vessels are not specifically included among the major vessels. Involvement of these vessels, particularly the jejunal branches of the SMA, may affect resectability. Although the assessment of tumor–vessel contact primarily relies on solid tumor contact, the consensus statement also recommends documenting the degree of contact associated with perivascular haziness. Perivascular hazy stranding may be observed after chemotherapy or radiation therapy, reflecting the replacement of solid tumor infiltration by fibrosis, inflammation, or viable tumors, and can also occur in concomitant pancreatitis [4].

Fig. 1. Degree of tumor–vessel contact. Schematic diagrams (upper row) illustrate the classification of the relationship between the tumor (T) and the vessel (V). No contact is characterized by a preserved fat plane between the tumor and the vessel. Abutment is defined as solid tumor contact involving ≤180° of the vessel circumference. Abutment with deformity indicates solid tumor contact (≤180°) accompanied by a visible deformity or indentation of the vessel wall. Encasement refers to solid tumor contact involving >180° of the vessel circumference. Corresponding CT images (lower row) demonstrate examples of pancreatic head cancer (arrowheads) with various degrees of portal or superior mesenteric vein (arrows) involvement.

Special attention should be paid to the identification of variants of the arterial anatomy. Among arterial variants, aberrant right or common HAs arising from the SMA are particularly important in PDACs of the pancreatic head, since these arteries typically traverse the portocaval space and are therefore susceptible to involvement in pancreatic head cancers (Fig. 2). Another important consideration in assessing tumor–vessel contact is that pancreatic head PDACs may extend along extra-pancreatic neural plexuses surrounding the peripancreatic vasculature, a pattern referred to as “perineural tumor spread” [21,22]. On CT and MRI, perineural tumor spread manifests as soft-tissue extension emanating from the primary tumor along the inferior pancreaticoduodenal artery toward the SMA, along the gastroduodenal artery toward the CA, and posteriorly from the pancreatic head [21]. Recognition of this mode of spread is critical, since it may result in non-concentric tumor growth and extensive perivascular tumor extension, despite a small primary tumor (Fig. 3).

Fig. 2. Pancreatic head cancer involving a replaced right hepatic artery arising from the superior mesenteric artery in a 77-year-old man. Axial arterial-phase CT images of two consecutive sections demonstrate a hypodense pancreatic head mass (arrowheads) abutting the replaced right hepatic artery (small arrows) arising from the superior mesenteric artery (large arrow) and traversing the portocaval space. Because the aberrant right or common hepatic arteries arising from the superior mesenteric artery typically course posterior to the pancreatic head, they are particularly susceptible to involvement by pancreatic head cancers.

Fig. 3. Extensive vascular invasion through perineural tumor spread in a 65-year-old man with a small pancreatic head tumor. A: An axial gradient-echo T1-weighted MR image shows a 1.5-cm mass (arrow) with lower signal intensity than the surrounding pancreatic parenchyma. B: An axial contrast-enhanced arterial-phase CT image demonstrates soft-tissue extension (arrowheads) emanating from the pancreatic head tumor (arrow) and extending toward the SMA (arrow). C: A coronal arterial-phase CT image reveals solid tumor infiltration (arrowheads) encasing both the CA and SMA (arrows). Despite the small size of the primary pancreatic head tumor, this extensive perivascular tumor infiltration is presumed to result from perineural tumor spread extending along the inferior pancreaticoduodenal artery toward the SMA and, subsequently, the celiac trunk. SMA = superior mesenteric artery, CA = celiac artery.

The surgical and pathological significance of the tumor–vessel contact categories has been evaluated in previous studies. Nakao et al. [23] classified radiological PV involvement as absent, unilateral narrowing, bilateral narrowing, and stenosis or obstruction. In their cohort of 358 patients with pancreatic head cancer, PV resection was performed in 19.8%, 93.9%, 99.0%, and 100% of patients, respectively. Among patients who underwent PV resection, pathological PV wall invasion was present in 0%, 51%, 74%, and 93% of the patients, respectively [23]. A subsequent study by Tran Cao et al. [24] reported consistent findings. Among patients who underwent upfront resection (UR) for PDAC, vein resection was required in 11.1%, 25.0%, and 100% of patients with no tumor–vein contact, abutment, and encasement, respectively, while pathologic vein invasion was present in 3.7%, 16.1%, and 100% of these patients, respectively. Collectively, these studies demonstrated that the absence of tumor–vein contact indicates the absence of tumor invasion, whereas tumor encasement or tumor contact with vein deformity indicates a high likelihood of pathological vein invasion.

