Abstract
It is often helpful to classify biomarker values into groups of different risk levels to facilitate evaluation of a biological, physiological, or pathological state. Stratification of patients into two risk groups is commonly seen, but there is always need for more than two groups for fine assessment. So far, there are no standard methods or tools to help decide how many cutoff points are optimal. In this study, we developed a comprehensive package that included methods to determine both the optimal number and locations of cutoff points for both survival data and dichotomized outcome. We illustrated workflow of this package with data from 797 patients with cervical cancer. By analyzing several risk factors of cervical cancer such as tumor size, body mass index (BMI), number of lymph nodes involved and depth of stromal invasion, in relation to survival and clinical outcome such as lymph nodal metastasis and lymphovascular invasion, we demonstrated that the best choice for BMI and stromal invasion was two cutoff points and one for the others. This study provided a useful tool to facilitate medical decisions and the analyses on cervical cancer may also be of interest to gynecologists. The package can be freely downloaded.
Highlights
Biomarkers have increasingly wide application in disease diagnosis, monitor of disease progression, assessment of prognosis, and development of pharmaceutical agents
We looked at the Akaike information criterion (AIC) values in case of body mass index (BMI) in relation to overall survival (OS), which were 749.9 with one cutoff and 744.9 with two, favoring two cutoffs to one
Stratification of patients is often useful for risk assessment or customized treatment plans
Summary
Biomarkers have increasingly wide application in disease diagnosis, monitor of disease progression, assessment of prognosis, and development of pharmaceutical agents. Biomarkers are usually represented by continuous values and ordinal numbers, but in practice, it is often helpful to classify biomarker values into groups of different risk levels to facilitate evaluation of a biological, physiological, or pathological state. A convenient tool to find an optimal cutoff point is, of high interest. Optimal number and location of cutoffs with application in cervical cancer
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have