RNA-based targeted sequencing aids the detection of several types of variants in hematologic and other malignancies, including splice-altering variants. However, accurately identifying clinically relevant mis-splicing events remains challenging because of the inherent complexity of the human transcriptome and the high prevalence of false-positive splice junctions in deep RNA-sequencing data. To address these challenges, SpliceChaser and BreakChaser were developed, which are bioinformatics tools designed to enhance the detection and characterization of relevant splice-altering events. SpliceChaser improves the identification of clinically relevant atypical splicing by analyzing read length diversity within flanking sequences of the mapped reads around the splice junctions. BreakChaser processes soft-clipped sequences and alignment anomalies to enhance the detection of targeted deletion breakpoints associated with atypical splice isoforms generated from intrachromosomal gene deletions. These tools were developed and validated using a cohort of >1400 RNA-sequencing samples from patients with chronic myeloid leukemia. Collectively, SpliceChaser and BreakChaser achieved a positive percentage agreement of 98% and a positive predictive value of 91% for the detection of clinically relevant atypical splice-altering variants or gene deletions in the targeted regions. By integrating splicing and breakpoint detection with robust filtering strategies, these tools facilitate precise identification of clinically relevant variants, paving the way for improved diagnostics and therapeutic strategies in chronic myeloid leukemia and other malignancies.

Genomic profiling in patients with chronic-phase chronic myeloid leukemia (CP-CML) demonstrated somatic variants in blood cancer-related gene variants (CGVs) and rearrangements associated with the formation of the Philadelphia chromosome (Ph-associated rearrangements) at diagnosis, collectively termed additional genetic abnormalities (AGAs). AGAs had a negative impact on failure-free survival (FFS) and molecular response in imatinib-treated patients. We investigated whether treatment with more potent therapies could overcome the negative impact of AGAs at diagnosis. Targeted RNA-based next-generation sequencing was performed on diagnostic samples of 315 patients consecutively enrolled in 4 clinical trials of frontline potent tyrosine kinase inhibitors (TKIs) in CP-CML. AGAs were present in 34% of patients at diagnosis, including 20% harboring CGVs and 18% with Ph-associated rearrangements (4% had both). Although the negative impact of Ph-associated rearrangements was overcome by more potent inhibitors, patients with CGVs continued to experience inferior outcomes. This result was largely attributable to patients with ASXL1 variants, observed in 7% overall. Patients harboring ASXL1 variants also had inferior outcomes compared with those with wild-type ASXL1 in terms of 12-month major molecular response (55% vs 83%; P = .001), 2-year FFS (61% vs 91%; P < .001), and notably, the development of treatment-emergent BCR::ABL1 kinase domain mutations at 2 years (35% vs 1%; P < .001). In multivariable models, both CGVs and ASXL1 variants were predictors of each outcome. Treatment with frontline potent TKIs overcame the negative impact of Ph-associated rearrangements observed with frontline imatinib. However, inferior outcomes were still associated with the presence of CGVs. The acquisition of TKI-resistant BCR::ABL1 mutations was almost exclusively associated with mutated ASXL1 at diagnosis.

