Pancreatic ductal adenocarcinoma (PDAC) is staged with two complementary systems that should be kept separate. The AJCC 8th edition TNM classification (2017) is an anatomic and pathologic system used mainly for prognosis and standardized reporting. In the 8th edition, T category is defined chiefly by tumor size, N category by the number of involved nodes, and T4 by arterial involvement regardless of size.^[1] The clinical resectability classification of the NCCN has four tiers: resectable, borderline resectable, locally advanced, and metastatic. It is based on tumor contact with the peripancreatic arteries and veins on pancreas-protocol imaging, and it determines initial management.^[2][3] Biologic markers (CA 19-9, nodal disease) and conditional factors (performance status) are increasingly added to anatomic resectability.^[4] Staging should be done with dedicated pancreas-protocol cross-sectional imaging and finalized by multidisciplinary consensus at a high-volume center.^[2] Pancreatic neuroendocrine tumors have a separate AJCC staging system and are not covered here.^[5]

• The 7th edition MX category is no longer used; clinical M is assigned only as M0 or M1.^[1]
• The most common sites of distant spread are the liver, peritoneum, lung, and nonregional lymph nodes.^[6]

• T category is size-based. The 7th edition criterion of “extension beyond the pancreas” was subjective and poorly reproducible, and it placed most resected tumors in T3. Multi-institutional validation supported the new size and nodal cut-points and gave a more even distribution across stages without loss of prognostic accuracy.^[7][8]
• T1 is subdivided (T1a/b/c) to capture small, potentially screen-detected cancers.^[1]
• N category is stratified by node count (N1 = 1–3, N2 = ≥4), reflecting the prognostic value of nodal burden. Stage III now includes N2 disease of any size.^[7]
• T4 is defined by arterial involvement (celiac axis, SMA, CHA) and is no longer labeled “unresectable.” Some arterially involved tumors become resectable after neoadjuvant therapy at expert centers.^[1]
• AJCC stage is now explicitly a prognostic and reporting tool. Treatment selection relies on the resectability classification.^[1]

Resectability determines initial therapy. It requires pancreas-protocol imaging and multidisciplinary review. A standardized radiology reporting template is preferred because inter-observer variability is high.^[2][9]

• A practical rule: arterial contact <180° suggests borderline resectable disease, and >180° suggests locally advanced disease. Unreconstructable venous involvement defines locally advanced disease.^[3]
• Locally advanced means anatomically unresectable at presentation. It does not mean incurable, because a subset of patients becomes resectable after induction therapy.^[1]
• Competing anatomic definitions exist (NCCN, MD Anderson, AHPBA/SSAT/SSO, Alliance, Dutch Pancreatic Cancer Group). They differ mainly in venous thresholds.^[10]

International consensus criteria extend anatomic resectability with two further dimensions. These are increasingly used in risk stratification and trial design.^[4]

• Biologically borderline disease (CA 19-9 ≥500 U/mL) is associated with worse survival even when the tumor is anatomically resectable.^[9]

• Pancreas-protocol multiphase contrast-enhanced CT is the preferred modality for detection, vascular mapping, and staging. It uses thin sections, pancreatic parenchymal and portal venous phases, and multiplanar reconstruction. Diagnostic accuracy is about 89%.^[2][9]
  • Imaging should be obtained before biliary stent placement when feasible, because stents cause artifact that degrades vascular assessment. Imaging should also be within 4 weeks of planned surgery.^[2]
  • Chest and pelvic CT complete staging.^[2]
• MRI/MRCP is a problem-solving adjunct. It is most useful for CT-indeterminate liver lesions, isoattenuating masses, and iodinated contrast allergy, and it is more sensitive than CT for small hepatic metastases.^[2][11]
• EUS is not a routine staging tool but is the preferred route for tissue acquisition. EUS-guided core biopsy is preferred over FNA and over percutaneous biopsy. Biopsy proof is not required before upfront resection when suspicion is high and should not delay it. Tissue is required before any chemotherapy or radiation.^[2]
• PET/CT or PET/MRI does not replace contrast-enhanced CT. Consider it for high-risk features (markedly elevated CA 19-9, large primary tumor, bulky regional nodes, equivocal imaging) to detect occult metastases, and for response assessment around neoadjuvant therapy.^[2]
• Staging laparoscopy is selective rather than routine. It is reserved for high-risk features, body/tail tumors (which carry a higher risk of occult peritoneal spread), ascites, and borderline resectable or locally advanced disease. With modern imaging, its yield for occult metastasis is about 15%.^[2][9]
• CA 19-9 should be measured after biliary decompression and normalization of bilirubin, because cholestasis falsely elevates it. It is uninformative in Lewis antigen–negative non-secretors (about 8–10%), in whom CEA and CA-125 may be used instead.^[2][6]
• Transabdominal ultrasound has no staging role. ERCP is used for biliary decompression and occasional brush cytology, not for staging.^[2]

