Comprehensive Genetic Characterization of 63 Myeloid/Lymphoid Neoplasms Associated with Eosinophilia and Rearrangement of PDGFR-α, PDGFR-β, FGFR1 or PCM1-JAK2

Background: The revision of WHO classification 2016 classifies myeloid or lymphoid neoplasms associated with eosinophilia (MLN-Eo) as a rare group of patients with genetic rearrangements of PDGFRα, PDGFRβ, FGFR1 and PCM1-JAK2. The current classification emphasizes the importance of genetics, because patients are also included in absence of eosinophilia but presence of aforementioned genetic events. Disease course and treatment options are heterogeneous and clinical presentation varies from myelofibrosis to T- or B-lymphoblastic leukemia depending on the rearrangement. Besides these driver rearrangements not much is known about the genetic landscape of MLN-Eo. A deeper understanding could shed light on disease biology and pathogenesis.

Aim: To molecularly characterize MLN-Eo cases using a panel of 48 genes known to be frequently mutated in myeloid or lymphoid malignancies.

Patients and Methods: We investigated 63 patients classified MLN-Eo (5 females, 58 males) with a median of 37% eosinophils in blood and a median age of 51 years (range: 19-79). Genetic rearrangements were detected by fluorescence in situ hybridization (FISH), chromosome banding and/or reverse transcriptase (RT) PCR. The cohort included the following aberrations: PDGFRα (n=37), PDGFRβ (n=13), FGFR1 (n=6) and PCM1-JAK2 (n=7). For mutation analysis, DNA was isolated from bone marrow (n=31) or peripheral blood (n=32) for sequencing on the NextSeq or MiSeq platform after custom TruSeq library preparation (all Illumina, San Diego, CA). SeqNext 4.3 (JSI Medical Systems, Kippenheim, Germany) was used for data analysis.

Results: In addition to the WHO defining aberrations, mutations were found in 14/63 (22%) patients. Three patients had more than one mutation (18 mutations in total). Most frequently mutated was RUNX1 (6/18 mutations, 33%). Nine of 18 mutations (50%) were found in epigenetic regulators (ASXL1, BCOR, DNMT3A and TET2). Other mutated genes were ETV6, NRAS or STAT5B (n=1 each). No mutation was detected in exons 8 and 17 of the KIT gene at a 3% sensitivity.

Frequencies of patients with mutations in the subgroups were the following: PDGFRα : 5/37 (14%), PDGFRβ : 3/13 (23%), FGFR1 : 5/6 (83%) and PCM1-JAK2 : 1/7 (14%). The rate of mutations in the group of FGFR1 rearrangements is significantly higher compared to the other subclasses (p=0.001). All mutations in the FGFR1 rearranged group were RUNX1 mutations. Among the other patients, only one RUNX1 mutation was found in a PDGFRα rearranged patient (p<0.001).

Next, we analyzed clonal composition using follow-up samples for mutated cases in 8 patients. We compared the development of mutation burdens (mutated/all reads) to malignant cell numbers. In three cases with a complete remission of PDGFRα rearrangement (RT-PCR negative), the DNMT3A (n=2) and TET2 (n=1) mutations were still present, indicating a clone that preceded the rearrangement or developed independently. Interestingly, DNMT3A and TET2 are genes frequently mutated in clonal hematopoiesis of indeterminate potential (CHIP), which should be considered for those cases.

For the other patients, the mutations are best explained to originate from MLN-Eo cells. Somatic mutations and PDGFRα - or PDGFRβ rearrangements were reduced/eradicated in parallel in three treated cases. Two patients with FGFR1 rearrangement showed a high number of malignant cells at initial diagnosis (92% and 97% by FISH), but only subclonal RUNX1 mutations (33% and 11% burden). FGFR1 rearranged cells were not significantly reduced under therapy, but RUNX1 burdens increased to 45% and 51%, indicating expansion of the mutated subclone.

Conclusions: (1) Of MLN-Eo patients, 22% showed at least one mutation in addition to WHO defining genetic aberrations. Five of six cases with FGFR1 rearrangement had a RUNX1 mutation. The mutation frequency was lower in the other subgroups and mutations were found mainly in epigenetic regulators. (2) Clonal development can be diverse and includes subclones which expand under therapy. The latter was observed for FGFR1 rearrangements adding another layer of complexity to treatment. (3) While the identification of PDGFRα, PDGFRβ, FGFR1 and PCM1-JAK2 rearrangements is critical for classification and targeted treatment, mutation screening adds to in depths understanding of pathogenesis and could even improve risk stratification and personalized therapy options.