Understanding of the heterogeneity in the formation of the Ph has evolved as technology has advanced. In the pre-TKI era, FISH revealed derivative 9 deletions that were associated with shorter survival. The poor outcomes were overcome with imatinib. Whole exome and transcriptome sequencing revealed greater complexity: inversions; novel 9;22 fusions; sequence fragmentation; and random reassembly. These were termed Ph-associated rearrangements. They occurred on the Ph and/or the derivative 9, were thought to occur at the same time as the Ph and correlated with poorer response to imatinib. To use whole genome sequencing (WGS) to examine derivative chromosomes and further resolve the complexity of Ph-associated rearrangements in patients at diagnosis and pan-TKI resistant cell lines. WGS was performed for 7 chronic phase CML patients at diagnosis and 7 matched sensitive/resistant Ph-positive cell lines. Resistance was induced by exposure to increasing concentrations of imatinib. Data was analyzed using Oncoanalyzer. Derivative chromosome structure was resolved using the LINX algorithm, which clusters structural variants that are highly unlikely to occur independently. Rearrangements were validated using orthogonal methods, including Optical Genome Mapping and SNP Array. Six of 7 patients had pre-characterized Ph-associated fusions identified using RNAseq. For most patients, a balanced reciprocal Ph translocation involves 2 rearrangements: 1 that generates the Ph and BCR::ABL1; and 1 reciprocal derivative 9 with ABL1::BCR. These structures were resolved for 1 patient without pre-characterized Ph-associated rearrangements and for 5 of 7 parental cell lines. A distinct class of complex structural rearrangement called chromoplexy was identified involving the Ph translocation in 11 samples. These Ph-associated rearrangements had the highest number of linked structural variants within the individual genomes of 10 of the 11 samples. Chromoplexy involves the formation of extensive chained rearrangements affecting multiple chromosomes that create an intricate series of inter/intra-chromosomal fusions, deletions, duplications and inversions. Many DNA strands were broken and ligated in new configurations in the affected samples. Multiple genes on the derivative chromosomes were disrupted, including recurrent cancer drivers. For 3 of 7 matched sensitive/resistant cell lines, Ph-associated rearrangements were acquired in the resistant lines. The most complex events occurred in the LAMA84 resistant line. The parallel sensitive line had the standard 2 rearrangements. However, the resistant line gained 17 rearrangements with highly amplified Ph sequences, inversions and novel genomic fusions. Remarkably, small fragments from 6 chromosomes (2, 4, 10, 13, 17, X) were inserted upstream and downstream of BCR::ABL1, and disrupted the NUP214 driver gene. The K562 sensitive line displayed 24 rearrangements with non-contiguous amplified Ph regions of different copy number. Events included NOTCH1 deletion, 9;22 NUP214 fusion and an adjacent 9;13 fusion. Six fragments from chromosome 13 were incorporated into the Ph. The K562 resistant line displayed further evolution with 31 Ph rearrangements. The AR230 sensitive line displayed Ph chromoplexy with 9 rearrangements. The resistant line evolved with 19 additional Ph rearrangements. A fragment from chromosome 21 was inserted upstream of BCR::ABL1 and disrupted the MAPK1 driver gene. The LZTR1 tumor suppressor was deleted. All other cell lines maintained the standard Ph configuration. Ph-associated chromoplexy was evident in 6 patient samples: average 10 rearrangements. The patient with the highest number (15) progressed to blast phase at 4 months of imatinib. Four Ph-associated subclones were detected at diagnosis with non-contiguous amplified regions, consistent with the evolved cell lines. Furthermore, the LZTR1 tumor suppressor gene was deleted, as detected in AR230. The blast phase sample was unavailable. Deep molecular response was achieved in 1 patient on imatinib and 4 on frontline or switch to nilotinib/dasatinib. We have demonstrated that Ph chromoplectic rearrangement can be present at diagnosis but can also be initiated during leukemic progression and disrupt multiple genes, including cancer drivers. The Ph may be prone to extensive DNA breakage. Ph chromoplexy could constitute a new mechanism that contributes to acquired TKI resistance.