For arterial invasion, Noda et al. [25] evaluated tumor–vessel contact on CT in 128 patients with PDAC and used a combination of surgical and pathological findings as the reference standard. They reported that the presence of any solid tumor contact was the optimal threshold for detecting arterial invasion, yielding a sensitivity of 100%, specificity of 93%, positive predictive value (PPV) of 36%, and negative predictive value (NPV) of 100% [25]. In the same study, applying a threshold of solid tumor contact >180° resulted in higher specificity (97%–99%) and PPV (44%–67%) for diagnosing arterial invasion. Another study including 105 patients demonstrated that tumor-artery contact >180° achieved a sensitivity of 88%, specificity of 94%, PPV of 56%, and NPV of 99% for detecting arterial tumor invasion [26]. These findings suggest that arterial encasement is highly predictive of true arterial invasion, whereas arterial abutment is a highly sensitive, but less specific imaging finding.

Anatomical Tumor Resectability

The NCCN criteria are most widely used for assessing anatomical resectability [3]. According to these criteria, non-metastatic PDAC is categorized as resectable, borderline resectable (BR), or locally advanced (LA) disease. The NCCN criteria defining resectability status at diagnosis of PDAC are presented in Table 1.

Table 1. Resectability status at diagnosis of PDAC according to the NCCN criteria.

Resectability status	Criteria
Resectable	No tumor contact with major arteries No tumor contact with major veins, or venous abutment without deformity
Borderline resectable	Tumor abutment with deformity or encasement of major veins (PV, SMV, and IVC), amenable to reconstruction Pancreas head tumors with abutment of common HA, SMA, or variant HA Pancreas body or tail tumors with abutment of CA
Locally advanced	Extensive, unreconstructible vein involvement Tumor encasement of major arteries Tumor involvement of the CA extending to aorta Tumor involvement of the HA extending to CA or HA bifurcation

Criteria are based on NCCN guidelines after minor modifications.

PDAC = pancreatic ductal adenocarcinoma, NCCN = National Comprehensive Cancer Network, PV = portal vein, SMV = superior mesenteric vein, IVC = inferior vena cava, HA = hepatic artery, SMA = superior mesenteric artery, CA = celiac artery

Resectable

PDACs are considered resectable in the absence of tumor contact with major arteries and either no contact with major veins or venous abutment without deformity.

Borderline Resectable

BR disease includes tumors showing abutment with deformity or encasement of the PV or SMV; tumor abutment to the IVC; pancreatic head tumors with abutment to the common HA, SMA, or variant HA; and body or tail tumors with abutment to the CA (Fig. 4). Venous involvement in BR disease should be limited to a longitudinal extent and should not involve extensive branch vessels, thereby allowing safe venous resection and reconstruction [3,6]. Regarding the limits of reconstructable venous invasion, earlier versions of the NCCN criteria defined tumor involvement of the most proximal jejunal branch as the criterion for LA disease [6]. Subsequent international consensus criteria for BR disease redefined this definition, classifying venous tumor infiltration exceeding the inferior border of the duodenum as LA disease [6]. Short-segment involvement of the common HA is classified as BR disease, provided the involvement does not extend to the CA or HA bifurcation (Figs. 4E, 5). Importantly, tumor involvement at the root of the gastroduodenal artery should be considered equivalent to HA involvement (Fig. 4E), since it may necessitate arterial resection during pancreaticoduodenectomy [6]. BR is sometimes further subclassified into BR-PV, defined as PV or SMV involvement alone, and BR-A, defined as arterial involvement [6]. The rationale for this subclassification is that BR-A tumors are associated with a worse prognosis and a greater risk of incomplete resection than BR-PV tumors [6]. In addition, pancreatic resection combined with arterial resection is associated with increased perioperative morbidity and mortality [27,28], whereas pancreatic resection with concomitant vein resection when performed in high-volume centers demonstrates postoperative morbidity and mortality rates comparable to those of pancreatic resection without vein resection [29,30]. Consequently, synchronous vein resection is commonly performed in high-volume centers to increase the likelihood of achieving margin-negative (R0) resection [30,31], whereas synchronous arterial resection is performed far less frequently [31].

