DIFFERENTIATING PANCREATIC CYSTS THROUGH FLUID GLUCOSE UTILIZING A VALIDATED LABORATORY GLUCOSE TEST IN A PROSPECTIVE COHORT STUDY

Background: Cyst fluid glucose (CFG) is increasingly being used in the differentiation of pancreatic cystic lesions (PCLs). Most prior studies lacked a uniform protocol for specimen acquisition and used point-of-care glucometers that may not have been validated for use with pancreatic cyst fluid (PCF). We sought to evaluate CFG for the differentiation of PCLs measured using a glucose assay on an automated and validated chemistry platform.

Methods: Glucose measured by a hexokinase photometric assay on the automated Beckman Coulter AU5800 (Beckman Coulter, Brea, Ca) was validated for PCF. Reproducibility, linearity, analytical sensitivity, analytical specificity, and stability were all evaluated using residual PCFs. Participants with available CFG levels and a reference diagnosis were selected from a prospective study cohort (NCT 03492151) evaluating PCLs by EUS-guided confocal laser endomicroscopy (EUS-nCLE). PCLs with CFG values < 50 mg/dL, CEA concentration of ≥ 192 ng/mL, cytological detection of mucin, EUS-nCLE visualization of papillae or epithelial bands were classified as mucinous. Reference diagnosis included PCL histopathology or the presence of KRAS or GNAS/BRAF or VHL mutations in fresh PCF with next-generation sequencing (NGS) analysis. The diagnostic accuracies were compared.

Results: Two samples of PCF at glucose concentrations of 72 mg/dL and 30 mg/dL demonstrated coefficient of variations (CV)s of 1.3% and 2.6%, respectively measured in 20 replicates over 5 days. CFGs were linear from 7-110 mg/dL and showed no significant matrix effects. A CV of 4.8% was measured at the limit of quantitation of 10 mg/dL.

79 participants (52% female, mean age 67.2 ± 14.8 years) with mean cyst size of 4.0 ± 1.7 cm met the selection criteria. Reference diagnosis included histopathology (n=53) and NGS detection of KRAS or GNAS/BRAF mutations (n=26).

The highest diagnostic accuracy was achieved by EUS-nCLE, with a sensitivity, specificity, and accuracy of 98%, 97%, and 98%, respectively. CFG had the next best diagnostic accuracy with sensitivity, specificity, and accuracy of 92%, 84%, and 87%, respectively. The cyst fluid CEA and/or cytology showed a sensitivity of 64%, a specificity of 100%, and an accuracy of 79% (Table 1).

While the diagnostic accuracy of glucose alone was not significantly higher than the combination of CEA and/or cytology (p=0.13), the combined accuracy of glucose, CEA and/or cytology was higher than that of CEA and/or cytology (92% vs. 79%, p=0.02) (Figure 1). The accuracy of EUS-nCLE was not significantly higher than the combination of glucose, CEA and/or cytology (98% vs. 92%, p=0.28).

Conclusion: Our study demonstrates the high diagnostic accuracy of CFG when measured using an automated chemistry analyzer validated for PCFs, supporting its adoption as a cost-effective diagnostic modality for differentiating PCLs.