Abstract
Follicular lymphoma (FL) is characterized by an indolent clinical course and frequent relapses. Transformation to more aggressive disease is common and one of the main causes of death. The time to transformation may vary widely and some patients show early transformation and a poor prognosis. Main questions in FL center on markers to predict transformation (and prognosis) at the time of diagnosis and to accurately assess the transformed phase to tailor the choice of therapy. Currently, prognostic stratification for actual aggressiveness and overall survival is based on histological grading and clinical criteria; however, in up to 30% of all cases these methods prove to be insufficient. Using 18k cDNA arrays, we analyzed gene-expression patterns of 135 cases of FL addressing both issues.
1) Using supervised classification on a training set of paired samples from the indolent and aggressive disease episodes and on an independent validation set, a 81-gene-expression profile was established that could, with an accuracy of 100% and 93% respectively distinguish indolent from aggressive disease. Importantly, in a third series of FL with ambiguous histological grade, precluding meaningful morphological guidance, the profile showed a 94% classification accuracy supporting the value of this method for practical clinical use. The profile consists of genes involved in cell cycle control, DNA synthesis and metabolism (upregulated in high-grade disease), T-cell related genes (upregulated in low-grade disease) and T-cell and macrophage activation markers, that are significantly upregulated upon transformation. This profile, however, did not predict future transformation in FL samples at diagnosis.
2) A separate analysis was performed in a selection of the biopsy samples of not previously treated patients with grade 1 and 2 FL with transformation to DLBCL within 7–22 months after diagnosis and no transformation with a minimum follow-up of 108 months. Supervised and unsupervised cluster analysis showed no differences between both groups exceeding the threshold of significance to construct a predictor profile with a validated significance, underlining the biological homogeneity of the study cohort. Direct comparison analysis on the basis of signal-to-noise ratio’s suggested a discriminative role of the immune response with enhanced cytokine- and chemokine-mediated T-cell activation and antigen-processing in the extremes of the spectrum. However, differences were slight and the immune-response seemed to play a less distinctive role in patients with transformed disease after more than 3 years.
In conclusion, actual indolent and aggressive clinical behavior can be more reliably distinguished using a FL-stratification profile than the currently used histological grading and clinical criteria, and may provide an important alternative to guide the choice of therapy in FL patients both at presentation and at relapse. Although there may be subtle differences between rapidly transforming disease and long-term stable indolent disease, possible inherent variations are insufficient to reliably predict future transformation.
Author notes
Corresponding author
