Toward an optimal algorithm for ovarian cancer screening with longitudinal tumor markers

Area of Science:

• Gynecology
• Oncology
• Biomarker Research

Background:

• Ovarian cancer screening remains a challenge due to low sensitivity and specificity of current methods.
• Longitudinal monitoring of tumor markers like CA125II may offer improved early detection capabilities.

Purpose of the Study:

• To develop and validate a novel screening test for ovarian cancer using longitudinal CA125II levels.
• To assess the performance of the developed test in terms of specificity and positive predictive value.

Main Methods:

• Reassayed stored samples for CA125II in postmenopausal women.
• Developed a screening test using linear regression of log(CA125II) on time, analyzing slope and intercept.
• Utilized Bayes' theorem to calculate risk of ovarian cancer (ROC), accounting for assay variability.

Main Results:

• The developed screening test achieved a specificity of 99.8% on the validation set.
• The ROC algorithm identified 83% of ovarian cancers detected within a year.
• The estimated positive predictive value was 16%, significantly higher than single-assay methods.

Conclusions:

• A screening test based on longitudinal CA125II levels and assay variability shows promise for improved ovarian cancer detection.
• The developed ROC algorithm offers a substantial increase in positive predictive value compared to traditional single-point assays.
• Further studies are warranted to confirm the sensitivity of this longitudinal approach.