Researchers from Taiwan have proposed a dual cut-off for the pancreatic tumor marker carbohydrate antigen 19-9 (CA19-9) in which the lower threshold identifies a small subgroup “nonproducers” with poor prognosis and the higher threshold detects the more common high-risk producers.
The approach “prevents the critical underestimation of disease severity and ensures more accurate risk stratification,” write Yung-Yeh Su, MD, from National Institute of Cancer Research in Tainan, and co-authors in Clinical Cancer Research.
Serum levels of CA19-9 (also known as sialyl Lewis antigen A) correlate with pancreatic cancer stage and prognosis in that higher levels indicate more advanced disease and a worse prognosis. Currently, a CA19-9 level below 37 units/mL is considered normal or, in the case of a known diagnosis of pancreatic cancer, prognostic for standard-risk disease.
However, approximately 10% of patients with pancreatic cancer do not have elevated CA19-9 levels, even in the presence of advanced disease. The Lewis antigen-negative status of these CA19-9 nonproducers comes from genetic polymorphisms in the FUT3 gene that impair the fucosyltransferase (FUT) activity required to produce CA19-9.
This is “clinically important because genotyping is rarely available, and Lewis-negative patients are often grouped with ‘low CA19-9’ cases, leading to an underestimation of risk,” Su et al. remark.
In 2012, a CA19-9 cut-off of five units/mL was proposed as a surrogate marker for CA19-9 nonproducers, but Su explained that this was largely an empirical observation based on clinicians noticing that Lewis-negative patients tended to exhibit extremely low baseline CA19-9 levels. It did not take genetic sequencing or genomic data into account.
Su and colleagues therefore set out to provide genomically validated evidence, by correlating CA19-9 levels directly with Lewis antigen genotypes, of a CA19-9 cutoff that would capture Lewis antigen-negative patients and better predict their prognosis.
The study included 615 patients with pancreatic ductal adenocarcinoma who underwent germline whole-exome sequencing to determine their FUT2/FUT3 genotypes.
Overall, 10.1% of participants were classified as FUT3-null, 33.0% as FUT-low, 35.6% as FUT-intermediate, and 21.3% as FUT-high.
The researchers report that median CA19-9 levels increased progressively across the FUT functional groups, from 2.4 units/mL in the FUT3-null group to 348 units/mL in the FUT-low group, 453 units/mL in the FUT-intermediate group, and 1300 units/mL in the FUT-high group.
By contrast, median overall survival (OS) was similar among the groups, at 13.5, 15.1, 14.5, and 14.3 months, respectively.
Using receiver operating characteristic analysis, Su and team identified seven units/mL as the optimal CA19-9 cutoff capable of characterizing the FUT3-null population within the clinical reference range of 37 units/mL or lower. At this cut-off, the sensitivity was 76.5% and the accuracy was 87.9%, which the researchers say represents 7.9 and 1.7 percentage point improvements, respectively, on the literature-defined threshold of five units/mL.
The investigators then grouped the patients by CA19-9 cutoff, without genotyping, and showed that OS was lower at each end of the spectrum.
Specifically, median OS was 13.5 months in participants with a CA19-9 of seven units/mL or lower, 23.2 months among those with a level of >7 to 37 units/mL, 22.0 months in those with a CA19-9 of >37 to 200 units/mL, and 12.8 months in the group with a level above 200 units/mL.
After adjusting for confounding factors, the risk for death was a significant 1.96 times higher in participants with the lowest CA19-9 levels and 1.69-fold higher in those with the highest levels, relative to individuals that had a CA19-9 of >7 to 37 U/mL.
“These data tell us that the conventional normal CA19-9 range of less than 37 units/mL does not distinguish between true low tumor burden and Lewis-negative status,” Su said.
He told Inside Precision Medicine that, at present, Lewis antigen genotyping is not standard clinical practice for pancreatic cancer patients due to added costs, turn-around times, and limited availability of routine genetic testing in many centers.
“This clinical gap is precisely why our findings are highly relevant,” Su remarked. “Because genetic testing is often unfeasible or unavailable, proposing a readily accessible serum CA19-9 cut-off (such as seven units/mL) provides clinicians with a highly feasible, cost-effective, and immediate surrogate marker to identify Lewis-negative patients using standard laboratory assays.”
He explained that robust external validation is now needed before the proposed lower cut-off can be adopted into routine clinical practice. “Because our current cohort may reflect specific regional or demographic characteristics, validating this seven U/mL cut-off in diverse cohorts—particularly in Western populations—is critical to ensure its global applicability.”
To address this, an international external validation study is currently underway.
The post Dual CA19-9 Threshold Improves Risk Stratification in Pancreatic Cancer appeared first on Inside Precision Medicine.