Measurement of both CEA and CA72–4at EUS-FNA of pancreatic cystic lesions may increase sensitivity of diagnosis

Mucinous and non-mucinous cystic tumours of the pancreas have different malignant potential and diagnosis of the former might be an indication for pancreatic resection.

CEA is the most commonly tested tumour marker in pancreatic cystic masses, with a level of >800ng/ml reportedly giving a sensitivity of 48% and specificity of 98% for mucinous vs. non-mucinous tumours (1). A more recently described marker, Ca72–4 has demonstrated a sensitivity of 80% and a specificity of 94% for mucinous vs. non-mucinous tumours at a level of >8U/ml. (2)

Aims:

To assess whether using either or both tumour markers gives the best sensitivity and specificity for mucinous over non-mucinous tumours.

Methods:

Since 2005 samples aspirated from pancreatic cystic masses at EUS-FNA have been routinely sent for analysis of CEA, Ca72–4 and amylase.

Abnormal cut of values are taken as CEA >800ng/ml and a Ca72–4 >8 U/ml.

Follow up was by chart review.

Statistical analysis to compare relationship between values was by Spearman's Rank Correlation.

Results:

Comparative paired results were available on 66 patients. Mean follow up was 18 months, but incomplete in 14.

Correlation of results was poor, r2=0.33, p>0.05.

In 9 cases both the CEA and Ca72–4 above threshold values; 2 adenocarcinoma, 2 IPMN, 2 mucinous cyst adenomas, one death of other causes, 2 lost to follow up.

In 12 cases there was discordance between the tumour marker values; 3 cases where only the CEA raised (One had cytology suggestive of IPMN), 9 cases just the Ca72–4 was abnormal (one proven IPMN at pancreatectomy, one adenocarcinoma at laparotomy).

In 45 cases neither tumour marker was raised; cytology and imaging also benign in 4, 2 glucagon positive neuroendocrine tumours, one paraganglioma and one mucinous adenoma on cytology. All others benign cytology and imaging.

Conclusion:

The discrepancy between the CEA or Ca72–4 levels in pancreatic cyst fluid indicates that a combination of tests will give a higher sensitivity. The limitation of this analysis is that the final diagnosis is only known on a minority of patients who have undergone pancreatectomy.