PF677 DNA HYPERMETHYLATION EMERGES AS THE STRONGEST PREDICTOR FOR POOR OUTCOME AFTER ALLOGENEIC HEMATOPOIETIC STEM CELL TRANSPLANTATION IN JUVENILE MYELOMONOCYTIC LEUKEMIA (JMML)

Background: JMML is a childhood hematopoietic malignancy characterized by hyperactive RAS signaling. For most patients early allogeneic hematopoietic stem cell transplantation (HSCT) is the therapy of choice. Risk factors for early death and relapse following HSCT are represented by age ≥ 2 yrs, low platelet count and HbF elevated for age. Most recently, genome‐wide DNA methylation profiles (Lipka, Nat Comm 2017) identified distinct methylation signatures predictive for outcome. Aims: This study evaluates the clinical significance of these epigenetic changes in a large cohort of uniformly transplanted children registered to the European Working Group of MDS in Childhood (EWOG‐MDS). Methods: We analyzed the outcome of 153 children with JMML (96 males, 57 females; median age at HSCT 2.7 yrs [0.3–11.5]) given a HSCT after a preparative regimen consisting of busulfan, cyclophosphamide, and melphalan. Results: Thirty‐seven patients (pts) were transplanted from an HLA‐identical sibling donor (MSD) and 116 from a matched unrelated donor (UD). Stem cell source was bone marrow (n = 122) or peripheral blood (n = 31). GVHD prophylaxis was CSA only in 25 (68%) MSD procedures, CSA/MTX/ATG was administered in 88 (76%) of UD‐HSCT recipients. All but 9 pts engrafted. The cumulative incidence of grades II‐IV acute GVHD was 44% (37–52). Thirty‐three of the 133 pts at risk developed chronic GVHD, which was limited in 20 and extensive in 13 pts. With a median follow‐up of 3.7 years (0.1–17.5), the 5‐year leukemia‐free survival (LFS) was 61% (53–69), while the cumulative incidence of non‐relapse mortality (NRM) and relapse (CIR) were 13% (8–20) and 26% (20–35), respectively. There was no difference in outcome between MSD and UD‐HSCT recipients. Of the 153 pts, 76 had a somatic PTPN11 mutation, 21 had neurofibromatosis type 1, 22 belonged to the NRAS‐ and 18 to the KRAS ‐mutated subgroup, while RAS pathway alterations were absent in 16 pts (“all negative”). 5‐year CIR was higher in PTPN11 ‐mutated JMML (35%) compared to that of the NF1 (24%), NRAS (23%), KRAS (11%) or all negative subtypes (14%). Methylation status was high (HM), intermediate (IM) or low (LM) in 61, 52 or 40 pts, respectively. 5‐year LFS and CIR of pts of the LM class was similar to that noted in the IM class with 76% vs 69%, and 8% vs 13%, respectively. In contrast, children of the HM class had a significantly inferior LFS (44%, 31–57, p  < 0.01) due to a high CIR (50%, 38–65, p  < 0.01). NRM in the 3 methylation groups (LM 16%, IM 19%, HM 6%) did not differ. There was a clear correlation between methylation class and molecular genetic subtype: PTPN11 ‐mutated JMML was less common in the LM class and the proportion of KRAS ‐mutated pts was highest in the LM class. Also, patients in the HM class had a higher median age at HSCT (3.8 yrs) than pts in the IM (2.1 yrs) or LM class (1.1 yrs). Multivariate analysis identified methylation class (HM vs IM: RR 5.3 [1.8–15.3], HM vs LM RR 7.6 [1.6 – 37.3]) as the only independent variable predicting relapse when analyzed with age at HSCT, platelet count and molecular genetic type ( PTPN11 ‐mutated vs others). Summary/Conclusion: Patient outcome in genome‐wide methylation classes nicely recapitulated earlier data applying a DNA methylation score of candidate gene regions (Olk‐Batz, Blood 2011). Thus, DNA methylation classifiers are exceptionally consistent allowing prospective assignment of DNA methylation categories for molecularly‐driven risk stratification. Future research needs to be directed at better understanding the mechanistic link of epigenetic dysregulation and resistance to treatment.