Introduction. Diffuse large B-cell lymphoma represents a group of entities
characterized by pathological and biological heterogeneity and different
clinical outcomes. Due to pronounced heterogeneity, prognostic biomarkers
are of great importance in identifying high-risk patients who might benefit
from more aggressive approaches or new therapeutic modalities. Several
prognostic score systems have been established and applied to predict the
survival of patients with diffuse B-large cell lymphoma. The first
established prognostic system for NHL patients is the International
Prognostic Index, its variations Revised International Prognostic Index and
National Comprehensive Cancer Network- International Prognostic Index were
subsequently introduced in the era of immunochemotherapy. As the
discriminative power of clinical scores is suboptimal, other strategies have
been explored in order to improve risk stratification, especially in the
high-risk group of patients who have the highest risk of treatment failure.
In this regard, there is a tendency to integrate genetic and molecular
biomarkers and prognostic somatic mutations into standardized and
personalized models for risk stratification that would have a wide
application in routine clinical practice. The results of recent studies
based on machine learning methods have shown that the best risk
stratification is achieved by a combination of clinical, genetic and
molecular parameters, as well as a combination of clinical parameters with
new quantitative Positron Emission Tomography parameters, such as Metabolic
Tumor Volume and dissemination features and analysis of circulating tumor
DNA levels. This paper provides an overview of studies in which these new
risk stratification models were analyzed.