Fig. 4. Representative examples of borderline resectable pancreatic cancer. A: An axial portal venous-phase CT image demonstrates a hypodense pancreatic head mass (T) abutting the superior mesenteric vein (arrow) with associated vessel deformity and stenosis. B: An axial portal venous-phase CT image demonstrates a hypodense pancreatic head mass (T) abutting the inferior vena cava (arrow), with loss of the intervening fat plane (arrowheads). C: An axial portal venous-phase CT image demonstrates a pancreatic head mass (T) abutting the superior mesenteric vein with deformity (small arrow) and the superior mesenteric artery (large arrow). Solid tumor contact with loss of the fat plane between the tumor and the superior mesenteric artery is noted (arrowhead). D: An axial portal venous-phase CT image demonstrates a pancreas body/tail mass (T) abutting the celiac artery (arrow). E: A coronal arterial-phase CT image demonstrates a small pancreatic head mass (arrowheads) abutting the gastroduodenal artery (small arrow) and a short segment of the common hepatic artery (large arrow). During surgery, tumor involvement of the common hepatic artery was confirmed, and short segmental resection with end-to-end anastomosis of the common hepatic artery was performed without clinically significant postoperative morbidity.

Fig. 5. Tumor involvement of the hepatic artery categorized as locally advanced disease in a 66-year-old woman with pancreatic head cancer. A: A coronal portal venous-phase CT image shows a hypoenhancing pancreatic head mass (arrowheads) encasing the gastroduodenal artery (small arrow) from its origin and involving the common hepatic artery (large arrow). B: Axial portal venous-phase CT images of two consecutive sections demonstrate the pancreatic head mass (arrowheads). Due to an anatomic variation with a short proper hepatic artery, the LHA (arrow) and RHA (arrow) arise directly from the common hepatic artery (large arrow). Soft-tissue infiltration is present around the common, left, and right hepatic arteries, indicating that hepatic artery involvement extends beyond hepatic artery bifurcation. Intraoperatively, tumor invasion of the common, left, and right hepatic arteries was identified. For complete tumor resection, pancreaticoduodenectomy with hepatic artery resection and ligation was performed since hepatic arterial reconstruction was not feasible because of long-segment tumor involvement and technical difficulty. C: An axial portal venous-phase CT image obtained five days after surgery demonstrates hepatic infarction involving the left hepatic lobe (arrows). LHA = left hepatic artery, RHA = right hepatic artery.

Locally Advanced

LA disease is defined by extensive involvement of major vessels that precludes complete surgical resection. This category includes tumors with encasement of the SMA or CA, tumor involvement of both the CA and abdominal aorta, and unreconstructible PV-SMV involvement (Figs. 5, 6). Notably, PDACs of the pancreatic body or tail with isolated CA invasion, without involvement of the aorta, gastroduodenal artery, or SMA, were previously classified as BR but are now considered LA disease [3,6]. Nevertheless, in carefully selected patients, curative-intent resection may be feasible using distal pancreatectomy with celiac axis resection (Appleby procedure) (Fig. 7). The success of this procedure requires a patent pancreaticoduodenal arcade, since hepatic arterial perfusion following celiac axis resection depends on collateral arterial flow from the SMA via the pancreaticoduodenal circulation [32].

Fig. 6. Representative examples of locally advanced pancreatic cancer. A, B: An axial arterial-phase CT image (A) and a coronal portal venous-phase maximal intensity projection image (B) demonstrate a hypodense pancreatic head mass (arrowheads) with tumor abutment and contour deformity of the superior mesenteric vein (small arrow) and tumor encasement of the superior mesenteric artery (large arrow). The maximal intensity projection image (B) clearly depicts stenosis of the superior mesenteric vein (small arrow). The tumor is considered to be locally advanced disease because of arterial encasement. C, D: An axial portal venous-phase CT image (C) and a coronal portal venous-phase ray-sum image (D) demonstrate a hypodense pancreatic head mass (arrowheads) with tumor abutment and contour deformity of the superior mesenteric vein (small arrows) and tumor abutment of the superior mesenteric artery (large arrows). The coronal ray-sum image (D) shows long-segment tumor involvement of the superior mesenteric vein and its tributaries, resulting in irregular stenosis. Dashed lines indicate the estimated extent of vein involvement. This unreconstructible long-segment vein involvement is considered locally advanced disease.

Fig. 7. Distal pancreatectomy with celiac artery resection (Appleby procedure) performed for curative-intent resection of pancreatic body cancer with celiac artery and common hepatic artery invasion in a 65-year-old woman. A, B: Preoperative axial arterial-phase CT images demonstrate a hypodense pancreatic body tumor (T) encasing the common hepatic artery (small arrows), with tumor extension to the celiac artery (large arrow). C: A preoperative axial arterial-phase CT image at the level of the pancreatic head shows a patent superior mesenteric artery (large arrow) and gastroduodenal artery (small arrow) without tumor involvement, indicating the feasibility of distal pancreatectomy with celiac artery resection. D, E: Axial portal venous-phase CT images obtained five days after surgery demonstrate metallic clips ligating the celiac artery at its origin (large arrow), the remaining pancreatic head (arrowheads), and a postoperative fluid collection (F) along the resection margin. No evidence of hepatic arterial insufficiency was observed postoperatively.

