CYFRA 21-1 in patients with suspected cancer: evaluation of an optimal cutoff to assess the diagnostic efficacy and prognostic value.

OBJECTIVES

Chosen cutoff for cytokeratin 19 fragment antigen (CYFRA 21-1) as a tumor biomarker considerably influences its diagnostic and prognostic usefulness. The aim of the present study is to determine an optimal cutoff value for diagnostic validity of CYFRA 21-1 by technology in patients with suspected cancer and also to determine if CYFRA 21-1 levels provide prognostic value.

METHODS

A consecutive 284 patients suggestive of malignant disease from six hospitals of Madrid were enrolled in a retrospective design. Optimal CYFRA 21-1 cutoff value was obtained by receiver operating characteristic curve and Youden test. The diagnostic validity was evaluated according to sensitivity, specificity, predictive values and likelihood ratios. The prognostic value of CYFRA 21-1 was checked using multiple logistic regression. Thirty-two diagnostic cancers were confirmed.

RESULTS

The most optimal cutoff was 3.15 ng/mL. This cutoff showed a better specificity 93.63% (95% confidence interval [CI], 89.66-96.16), positive predictive value 60.98% (95% CI, 44.54-75.38) and positive likelihood ratio 12.65 (95% CI, 7.64-20.95) than the cutoff recommended by Fujirebio (1.8 ng/mL) (specificity: 73.71% [95% CI, 67.72-78.95], positive predictive value: 29.79% [95% CI, 21.02-40.23] and positive likelihood ratio 3.43 [95% CI, 2.71-4.35]), improving the current diagnostic accuracy. In multivariate analysis, elevated levels of CYFRA 21-1 (>3.15 ng/mL) was confirmed as an unfavorable prognostic factor.

CONCLUSIONS

The best cutoff for CYFRA 21-1 obtained was 3.15 ng/mL in patients with suspected cancer. This new cutoff decreases the false positive rate and improves the diagnostic efficacy of CYFRA 21-1 as a tumor marker as well as its association with death events.