1. Stage anatomy with pancreas-protocol CT (or MRI). Assign resectability by multidisciplinary consensus before any treatment.^[12][2]
2. Select initial therapy with the four-tier resectability classification, not the AJCC stage. Resectable disease may proceed to upfront surgery or neoadjuvant therapy. Borderline resectable and locally advanced disease receive neoadjuvant or induction systemic therapy first.^[2][13]
3. Interpret CA 19-9 after biliary drainage and use it as a biologic staging variable. A value ≥500 U/mL flags high-risk, biologically borderline disease.^[4][2]
4. Treat positive peritoneal cytology as M1 disease, regardless of anatomic resectability.^[2]
5. Refer patients to a high-volume center (≥15–20 pancreatectomies per year) for staging and management.^[2]
6. Do not use post-neoadjuvant imaging alone to deny surgery. Cross-sectional imaging poorly distinguishes viable tumor from treatment-related fibrosis.^[2]

• More than half of patients (about 53–57%) present with stage IV disease. Most clinically localized tumors harbor occult micrometastases, which drives the high recurrence rate after resection.^[9][13]
• Five-year relative survival is about 12% across all stages and depends strongly on stage. It is about 44% for localized disease, 16% for regional disease, and 3% for distant disease. Only about 15% of cases are diagnosed at the localized stage.^[14][15][16]
• Overall 5-year survival rose from about 6% (2004) to about 12% (2015), driven mostly by better outcomes in localized disease.^[14]
• After neoadjuvant therapy and resection, actual 5-year survival is about 34% overall: about 42% for baseline resectable, 29% for borderline resectable, and 18% for locally advanced disease.^[17]

• Competing resectability definitions: In a NORPACT-2 subgroup analysis, Dutch Pancreatic Cancer Group criteria predicted both resectability and survival. NCCN criteria predicted only resectability.^[18]
• Prognostic performance of the 8th edition: One single-center validation found that the 8th edition better stratified early recurrence and disease-free survival, while the 7th edition kept stronger overall-survival discrimination.^[19] The size-based pT category improved overall-survival stratification in several cohorts. The added prognostic value of the N1/N2 split has been inconsistent.^[8]
• Heterogeneity of stage III: N2 and T4 tumors are both stage III but have different treatment pathways and outcomes. Subdivision into IIIA (T1–3 N2 M0) and IIIB (T4 any N M0) has been proposed.^[20][21]
• Restaging after neoadjuvant therapy is unreliable, so radiographic resectability after treatment should not be the sole determinant of surgery.^[2]
• CA 19-9 threshold for staging laparoscopy: Reported thresholds range from ≥130 U/mL to ≥824 U/mL, and about 150 U/mL is most often cited.^[22][23]

• AJCC stage answers “what is the prognosis?” Resectability answers “what do we do first?”^[1]
• Using routine, non-protocol CT misses subcentimeter metastases and underestimates vascular contact.^[2]
• Equating T4 or locally advanced disease with incurable disease is an error, because some patients respond to induction therapy and undergo resection.^[10]
• Applying 7th edition T and N definitions, which persist in older references and templates, misclassifies tumors.^[1]
• Over-ordering staging laparoscopy or PET in low-risk, clearly resectable disease yields little.^[2]