Lymph Node Metastasis

Imaging features suggestive of metastatic lymph nodes include a short-axis diameter of >1 cm, abnormal round morphology, heterogeneity, or central necrosis. Only lymph nodes within the lymphatic drainage pathways of the primary tumor and surgical resection boundaries are considered regional lymph nodes, and lymph node metastases beyond the surgical fields are classified as distant metastases [4]. Notably, CT and MRI have limited sensitivity for detecting lymph node metastasis. In a previous study, the sensitivity and specificity of CT for identifying regional lymph node metastasis were 23.5% and 89.9%, respectively [20]. PET/CT may complement CT or MRI in the detection of metastatic lymph nodes. However, as a standalone modality, PET/CT also has limited sensitivity [20], and even the combined use of contrast-enhanced CT and PET/CT yields low sensitivity, ranging from 22.4% to 32.0% [20,33]. Despite this limited sensitivity, the presence of suspicious regional lymph node metastases on CT or PET/CT has consistently been reported as an adverse prognostic factor following curative resection in patients with PDAC [20,33,34].

Adjacent Organ Invasion

According to the current guidelines, technically resectable extra-pancreatic organ invasion is not considered a determinant of PDAC resectability [3]. Nevertheless, invasion of extra-pancreatic organs necessitates additional resection and modification of the surgical approach. Extra-pancreatic organ invasion is of particular concern in PDACs of the pancreatic tail, since these tumors are less likely to involve major vessels owing to the relative distance of major vessels from the pancreatic tail, but more frequently invade adjacent organs such as the spleen, stomach, colon, jejunum, kidney, and adrenal gland (Fig. 8). A previous study identified the presence of solid tumor contact with adjacent organs, with or without organ contour deformity, as the optimal CT criterion for diagnosing pathological extra-pancreatic organ invasion, yielding a sensitivity of 91.2% and specificity of 96.0% [35]. This study also demonstrated that radiological or pathological extra-pancreatic organ invasion, when technically resectable with additional organ resection, did not adversely affect resection margin status, recurrence, or survival after curative resection.

Fig. 8. Pancreatic tail cancer with central tumor necrosis and invasion of the stomach and jejunum in a 65-year-old man. A: An axial portal venous-phase CT image demonstrates a pancreatic tail mass (arrows) with upstream pancreatic duct dilatation (arrowheads). The tumor had a maximal dimension of 4.5 cm in maximal dimension and showed central necrosis appearing as a central non-enhancing area within the tumor (*). The serum CA 19-9 level was markedly elevated at 2,216 U/mL. B: A coronal portal venous-phase CT image demonstrates the pancreatic tail mass (T) with suspected invasion of the stomach (S) and proximal jejunum (J). Peripancreatic tumor infiltration abuts the stomach with loss of the intervening fat plane (small arrow) and also abuts the proximal jejunum with focal jejunal wall deformity (large arrow). Distal pancreatectomy and splenectomy were performed, and surgical pathology revealed undifferentiated carcinoma of sarcomatoid type. C: An axial portal venous-phase CT image obtained two months after surgery demonstrates hepatic metastases (arrows).

Stenosis of the Celiac Artery or Superior Mesenteric Artery

The recognition of CA or SMA stenosis is important in patients undergoing pancreaticoduodenectomy. In the presence of CA or SMA stenosis, arterial perfusion of the hepatic, gastric, and mesenteric circulations may depend on collateral circulation through the pancreaticoduodenal arcade arising from the SMA or CA, respectively. Since the pancreaticoduodenal arcade is typically resected during pancreaticoduodenectomy, unrecognized CA or SMA stenosis may lead to hepatic, gastric, or mesenteric ischemia after surgery (Fig. 9) [36]. When CA or SMA stenosis is identified preoperatively, adequate arterial perfusion can be maintained by preoperative endovascular intervention or intraoperative dissection of the median arcuate ligament to relieve celiac axis compression [37].