Background Treatment failure in CML is associated with additional variants in genes that drive acute myeloid and lymphoid leukemia. Some variants are detected at diagnosis where ASXL1 is the most commonly mutated gene (~9% of patients [pts]), whereas others are acquired during therapy. Notably, multiple studies have shown a high frequency of concurrence of BCR::ABL1 kinase domain (KD) mutations and blood cancer-related gene variants (herein named cancer variants). Little is known about the dynamics of cancer variants over time and whether they precede the acquisition of KD mutations. Aim To interrogate longitudinal samples to assess the dynamics and time differential between cancer variant detection, KD mutations and treatment failure. Method A sensitive RNA-based hybridization capture next-generation sequencing method was used to target 53 myeloid and lymphoid blood cancer genes. Single nucleotide variants, small insertions/deletions, RNA splice altering variants, and specific gene fusions and deletions were detectable. Serial samples of 35 pts with treatment failure (2020 European LeukemiaNet criteria) were tested. Samples sequenced had BCR::ABL1 ≥1% IS by qRT-PCR. Only variants that met strict criteria for pathogenicity were retained. Results Pts were treated with frontline BCR::ABL1 inhibitors and received treatment intervention for lack of milestone molecular responses. The first failure criterion for the 35 pts occurred at a median of 6 months after commencing therapy, range 6-70 months: 22 pts failed milestone molecular responses; 9 lost response with a BCR::ABL1 rise above 1% IS; and 4 developed sudden blast phase. At last follow up, 8 pts had developed lymphoid blast phase, 10 myeloid blast phase, 2 accelerated phase, 3 received allogeneic transplants in chronic phase, 2 died in chronic phase and 10 maintained chronic phase. At diagnosis, 32 cancer variants were detected in 18 of 35 pts (51%). ASXL1 variants were enriched at diagnosis: 11 of 35 pts (31%). However, only 2 of these 11 pts later developed blast phase compared to 16 of the remaining 24 pts. The emergence of cancer variants over time was assessed. Pts had a median of 4 tests, range 2-18. BCR::ABL1 fusion transcripts were detected in all samples using the RNA-based gene panel. Overall, 31 of 35 pts (89%) with treatment failure had cancer variants detected in the follow up samples. In total, 72 cancer variants were gained after commencing therapy. Additionally, 29 BCR::ABL1 KD mutations were gained in 15 of 35 pts (43%). Every pt with a KD mutation also had 1-8 additional cancer variants. Notably, of the 15 pts who gained KD mutations, 9 had cancer variants at diagnosis and 2 other pts gained cancer variants prior to KD mutation detection. Before the first failure criterion occurred in the 35 pts, cancer variants were detected in 19 pts at a median of 6 months before failure, range 2-43 months. All 18 pts in blast phase had cancer variants. Gene fusions and deletions were a common feature of blast phase and were detected in 13 of 18 pts (72%). RUNX1 variants were also enriched in blast phase, 8 of 18 pts (44%), but emerged in only 1 of the other 17 pts with treatment failure. The time differential between the onset of blast phase and the acquisition of cancer variants was examined. Deletions affecting IKZF1, RB1, CDKN2A, BTG1 and SETD2 were detected in 8 pts and coincided with the onset of blast phase in 6. The onset of blast phase occurred within 4 months of first detection for 6 of 7 gene fusions involving KMT2A, RUNX1, ETV6, CBFB and MECOM. The latency of blast phase for the 8 pts with RUNX1 variants was variable: median 5.7 months after first detection of the RUNX1 variant, range 0-48. Cancer variant dynamics over time revealed: 1) several cases of convergent evolution where multiple different variants in the same gene were detected in an individual pt; 2) new variants that emerged with a change of BCR::ABL1 inhibitor; 3) clonal competition where some variants dominated in blast phase; 4) specific variants associated with the onset of blast phase; and 5) cancer variant detection prior to KD mutations in most pts. Conclusion Our study highlights the potential for sensitive genomic testing to enhance prediction of relapse and transformation. Cancer variants were frequently detected prior to treatment failure and KD mutations. The detection of specific variants associated with progression could warrant early consideration of allograft options.

Background We previously demonstrated that additional genetic abnormalities (AGAs) detected at diagnosis had a negative impact on failure-free survival (FFS, as defined by ELN 2020) and molecular response for 200 frontline imatinib-treated patients (pts), despite proactive treatment intervention. The effect was evident for blood cancer related gene variants (CGVs) and aberrant structural rearrangements associated with the formation of the Philadelphia chromosome (Ph-associated). We have now sequenced the diagnosis samples of 315 pts treated with more potent frontline BCR::ABL1 inhibitors. Aim To determine if frontline more potent BCR::ABL1 inhibitors overcome the negative impact of AGAs detected at diagnosis. Methods An RNA-based targeted myeloid and lymphoid cancer-related gene next-generation sequencing method was used to detect single nucleotide variants, RNA splice altering variants, small insertions/deletions, gene fusions and focal gene deletions. Blood samples from pts enrolled in consecutive Australasian frontline CML clinical trials were investigated; 200 imatinib-treated pts (TIDEL II) and 315 of 335 pts treated with more potent BCR::ABL1 inhibitors (dasatinib: DIRECT n=76; nilotinib: PINNACLE n=59 and ENESTxtnd n=81; asciminib: ASCEND n=99). Kaplan-Meier