Detection of cervical cancer recurrence during follow-up: a multivariable comparison of 9 frequently investigated serum biomarkers.

ABSTRACT

OBJECTIVE: To assess the diagnostic accuracy and model the optimal combination of commonly studied serum biomarkers aimed at identifying recurrence in cervical cancer patients. METHODS: From a systematic literature search, nine biomarkers (CA-15.3, CA-125, CEA, CYFRA 21-1, hsCRP, IL-6, SCC-Ag, TNF-α and VEGF) were selected for a serum analysis. Samples were derived from a historical cervical cancer cohort. Subjects with serum samples stored in a biobank were included when quality criteria were met, and one sample preceding and at least one following primary treatment were available. In case of recurrence, two additional post-recurrence samples were analyzed. Biomarker serum levels were quantified by enzyme linked or chemiluminescence microparticle immunoassays. Logistic regression and receiver operating curve analysis were employed for selection, modeling and comparison on the diagnostic accuracy of the tested biomarkers. RESULTS: 205 samples were analyzed from 75 subjects, of whom 19 (25.3%) had a recurrence. The area under the curve (AUC) of CA-15.3, CA-125, CEA, CYFRA 21-1, IL-6, TNF-α and VEGF were all <0.750. Only SCC-Ag and hsCRP were included in the final model with an AUC of 0.822 (95% CI 0.744-0.900) and 0.831 (95% CI 0.758-0.905) respectively. Combined AUC was 0.870 (95% CI 0.805-0.935). Rises in SCC-Ag and hsCRP significantly increased the odds for recurrence. Each ng/ml of SCC-Ag increase, related to an odds ratio (OR) of 1.117 (95% CI 1.039-1.200). Comparably, the OR for hsCRP (in mg/ml) was 1.025 (95% CI 1.012-1.038). CONCLUSION: Combined testing of SCC-Ag and hsCRP yields the highest detection rate of disease recurrence during cervical cancer follow-up.