Fig. 9. Nonocclusive mesenteric ischemia after pancreaticoduodenectomy for IPMN in a 74-year-old woman. A: An oblique coronal T2-weighted thick-slab MR cholangiopancreatography image demonstrates a 4-cm cystic mass (arrows) with intra-cystic filling defects (arrowheads). B: A coronal contrast-enhanced portal venous-phase T1-weighted image shows the same cystic mass (arrows) with enhancing mural nodules (arrowheads), consistent with IPMN with high-risk stigmata. C: A sagittal arterial-phase maximal intensity projection image demonstrates severe stenosis at the orifice of the superior mesenteric artery (small arrow). The celiac artery (large arrow) and gastroduodenal artery (arrowhead) are enlarged, likely functioning as a collateral pathway supplying superior mesenteric artery flow. The expected flow direction is indicated by a dotted line. D, E: Coronal portal venous-phase CT images obtained five days after pylorus-preserving pancreaticoduodenectomy demonstrate ileus of the small bowel and colon, with markedly decreased enhancement of the ileum and ascending colon (*) in comparison with the relatively preserved enhancement of the proximal jejunum (arrowheads). The superior mesenteric artery (arrow) remains patent (C). These findings are suggestive of nonocclusive mesenteric ischemia, likely due to mesenteric hypoperfusion following loss of collateral perfusion in the setting of superior mesenteric artery stenosis. IPMN = intraductal mucinous papillary neoplasm.

BIOLOGICAL AND CONDITIONAL FACTORS

Achieving margin-negative resection has traditionally been the primary goal of curative-intent surgery for PDAC, and surgical candidacy has been determined on the basis of anatomical resectability. However, recurrence after curative-intent pancreatic resection remains common even in patients who have achieved R0 resection. In a previous study, the 1-year recurrence-free survival rate was only 45% in patients who underwent UR for resectable PDAC and 49% in those who achieved R0 resection [34]. These high postoperative recurrence rates underscore the limitations of treatment decision-making based solely on anatomical resectability and have led to the emergence of an expanded concept of resectability that incorporates not only anatomical resectability but also biological and conditional factors.

The concept of biological resectability emphasizes the consideration of high-risk features suggestive of aggressive tumor biology, in addition to the anatomic tumor extent, when selecting an optimal treatment strategy [6,7,20,38,39]. This also suggests that patients with anatomically resectable disease who exhibit high-risk features may be more appropriately classified as having BR disease [6,7]. Among the factors reflecting tumor biology, the serum CA 19-9 level is the most widely used [38]. Other potential biological factors include radiologic tumor phenotypes, presence of suspicious regional lymph nodes, and maximum standardized uptake value (SUVmax) of the primary tumor on ¹⁸F-FDG PET/CT. In addition, host-related conditional factors should be considered in therapeutic decision-making, since they are associated with the risk of postoperative morbidity and mortality. Among them, the Eastern Cooperative Oncology Group (ECOG) performance status is the most commonly used [38]. Other conditional factors include the modified Glasgow prognostic score, neutrophil-to-lymphocyte ratio, and Onodera prognostic nutritional index [6,40]. Accordingly, an A-B-C approach to clinical staging that incorporates anatomical resectability, biologic factors, and patient condition has been proposed [5,38]. In a cohort study of patients with localized PDAC initially treated with modified FOLFIINOX, staging based on the A-B-C approach, combining anatomical resectability, CA 19-9 level (>500 U/mL), and ECOG performance status provided better survival stratification than anatomical resectability alone [38].

Imaging Phenotype

Multiple previous studies have suggested that the imaging phenotypes of PDAC may predict tumor aggressiveness and patient prognosis. Koay et al. [41] reported a quantitative delta metric on CT that represents the degree of enhancement difference between the tumor and surrounding pancreatic parenchyma as a potential prognostic biomarker for PDAC. They demonstrated that conspicuous tumors with high delta values were associated with lower stromal content, higher prevalence of common pathway mutations, increased likelihood of early metastasis, and shorter survival than inconspicuous tumors with low delta values [41]. Consistent with these findings, other studies have reported that tumors exhibiting necrosis on CT [34,42] and rim enhancement on MRI [43] are associated with poor prognosis and aggressive pathological features, including poor differentiation and lymph node metastasis (Fig. 8). In contrast, isoattenuation on CT [13] and a larger extracellular volume, reflected by increased contrast enhancement on equilibrium-phase CT [44], have been suggested as favorable prognostic features. In addition to the morphological characteristics of the primary tumor, the presence of suspicious metastatic lymph nodes has been identified as a high-risk factor for early recurrence or death.

Serum CA 19-9

CA 19-9 is a tumor-associated antigen widely used to assess PDAC. The pretreatment serum CA 19-9 level has been shown to be correlated with tumor stage [45] and survival outcomes following surgical resection [20,39,45] and chemotherapy [38,46] in patients with PDAC. Importantly, the pretreatment serum CA 19-9 level enables prognostic stratification of patients with resectable PDAC undergoing curative-intent surgery. In a study by Ushida et al. [39], patients with PDAC were stratified according to anatomical resectability and a serum CA 19-9 threshold of 500 U/mL. The authors demonstrated that both anatomical resectability and serum CA 19-9 category were independent prognostic factors for overall survival. Notably, patients with resectable PDAC and high CA 19-9 levels exhibited worse survival than those with resectable PDAC and low CA 19-9 levels but comparable survival to patients with BR disease. These findings suggest that markedly elevated CA 19-9 levels in anatomically resectable PDAC are a high-risk feature and may serve as a potential indicator for NAT.