and cumulative incidence analyses were performed to assess the effect of AGAs on FFS, kinase domain (KD) mutation acquisition (including ATP-site and myristoyl-site) and molecular response. Results The incidence of AGAs in the total cohort of 515 pts was 32%: cancer gene variants (CGVs 18%) and Ph-associated rearrangements (Ph-ass 18%). ASXL1 variants were most frequently detected: 41/515 pts (8%). We previously demonstrated that AGAs at diagnosis were associated with inferior FFS and molecular response for the imatinib-treated pts. Notably, Ph-associated rearrangements were strongly associated with the acquisition of KD mutations. No differences were observed for overall or transformation-free survival. For pts treated with more potent inhibitors, the negative impact of Ph-ass for FFS, KD mutations and molecular response was overcome. However, CGVs were associated with a significantly higher rate of treatment failure (2-year FFS 76 vs 92%, P<0.001), acquisition of KD mutations at 2 years (11% vs 0.3%, P<0.001), and a lower rate of major molecular response (12-month MMR 63% vs 82%, P=0.002) for pts treated with more potent inhibitors with a median follow up of 24 months. These inferior responses were observed irrespective of whether pts were treated with nilotinib/dasatinib or asciminib. Notably, ASXL1 variants at diagnosis were associated with inferior 12-month MMR (55% vs 82%, P=0.033), 2-year FFS (68% vs 93%, P<0.001) and higher 2-year KD mutation acquisition (27% vs <0.3%, P<0.001) compared to pts without CGVs. Among asciminib-treated pts with ASXL1 variants, the cumulative incidence of KD mutations at 24 months was 37%. Non-ASXL1 variants also predicted for inferior 12-month MMR. Irrespective of whether pts with CGVs at diagnosis were treated with frontline imatinib or more potent inhibitors, there was no significant difference in 12-month MMR (53% vs 63%, P=0.34), 24-month FFS (71% vs 76%, P=0.57) or 24-month KD mutation acquisition (12% vs 11%, P=0.95). Prognostic factors for pts treated with more potent inhibitors were examined in multivariable models. CGVs at diagnosis were the only predictors of all 3 outcomes: 12-month MMR [HR 0.54 (0.36-0.80), P=0.003], 2-year FFS [HR 3.57 (1.55-8.19), P=0.003] and KD mutations [HR 30.4 (3.75-246.9), P=0.001]. In a separate model, ASXL1 variants also predicted 12-month MMR [HR 0.47 (0.24-0.92), P=0.003], 2-year FFS [HR 1.65 (1.04-2.64), P=0.035] and KD mutations [HR 73.9 (8.8-619), P<0.001]. High risk ELTS also predicted 12-month MMR [HR 0.50 (0.30-0.80), P=0.009]. No difference was observed for 24-month MR4 and overall or transformation-free survival. Conclusion CGVs at diagnosis were associated with inferior outcomes, even with more potent frontline inhibitor therapy. KD mutations almost exclusively developed in pts with CGVs at diagnosis. However, the negative impact of Ph-associated rearrangements observed for imatinib-treated pts was abrogated by more potent inhibitors. The data suggest a potential beneficial role for combination therapy in this cohort of pts where the risk of treatment failure driven by KD mutations is high.

Chronic myeloid leukemia (CML) is a model of genomically based diagnosis and management where BCR :: ABL1 is successfully targeted by tyrosine kinase inhibitor (TKI) therapy in most patients. The dynamics of BCR :: ABL1 transcript decline during therapy is a dependable biomarker of response, relapse, and drug resistance. Missense mutations acquired within the BCR :: ABL1 kinase domain that disrupt TKI binding can evolve during therapy and are frequently detected in patients for whom TKI treatment fails. Importantly, specific BCR :: ABL1 missense mutations are targetable alterations and direct therapeutic decisions based on the individual mutant TKI sensitivity profile. Nevertheless, BCR :: ABL1 mutations are only implicated in approximately half of the cases of acquired resistance. Furthermore, not all patients with a single BCR :: ABL1 mutation that is predicted to be sensitive to a specific TKI will experience a response when switched to that TKI. Progression to blast phase heralds independence from BCR :: ABL1 , and this phase of the disease is notoriously difficult to treat. The independent drivers of resistance and disease progression have long been investigated to both predict progression and to find targets for therapeutic intervention. Recent data reaffirm that drug resistance and disease progression is a mutation-driven process in CML, and somatic variants in genes that are known to drive acute myeloid and lymphoid leukemia have been detected in patients in the advanced phases of CML. Genomic testing over the last few decades for patients with blood cancer has revealed of variety of genomic aberrations that drive disease. Consequently, incorporation of genomic factors into patient management for a range of blood cancers has led to the implementation of high-throughput gene testing to detect clinically actionable variants. Is it time to integrate broader genomic screening into clinical management strategies for patients with CML?

