Predicting patient outcome in lung cancer by a cluster analysis of lymphocyte population

A number of papers have established that a high density of tumor-infiltrating lymphocytes (TILs) is highly correlated with a better prognosis for many different cancer types. Recently, some studies have shown that the spatial interplay between different subtypes of TILs (e.g. CD3, CD4, CD8) is more...

Full description

Autores:: Barrera Monje, Cristian Raul

Tipo de recurso:

Fecha de publicación:: 2019

Institución:: Universidad Nacional de Colombia

Repositorio:: Universidad Nacional de Colombia

Idioma:: spa

Description
Summary:	A number of papers have established that a high density of tumor-infiltrating lymphocytes (TILs) is highly correlated with a better prognosis for many different cancer types. Recently, some studies have shown that the spatial interplay between different subtypes of TILs (e.g. CD3, CD4, CD8) is more prognostic of disease outcome compared to just metrics related to TIL density. A challenge with TIL subtyping is that it relies on quantitative immunofluoresence or immunohistochemistry, complex, expensive, and tissue-destructive techniques. In this thesis, we present a new approach to identify TIL sub-groups and quantify the interplay between these sub-populations to analyze the association of these interplay features with recurrence in early stage lung cancer. The approach comprises a Dirichlet Process Gaussian Mixture Model that clusters lymphocytes on H and E images according to the set of 186 features. The approach was evaluated on a cohort of 178 early stage non-small cell lung cancer patients, 100 were used for model training and 78 for independent validation. Recurrence information was extracted from the patient medical records, where 5 years survival is the established range for remained recurrence-free. A Linear Discriminant Analysis classifier was trained in conjunction with clustered lymphocytes to predict the likelihood of recurrence in the test set. The features yielded an AUC=0.84 compared to an approach involving just TIL density alone (AUC=0.58). In addition, a Kaplan-Meier analysis showed that those features were able to statistically distinguish early recurrence from late recurrence (p = 4 ∗ 10 −5 with cutoff alpha of 0.05).

Predicting patient outcome in lung cancer by a cluster analysis of lymphocyte population

Publicaciones similares