Nevertheless, assessments based on CA 19-9 levels have several limitations. First, elevated CA 19-9 levels are not specific to PDAC and may occur in benign conditions such as biliary obstruction or cholangitis. Accordingly, in patients with biliary obstruction, serum CA 19-9 levels may not reliably reflect tumor burden or prognosis. For a valid interpretation, the CA 19-9 level should be measured after adequate biliary drainage, with a serum bilirubin level <2.0 mg/dL commonly used as the threshold for reliability [20,45]. Second, individuals who are Lewis antigen–negative, who account for approximately 5%–10% of the East Asian population, are unable to synthesize CA 19-9 [47]. Consequently, a CA 19-9–based prognostic assessment is generally not applicable in patients with baseline CA 19-9 levels below the low-normal range. Third, a universally established cutoff serum CA 19-9 level indicative of poor prognosis has not been established. However, a threshold CA 19-9 level >500 U/mL has been commonly used in previous studies [7,38,39]. Finally, the absence of a universally accepted international reference standard or calibrator for CA 19-9 has led to inter-assay variability, precluding interchangeability of results from different assay platforms [48].

¹⁸F-FDG PET/CT Findings

Several studies have suggested that the SUVmax of the primary tumor, measured using ¹⁸F-FDG PET/CT, may serve as a prognostic biomarker for PDAC. Moon et al. [49] demonstrated that a pretreatment tumor SUVmax >5.5 was an independent predictor for worse survival after surgical resection in patients with resectable PDAC. Similarly, Barnes et al. [50] reported that a pretreatment tumor SUVmax >7.5 was associated with worse survival outcomes following NAT in patients with non-metastatic PDAC. However, evidence supporting the use of the tumor SUVmax as a prognostic marker remains limited, and additional well-designed large-scale studies are warranted to clarify its clinical utility.

Multivariable Risk Models for Resectable Disease

Pretreatment risk stratification is clinically relevant in patients with resectable PDAC, for whom both UR and NAT are valid treatment options. To improve the prediction performance, multivariable prediction models have been developed to predict the risk of early recurrence or death following UR in patients with resectable PDAC [20,34,51,52,53]. The representative preoperative models are shown in Table 2.

Table 2. Multivariable risk prediction models for upfront resection in patients with resectable pancreatic cancer.

Study	Primary outcomes	Predictors	Key findings
Nakamura et al., 2018 [52]	OS	Tumor location (head), tumor size on CT (>2 cm) serum CA 19-9 level (>100 U/mL)	Patient stratification according to OS
Kim et al., 2020 [34]	Recurrence free survival	CT variables of tumor size, tumor density in portal venous phase, tumor necrosis, peripancreatic infiltration, suspicious metastatic lymph nodes	C-index of 0.68 in the temporal validation
Crippa et al., 2024 [51]	Futile pancreatectomy*	Tumor size on CT, serum CA 19-9 level, American Society of Anesthesiology class	C-index of 0.65 in the external validation
Jeong et al., 2024 [20]	Recurrence free survival	Tumor size on CT, tumor SUVmax on PET/CT, serum CA 19-9, suspicious regional lymph node on CT or PET/CT, suspicious metastasis on PET/CT	C-index of 0.61 in the temporal validation
Schouten et al., 2024 [53]	OS	Anatomic resectability, serum CA 19-9 level (≥500 U/mL), and ECOG performance status (≥2)	Patient stratification according to OS

^*Defined as the occurrence of death or recurrence within 6 months of surgery.

CA = carbohydrate antigen, OS = overall survival, SUVmax = maximum standardized uptake value, ECOG = Eastern Cooperative Oncology Group

TREATMENT STRATEGY BASED ON ANATOMICAL RESECTABILITY

Anatomical tumor resectability status stratifies the likelihood of margin-negative R0 resection. Data from a cohort of 616 patients with PDAC [31] demonstrated that the R0 resection rates in patients who underwent UR were 73%, 55%, and 16% for resectable, BR, and LA disease, respectively; however, when all patients were considered, including those who did not undergo surgical resection, the corresponding R0 resection rates were 71%, 45%, and 4%, respectively. Given the obvious differences in surgical outcomes in relation to the resectability status, treatment strategies for PDAC have been determined primarily by tumor resectability status, with consideration of patient performance status and biological factors [3,5].