Background The gain of mutations with a competitive advantage in hematopoietic cells of healthy individuals is common with aging and termed clonal hematopoiesis of indeterminate potential (CHIP). Through sequencing studies we detected CHIP related mutants in chronic myeloid leukemia (CML) patients in deep molecular response. CHIP mutants may provide a selective growth advantage to non-leukemic cells, driving their clonal expansion in remission and potentially influencing treatment-free remission (TFR) if CHIP clones outcompete an emerging BCR:: ABL1 clone. Aim To determine if CHIP mutants detected at the time of stopping tyrosine kinase inhibitor (TKI) therapy could influence achievement of TFR. Methods An RNA-based sequencing method targeting 17 of the most frequently mutated CHIP genes was developed and applied to blood samples of 150 patients who attempted TFR. The median follow up of patients in TFR was 71 months. CHIP mutants met strict criteria for pathogenicity based on their potential to confer growth and survival advantages, i.e. the same criteria as cancer driver mutations. Variant allele frequency (VAF) of ≥0.4% was reproducibly detected. CHIP at the time of cessation, termed CHIP TOC, was defined as 1) ≥1 CHIP mutant with VAF ≥2.0% (n=30 patients) or 2) a novel criterion for patients with multiple CHIP mutants where the combined VAF was ≥2.0% (n=7). Overall, 25/37 patients (68%) with CHIP TOC had multiple mutations, range 2-7. Results CHIP TOC was detected in 37/150 patients (25%). The average age of patients with CHIP TOC was 66.3 years vs 58.0 for those without, P=.0004, which is consistent with the age related nature of CHIP. No patient aged <40 had CHIP TOC, whereas 44% of patients ≥70 had CHIP TOC. Twelve genes were mutated and the most frequent were TET2, DNMT3A, ASXL1 and PPM1D. The overall probability of TFR was 60.4% at 12 months and 53.0% at 84 months. CHIP TOC was associated with TFR. At 12 months, 83.6% of patients with CHIP TOC maintained TFR versus 52.9% of patients without CHIP TOC, P=.002. At 84 months, 62.9% of patients with CHIP TOC maintained TFR versus 49.6% of patients without CHIP TOC, P=.014. We previously demonstrated that e13a2 transcripts and a slow initial decline of BCR:: ABL1 after starting frontline TKI therapy, measured as the number of days over which BCR:: ABL1 halved, were our strongest predictors of relapse. Prognostic factors for the achievement of TFR at 12 and 84 months were examined in multivariable analysis. Candidate prognostic factors were CHIP TOC, BCR:: ABL1 halving time, BCR:: ABL1 transcript type, age at diagnosis, gender, duration of MR4.5 and time on TKI before TKI stop. The independent predictors of TFR at 12 and 84 months were the presence of CHIP TOC at TKI stop, a shorter BCR:: ABL1 halving time (indicating a more rapid initial BCR:: ABL1 decline) and the e14a2 (plus e14a2/e13a2) BCR:: ABL1 transcript type (Table 1). The rate of late relapse, defined as loss of MMR >12 months after cessation, was higher in patients with CHIP TOC. Overall, 85 patients had TFR at 12 months and late relapse occurred in 8. A landmark analysis demonstrated that the probability of maintaining TFR at 84 months for these 85 patients was 93.8% for those without CHIP TOC (n=57) and 75.2% for those with CHIP TOC (n=28), P=.007. The only variable associated with late relapse among the patients with CHIP TOC was fluctuation of BCR:: ABL1 within the first 12 months after cessation, defined as irregular rise and fall comprising 2 consecutive BCR:: ABL1 ratios with loss of MR4.5, or a single loss of MR4. Overall, 13 patients had fluctuation within the first 12 months and 5 of these had late relapse. All 5 had CHIP TOC. Patients with CHIP TOC and stable MR4.5 by 12 months had a 93.8% probability of TFR at 84 months (Table 2). Conclusion The presence of CHIP TOC at TKI cessation, a shorter BCR:: ABL1 halving time and e14a2 BCR:: ABL1 transcripts were independent predictors of TFR at 12 and 84 months. Delayed loss of MMR that occurred for a minority of patients with CHIP TOC may be associated with competition between CHIP clones and a residual leukemic clone. These patients had fluctuating BCR:: ABL1 over the first 12 months after cessation and prior to late relapse. Novel strategies may be warranted to reduce the risk of late relapse for these patients. Patients who sustained TFR by 12 months in stable MR4.5 had a reassuringly low risk of late relapse. CHIP status could be a key variable to guide TFR decision-making and monitoring.

