Nonparametric estimation of the size-metastasis relationship in solid cancers

Area of Science:

• Oncology
• Biostatistics
• Mathematical Biology

Background:

• Metastasis is a key factor in cancer mortality.
• Understanding the relationship between primary tumor size and metastasis is crucial for prognosis and treatment.
• Current methods for estimating tumor size at metastasis are limited.

Purpose of the Study:

• To develop and explore probabilistic models for the relationship between primary tumor size and metastasis.
• To estimate tumor size at the point of metastatic transition.
• To estimate the probability of detectable metastases at diagnosis.

Main Methods:

• Developed a probabilistic equation relating two functions: tumor size at metastatic transition and probability of detectable metastases.
• Employed the Expectation-Maximization (EM) algorithm for estimating unobservable tumor sizes at metastasis.
• Utilized nonparametric estimation methods, accounting for unmeasured tumors, especially metastatic ones.
• Applied methods to lung (epidermoid, adenocarcinoma), colorectal, and breast cancer data.

Main Results:

• Successfully applied estimation methods to various cancer types, including lung, colorectal, and breast cancer.
• The developed methods provide estimates for primary tumor size at the point of metastasis.
• Monte Carlo simulations demonstrated acceptably small bias in the estimation methodology.

Conclusions:

• The study provides a robust framework for analyzing the primary tumor size-metastasis relationship.
• The methods are applicable to real-world cancer data, offering insights into metastatic processes.
• The findings contribute to a better understanding of cancer progression and inform clinical strategies.