Borderline Resectable and Locally Advanced Disease

The introduction of multiagent chemotherapy regimens has led to the increasing adoption of NAT as the initial treatment for non-metastatic PDAC. For BR PDAC, prior studies have demonstrated that NAT is associated with higher R0 resection rates and improved survival in comparison with upfront surgery [54,55]. Accordingly, NAT has become the standard initial treatment strategy for BR PDAC, with FOLFIRINOX or gemcitabine combined with albumin-bound paclitaxel being the preferred regimens [3]. Patients who do not experience disease progression during NAT should be considered for conversion surgery following restaging, which is typically determined through a multidisciplinary discussion. For LA PDAC, systemic chemotherapy with or without stereotactic radiotherapy is recommended. Patients who demonstrate a favorable treatment response may subsequently undergo surgical resection.

Resectable Disease

The treatment paradigm for resectable PDAC has evolved over time. Traditionally, UR followed by adjuvant chemotherapy has been considered the standard of care. Recently, however, NAT has been increasingly adopted for patients with resectable PDAC. Earlier guidelines recommended UR as the primary therapy, allowing NAT selectively for patients with high-risk features such as a large tumor size, suspicious metastatic lymph nodes, and markedly elevated CA 19-9 levels. In contrast, guidelines updated since 2024 endorse NAT as an initial treatment strategy for resectable PDAC while recommending UR only for patients without high-risk features [3].

Unresolved Issues

Despite the increasing adoption of NAT as the initial treatment for PDAC in clinical guidelines [3,8,11], its implementation in real-world practice remains variable across institutions, and UR continues to be performed for resectable and even BR PDAC [31]. Although multiple factors may influence the clinical implementation of NAT, one important reason in Korea may be that multiagent NAT for potentially resectable PDAC is not fully reimbursed by the National Health Insurance system.

Other than BR PDAC, for which NAT has consistently shown favorable surgical and survival outcomes in comparison with UR [54,56,57], the survival benefit of NAT over UR in resectable PDAC remains controversial. Several retrospective studies have suggested improved survival with NAT in comparison with UR [58,59,60]. However, these studies included only patients who underwent NAT followed by resection and may have been subject to selection bias due to the exclusion of patients who did not undergo surgery. Moreover, the evidence from randomized controlled trials has been inconsistent. Two trials that enrolled patients with both resectable and BR PDAC demonstrated superior survival rates with NAT [55,61]. In contrast, a recently completed randomized trial that exclusively enrolled patients with resectable PDAC failed to demonstrate a survival benefit for NAT over UR [62]. Thus, current evidence does not fully support the preference for NAT as an initial treatment strategy for resectable PDAC, as recommended by the current NCCN guidelines [3]. Furthermore, multiple studies have identified high-risk features associated with recurrence or death after UR in patients with resectable PDAC. However, conclusive evidence showing that NAT improves the survival outcomes in these high-risk patients is lacking. Further studies are needed to identify the optimal treatment strategy and enable personalized treatment selection through the identification of robust biological and imaging biomarkers.

RESTAGING AFTER NEOADJUVANT THERAPY FOR BORDERLINE RESECTABLE OR LOCALLY ADVANCED TUMORS

Curative-intent pancreatic resection improves survival outcomes in patients with BR or LA PDAC treated with NAT [63,64]. Therefore, the feasibility of conversion surgery should be evaluated in patients who do not progress during NAT [3]. However, restaging using CT and MRI after NAT is more challenging than initial staging and is associated with lower accuracy in predicting the likelihood of R0 resection [65]. Previous studies have suggested that CT and MRI tend to overestimate vascular invasion after NAT, likely because imaging cannot reliably distinguish viable tumors from post-treatment fibrosis [65,66,67]. Consequently, R0 resection may be achieved after NAT in patients with BR or even LA disease. Ferrone et al. [66] reported that R0 resection could be achieved in 92% of patients who initially had BR or LA disease, although radiological downstaging to resectable disease occurred in only 35% of the patients. However, these findings should be interpreted with caution, because many previous studies reporting R0 resection rates after NAT included only patients who underwent both NAT and subsequent surgery, potentially overestimating the R0 resection rates. In a recent prospective study using high-resolution, low-tube-voltage CT imaging, the R0 resection rates for resectable, BR, and LA disease were 88.7%, 52.4%, and 0%, respectively, among patients who underwent UR and 90.9%, 76.7%, and 25.0%, respectively, among those who underwent surgery after NAT [67]. Taken together, these results suggest that higher R0 resection rates may be achieved for BR and LA disease after NAT than those observed at initial staging.