The BCR::ABL1 gene fusion initiates chronic myeloid leukemia (CML); however, evidence has accumulated from studies of highly selected cohorts that variants in other cancer-related genes are associated with treatment failure. Nevertheless, the true incidence and impact of additional genetic abnormalities (AGA) at diagnosis of chronic phase (CP)-CML is unknown. We sought to determine whether AGA at diagnosis in a consecutive imatinib-treated cohort of 210 patients enrolled in the TIDEL-II trial influenced outcome despite a highly proactive treatment intervention strategy. Survival outcomes including overall survival, progression-free survival, failure-free survival, and BCR::ABL1 kinase domain mutation acquisition were evaluated. Molecular outcomes were measured at a central laboratory and included major molecular response (MMR, BCR::ABL1 ≤0.1%IS), MR4 (BCR::ABL1 ≤0.01%IS), and MR4.5 (BCR::ABL1 ≤0.0032%IS). AGA included variants in known cancer genes and novel rearrangements involving the formation of the Philadelphia chromosome. Clinical outcomes and molecular response were assessed based on the patient's genetic profile and other baseline factors. AGA were identified in 31% of patients. Potentially pathogenic variants in cancer-related genes were detected in 16% of patients at diagnosis (including gene fusions and deletions) and structural rearrangements involving the Philadelphia chromosome (Ph-associated rearrangements) were detected in 18%. Multivariable analysis demonstrated that the combined genetic abnormalities plus the EUTOS long-term survival clinical risk score were independent predictors of lower molecular response rates and higher treatment failure. Despite a highly proactive treatment intervention strategy, first-line imatinib-treated patients with AGA had poorer response rates. These data provide evidence for the incorporation of genomically-based risk assessment for CML.

Background Treatment-free remission (TFR) is a goal of therapy. However, ~50% of patients who attempt TFR will relapse, which mostly occurs by 6 months. The acquisition of somatic mutations in hematopoietic cells is common with age (clonal hematopoiesis, CH). Through next-generation sequencing studies we identified CH in patients in deep molecular remission. CH variants provide a selective growth advantage to cells, driving their clonal expansion. The clones also give rise to mutated immune effector cells with a proinflammatory profile that exacerbate diseases with a chronic inflammatory component, such as atherosclerosis. CH variants could potentially influence TFR, either by outcompeting an emerging residual CML clone or by altering immune function, which is known to play an important role in TFR. Aim To determine whether CH variants present at the time of therapy cessation could influence TFR. Methods An RNA-based hybridization capture sequencing method targeting 17 of the most frequently mutated CH genes was developed and applied to samples of 138 patients who attempted TFR. CH variants met strict criteria for pathogenicity based on their potential to confer growth and survival advantages. The CH status was confirmed by repeat testing of samples of 28 patients. The method reproducibly detected variants with a variant allele frequency (VAF) of ≥0.4%. Results The median follow-up for the 138 patients was 66 months from the TFR attempt. Molecular recurrence was defined as loss of major molecular response (MMR, BCR::ABL1 ≤0.1%). The probability of TFR was 64.5% at 6 months, 59.4% at 12 months and 51.0% at 84 months. CH variants were detected in 61/138 patients (44%): 107 variants in 12 genes. Multiple variants, range 2 - 7, were detected in 26 patients (19%). The median VAF was 0.9%, range 0.4 - 44%. Variants most frequently occurred in TET2 (44% of variants), DNMT3A (25%), PPM1D (7.4%) and ASXL1 (6.5%). The average age at cessation was 59.6 years, range 33 - 85. The average age of patients with CH was higher than those without CH: 64.7 versus 55.5 years, P < .0001. Only 3/13 patients (15%) aged ≤40 had CH compared with 6/7 patients (86%) ≥80. CH was significantly associated with TFR to 36 months after cessation. The optimal criteria for TFR prediction were at least one CH variant of VAF >2% or >2 variants (CH >2%/>2). Thirty-one of 138 patients (22%) had CH of VAF >2% (n = 24) or >2 variants (n = 7). The probability of TFR for these 31 patients was significantly higher than for the remaining patients when assessed at 6, 12, 24 and 36 months after cessation, Figure. The probability of TFR at 6 months was 90.3% for patients with CH >2%/>2 versus 57.0% for the remaining patients, P = .001. The probability of TFR at 36 months was 65.9% versus 51.2%, P = .049. A landmark analysis was performed at 12 months for the 74 patients who maintained TFR at 12 months. MMR was subsequently lost in 8/74 patients and the probability of TFR at 84 months was 86.9%. CH >2%/>2 occurred for 23/74 patients (31%). The probability of late molecular recurrence was higher