Abstract
BACKGROUND: This study aimed to establish a reference interval (RI) for the tumor marker carbohydrate antigen 72-4 (CA72-4) in healthy adults in Shenzhen, China, and to validate its applicability. METHODS: Serum CA72-4 results from 2614 adults undergoing routine check-ups were analyzed per CLSI EP28-A3c. High outliers were removed with Tukey's 1.5 × IQR rule. Distribution normality was tested by Shapiro-Wilk, and the need for partitioning by sex or age was assessed with the Harris-Boyd Z-test. The RI was derived non-parametrically from the 2.5th and 97.5th percentiles; a 90% confidence interval (CI) for the upper limit was obtained by 2000-bootstrap resampling. Performance was evaluated in an independent set of 206 adults (Apr 2025) by the exceedance rate (< 5% acceptable). RESULTS: CA72-4 values were markedly right-skewed. After excluding 258 outliers (9.9%), 2356 results remained. No significant sex- or age-related differences were found (Z < 3); thus, a combined RI was adopted. The 2.5th percentile was 0.54 U/mL (90% CI, 0.52-0.55) and the 97.5th percentile 8.5 U/mL (90% CI, 8.29-8.90). The oldest subgroup (61-85 year) showed a modest increase, but it did not affect the overall RI. In validation, 4.85% (10/206) exceeded 8.5 U/mL, meeting the preset criterion. CONCLUSIONS: An upper reference limit of 8.5 U/mL for CA72-4 in Shenzhen healthy adults-higher than the manufacturer's value-reduces false positives in health screening. Slight age-related increases suggest cautious interpretation in the elderly. This study provides an important reference for region-specific tumor marker intervals in Shenzhen and similar populations.