Therefore, the interpretation of anatomical resectability at initial diagnosis may not be directly applied to restaging after NAT, and some patients with BR or LA disease at restaging may benefit from curative-intent resection. Various approaches that facilitate the selection of candidates for conversion surgery after NAT have been reported. Changes in the radiologic tumor extent may have an incremental value for anatomical resectability in predicting R0 resection at restaging. Regression of tumor–vessel contact after NAT, even with residual contact, has been reported to predict R0 resection with high specificity [68]. Objective tumor response in comparison with baseline has also been reported to be associated with a higher likelihood of R0 resection (Fig. 10). In addition to the prognostic value of pretreatment CA 19-9 levels, post-treatment CA 19-9 levels and their dynamic changes may have clinical implications. Previous studies have shown that normalization or reduction of CA 19-9 levels by more than 50% from their baseline value is associated with higher R0 resection rates and improved survival [69,70,71,72]. A recent study demonstrated that the pattern of changes in CA 19-9 levels is also important. Constant or bidirectional reductions to normal levels were associated with the best survival outcomes, followed by persistently normal CA 19-9 levels or reductions without normalization, whereas persistently increasing CA 19-9 levels without normalization were associated with the worst survival outcomes [73]. Furthermore, tumor metabolic activity on post-treatment PET/CT may help select candidates for conversion surgery. In a study by Yoo et al. [74], tumor–vessel contact without FDG uptake was more likely to be associated with R0 resection than residual contact with FDG uptake (Fig. 10).

Fig. 10. Margin-negative tumor resection following neoadjuvant chemotherapy in a 75-year-old man with borderline resectable pancreatic cancer. A, B: Axial portal venous-phase CT images obtained at initial staging demonstrate an ill-defined hypodense pancreatic head mass (arrowheads) with tumor abutment associated with contour deformity of the superior mesenteric vein (small arrows) and tumor abutment of the superior mesenteric artery (large arrows). C: A ¹⁸F-FDG PET/CT image demonstrates a hypermetabolic tumor (arrowhead) with a maximum standardized uptake value of 11.4. The baseline serum CA 19-9 level was 51 U/mL. The tumor was considered borderline resectable. FOLFIRINOX therapy was administered as neoadjuvant therapy. D, E: Axial portal venous-phase CT images obtained after six cycles of FOLFIRINOX demonstrate a decrease in tumor size with an indistinct mass in the pancreatic head. Tumor contact with the superior mesenteric vein (small arrows) improved with no residual vein contour deformity (D). However, residual tumor infiltration (arrowheads) persisted along the medial aspect of the pancreatic head abutting the superior mesenteric artery (large arrows) (E). F: A ¹⁸F-FDG PET/CT image obtained after completion of FOLFIRINOX therapy demonstrates a complete metabolic response with no hypermetabolic tumor uptake (arrowhead). The post-treatment CA 19-9 level normalized to 18.4 U/mL. The patient subsequently underwent pancreaticoduodenectomy, achieving a margin-negative resection, and has remained recurrence-free for 8 years following surgery. FDG = fluorodeoxyglucose.

These findings suggest that changes in the tumor extent, tumor–vessel contact, CA 19-9 level, and metabolic activity may refine the interpretation of residual tumor–vessel contact after NAT. In patients with favorable radiologic, CA 19-9, or metabolic responses, residual tumor–vessel contact may not represent true vascular invasion, and such patients may be considered for conversion surgery. However, well-established criteria for selecting patients who may benefit from surgery after NAT are lacking. Further large-scale studies are required to establish robust criteria for the selection of patients for conversion surgery after NAT.

CONCLUSIONS

Accurate imaging evaluations are essential for the diagnosis and management of PDAC, particularly for tumor staging and the assessment of tumor resectability. The paradigm for evaluating tumor resectability is evolving from a purely anatomical framework to a comprehensive approach that incorporates tumor biology and patient-related conditional factors. Combined assessment of anatomical resectability with biological and conditional factors can provide improved prognostic stratification in comparison with reliance on anatomical criteria alone. For patients with resectable PDAC, both UR and NAT represent viable treatment strategies. However, the current evidence remains insufficient to clearly favor one approach over the other. Future research should focus on the identification and validation of robust biomarkers that enable personalized treatment selection and the optimization of outcomes in patients with resectable PDAC. Radiological restaging after NAT is more challenging than initial staging and may overestimate vascular invasion. In this context, consideration of changes in tumor–vessel contact, objective tumor response, and biological and metabolic factors may help address the limitations of resectability assessment after NAT. Robust criteria are required to guide the selection of the optimal candidates for conversion surgery after NAT.