among the patients with CH >2%/>2: 27.3% versus 5.5% for the remaining patients, P = .009. Overall, molecular recurrence occurred for 12/31 patients (39%) with CH >2%/>2 and 53/107 of remaining patients (50%). Time to molecular recurrence after cessation was longer for patients with CH >2%/>2 than the remaining patients: median 12.5 months versus 3.5 months, P < .0001. The longer time to molecular recurrence was linked to a different pattern of loss of MMR. The median time between first loss of MR4 (BCR::ABL1 ≤0.01%) and loss of MMR was 145 days for patients with CH >2%/>2 and 28 days for remaining patients, P = .0007. BCR::ABL1 values more frequently fluctuated before loss of MMR in patients with CH >2%/>2: 8/12 patients (67%, fluctuation over 120-1335 days) versus 6/53 (11%, fluctuation over 85-715 days) in the remaining patients, P =.016. Conclusion CH may improve the probability and duration of TFR, delaying the time to molecular recurrence, which has implications for patient selection and monitoring. Patients with CH who are eligible for therapy cessation have an excellent prospect of remaining drug-free for 1-2 years but their long term prospects of sustained TFR may be less certain compared to other patients, given the slower dynamics of relapse. Delayed relapse could be due to competition for clonal dominance between residual leukemic cells and larger CH clones, or modulation of the immune microenvironment. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

The chronic myeloid leukemia (CML) treatment success story is incomplete as some patients still fail therapy, leading to end-stage disease and death. Here we discuss recent research into CML incidence, the role of comorbidities on survival and detecting patients at risk of failing therapy. The incidence of CML has fallen markedly in high social-demographic index (SDI) regions of the world but there is disturbing evidence that this is not the case in low and low-middle SDI countries. Now that CML patients more frequently die from their co-morbid conditions than from CML the Adult Comorbidity Evaluation-27 score can assist in risk assessment at diagnosis. Non-adherence to therapy contributes greatly to treatment failure. A good doctor-patient relationship and social support promote good adherence, but patient age, gender, and financial burden have negative effects, suggesting avenues for intervention. Mutations in cancer-associated genes adversely affect outcome and their detection at diagnosis may guide therapeutic choice and offer non-BCR::ABL1 targeted therapies. A differential gene expression signature to assist risk detection is a highly sought-after diagnostic tool being actively researched on several fronts. Detecting patients at risk of failing therapy is being assisted by recent technological advances enabling highly sensitive genomic and expression analysis of insensitive cells. However, patient lifestyle, adherence to therapy, and comorbidities are critical risk factors that need to be addressed by interventions such as social and financial support.

Background We previously performed an integrative genomic analysis for 47 chronic phase (CP)-CML patients at diagnosis who were specifically selected based on optimal or very poor outcome to tyrosine kinase inhibitors (TKIs). Patients who subsequently progressed to blast phase (BP) had a significantly higher frequency of cancer-related gene mutations at diagnosis compared with patients who achieved an optimal response (54% versus 16%). A novel mutational subtype associated with the formation of the Philadelphia (Ph) chromosome, termed Ph-associated rearrangements, were also described and included novel fusions, deletions and inversions adjacent to the BCR::ABL1 junction. These occurred at a higher frequency of patients who progressed to BP compared to those with an optimal response (33% versus 11%). Consequently, we sought to investigate the true incidence and effect of cancer-gene mutations and Ph-associated rearrangements (collectively termed additional mutational events, AMEs) detected at diagnosis and whether outcomes differ based on the TKI initiated. Aim To assess the impact on outcome and molecular response of AMEs detected at diagnosis for patients consecutively treated with imatinib compared to patients treated with second-generation TKIs (2G-TKI). Methods A recently validated hybridization capture sequencing method targeting genes involved in both myeloid and lymphoid malignancies was applied to available diagnostic RNA of consecutive clinical trial patients treated with either frontline imatinib or 2G-TKIs. The 200 imatinib-treated patients (TIDEL II; CML9) were commenced on 600mg daily of imatinib and managed with a highly proactive treatment intervention strategy for suboptimal response, which was either imatinib dose increase or switch to nilotinib. The 2G-TKI patients were enrolled in either DIRECT;CML12 (n=72), where patients commenced dasatinib 100mg daily before dose modification, or either ENESTxtnd