Identifying clinically relevant biological characteristics in follicular lymphoma

The focus of this thesis was to identify tissue-derived clinically relevant biomarkers, for follicular lymphoma (FL) patients at diagnosis. Such knowledge is required for biomarker-guided decision-making and could contribute to the development of personalized treatment choices in daily hematological practice. For this, we collected diagnostic biopsy samples of FL patients, on which we explored the molecular landscape of FL tumor cells and the features of the non-malignant tumor microenvironment (TME). An end-of-spectrum study design was chosen, because it is particularly challenging to discover clinically meaningful prognostic or predictive biomarkers in FL, as a consequence of the extreme asymmetrical distribution of the prognosis. Only a minority of the FL patients have a poor outcome. Thereby such cases are highly underrepresented in consecutive series and very large numbers of patients need to be included to achieve sufficient numbers and statistical power. By extraction cases at the end of the survival spectrum for a head-to-head comparison, sensitivity and statistical power to discover prognostic or predictive biomarkers increases. The end-of-spectrum cohorts described in this thesis were FL patients; (1) with >5 years of response after 1st line-treatment versus patients progression, or death <2 years after or refractory on 1st line-treatment, (2) starting 1st line-treatment <3 months after diagnosis versus patients who deferred treatment for >5 years, and (3) stage I versus stage III/IV. This research was conducted within the "Lunenburg Lymphoma Biomarker Consortium" which is an international collaboration, because it is impossible for individual research groups to identify a sufficient number of patients who are at the end of the clinically defined spectrum. The non-malignant TME was visualized by utilizing immunohistochemistry markers for computerized single cell counts of T-cell populations and macrophages. A bioinformatics pipeline was developed to interpret the large amount of next generation sequencing (NGS) data, to study copy number aberrations (CNAs), mutations, and translocations. In this thesis we discovered that the difference in cell count of the different cell populations were, even in these clinically extreme cohorts, minor and too small to be used in daily practice to differentiate. With the NGS data we showed that EZH2 wild-type and gain of chromosome 18 were associated with early failure after 1st line-treatment. There were no genetic differences that could distinguish between either deferred treatment or requiring immediate treatment. In the stage I cases, which could be cured in about 45-65%, we revealed, with unsupervised cluster analysis on all genetical alterations, three distinct clusters. The first cluster is consistent with the BCL2-translocation positive cluster of reference stage III/IV cohort. The two other clusters had no comparable cluster in the stage III/IV cohort. Both these clusters were BCL2 translocation negative. One of these clusters was a heterogeneous cluster, characterized by a higher level of CNAs and mutations in linker histone genes, the other had a high frequency of STAT6 and CREBBP mutations. These two unique clusters suggest a different and characteristic pathogenesis. FL is characterized by a complex interaction between the TME and the malignant FL cells. The IHC technique used in this thesis is insufficient to explore the underlying complex interaction of the TME. Newer techniques such as combining in situ imaging, RNA sequencing and functional assays might be useful to explore the TME more in depth. The in depth TME information, combined with the NGS data might lead to a more personalized prognostic and or predictive tool in the future. The results from our studies contribute modest clinically relevant information with regard to the genetic and biological landscape of FL, to recognize patients at the end of the spectrum, and await further validation.