OBJECTIVES: To determine the sensitivity and specificity of elevated serum IgG4 level for the diagnosis of autoimmune pancreatitis (AIP) and its ability to distinguish AIP from pancreatic cancer, its main differential diagnosis. METHODS: We measured serum IgG4 levels (normal 8-140 mg/dL) in 510 patients including 45 with AIP, 135 with pancreatic cancer, 62 with no pancreatic disease, and 268 with other pancreatic diseases. RESULTS: Sensitivity, specificity, and positive predictive values for elevated serum IgG4 (>140 mg/dL) for diagnosis of AIP were 76%, 93%, and 36%, respectively, and 53%, 99%, and 75%, respectively, for IgG4 of >280 mg/dL. Among subjects with elevated IgG4, non-AIP subjects (N = 32) differed from AIP subjects (N = 34) in that they were more likely to be female (45%vs 9%, P < 0.001), less likely to have serum IgG4 >280 mg/dL (13%vs 71%, P < 0.001), or elevation of total IgG (16%vs 56%, P < 0.001). Serum IgG4 levels were elevated in 13/135 (10%) pancreatic cancer patients; however, only 1% had IgG4 levels >280 mg/dL compared with 53% of AIP. Compared with AIP, pancreatic cancer patients were more likely to have CA19-9 levels of >100 U/mL (71%vs 9%, P < 0.001). CONCLUSION: Elevated serum IgG4 levels are characteristic of AIP. However, mild (<2-fold) elevations in serum IgG4 are seen in up to 10% of subjects without AIP including pancreatic cancer and cannot be used alone to distinguish AIP from pancreatic cancer. Because AIP is uncommon, IgG4 elevations in patients with low pretest probability of having AIP are likely to represent false positives.
OBJECTIVES: To determine the sensitivity and specificity of elevated serum IgG4 level for the diagnosis of autoimmune pancreatitis (AIP) and its ability to distinguish AIP from pancreatic cancer, its main differential diagnosis. METHODS: We measured serum IgG4 levels (normal 8-140 mg/dL) in 510 patients including 45 with AIP, 135 with pancreatic cancer, 62 with no pancreatic disease, and 268 with other pancreatic diseases. RESULTS: Sensitivity, specificity, and positive predictive values for elevated serum IgG4 (>140 mg/dL) for diagnosis of AIP were 76%, 93%, and 36%, respectively, and 53%, 99%, and 75%, respectively, for IgG4 of >280 mg/dL. Among subjects with elevated IgG4, non-AIP subjects (N = 32) differed from AIP subjects (N = 34) in that they were more likely to be female (45%vs 9%, P < 0.001), less likely to have serum IgG4 >280 mg/dL (13%vs 71%, P < 0.001), or elevation of total IgG (16%vs 56%, P < 0.001). Serum IgG4 levels were elevated in 13/135 (10%) pancreatic cancerpatients; however, only 1% had IgG4 levels >280 mg/dL compared with 53% of AIP. Compared with AIP, pancreatic cancerpatients were more likely to have CA19-9 levels of >100 U/mL (71%vs 9%, P < 0.001). CONCLUSION: Elevated serum IgG4 levels are characteristic of AIP. However, mild (<2-fold) elevations in serum IgG4 are seen in up to 10% of subjects without AIP including pancreatic cancer and cannot be used alone to distinguish AIP from pancreatic cancer. Because AIP is uncommon, IgG4 elevations in patients with low pretest probability of having AIP are likely to represent false positives.