or PINNACLE;CML11 (n=44) where patients commenced nilotinib 300mg twice daily. Variants in cancer genes with variant allele frequencies ≥5%, including single nucleotide variants, RNA splice altering variants, small insertions/deletions, gene fusions and focal gene deletions were assessed for pathogenicity using strict criteria. Ph-associated rearrangements were identified. Clinically relevant variants were confirmed as somatic and validated. Kaplan-Meier and cumulative incidence analyses were performed to assess the effect of AMEs on failure-free survival and key molecular outcomes by 36 months. Failure events were as defined by the European LeukemiaNet 2020. Results The incidence of AMEs for imatinib-treated patients was 31% (61/200): 33 (16%) cancer gene mutations and 36 (18%) Ph-associated rearrangements; 8 patients had both subtypes. In the 2G-TKI group, 41 AMEs were identified in 29% (34/116) of patients: 20 (17%) cancer gene mutations and 16 (14%) Ph-associated rearrangements; 2 patients had both subtypes. In total, 14 cancer genes were mutated. Consistent with other studies, ASXL1 mutations were most frequently detected: 18/200 imatinib patients (9%) and 9/116 2G-TKI patients (8%). Using 36-month outcomes as comparison, imatinib-treated patients with AMEs had inferior failure-free survival (84% versus 69%, P=.03), major molecular response (84% versus 72%, P=.02) and MR4 (62% versus 37%, P=.001). In contrast, outcomes in the 2G-TKI cohort were similar irrespective of the presence of AMEs, with regards to failure-free survival (94% versus 91%, P=.47) and achievement of major molecular response (88% versus 83%, P=.71). Interestingly, achievement of MR4 was delayed for patients with AMEs compared to those without AMEs at diagnosis: 51% versus 69%, P=.02 by 24 months; and 63% versus 70%, P=.40 by 36 months. No differences were observed for overall survival or transformation-free survival in either the frontline imatinib or 2G-TKI cohorts. Conclusion These results indicate that while the presence of AMEs do predict for inferior outcomes and molecular responses when patients are treated with frontline imatinib, the more potent 2G-TKIs may overcome the negative impact of the AMEs. The effect for imatinib treated patients was evident despite a starting dose of 600mg and more stringent treatment intervention than currently recommended. This supports a strategy of mutation analysis at diagnosis of CP-CML for optimal TKI selection. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Mutation detection is increasingly used for the management of hematological malignancies. Prior whole transcriptome and whole exome sequencing studies using total RNA and DNA identified diverse mu-tation types in cancer-related genes associated with treatment failure in patients with chronic myeloid leukemia. Variants included single-nucleotide variants and small insertions/deletions, plus fusion transcripts and partial or whole gene deletions. The hypothesis that all of these mutation types could be detected by a single cost-effective hybridization capture next-generation sequencing method using total RNA was assessed. A method was developed that targeted 130 genes relevant for myeloid and lymphoid leukemia. Retrospective samples with 121 precharacterized variants were tested using total RNA and/or DNA. Concordance of detection of precharacterized variants using RNA or DNA was 96%, whereas the enhanced sensitivity identified additional variants. Comparison between 24 matched DNA and RNA samples demonstrated 95.3% of 170 variants detectable using DNA were detected using RNA, including all but one variant predicted to activate nonsense-mediated decay. RNA identified an addi-tional 10 variants, including fusion transcripts. Furthermore, the true effect of splice variants on RNA splicing was only evident using RNA. In conclusion, capture sequencing using total RNA alone is suitable for detecting a range of variants relevant in chronic myeloid leukemia and may be more broadly applied to other hematological malignancies where diverse variant types define risk groups.

Biomarkers which better match anticancer drugs with cancer driver genes hold the promise of improved clinical responses and cure rates. We developed a precision medicine platform of rapid high‐throughput drug screening (HTS) and patient‐derived xenografting (PDX) of primary tumor tissue, and evaluated its potential for treatment identification among 56 consecutively enrolled high‐risk pediatric cancer patients, compared with conventional molecular genomics and transcriptomics. Drug hits were seen in the majority of HTS and PDX screens, which identified therapeutic options for 10 patients for whom no targetable molecular lesions could be found. Screens also provided orthogonal proof of drug efficacy suggested by molecular analyses and negative results for some molecular findings. We identified treatment options across the whole testing platform for 70 A precision diagnostic platform integrating genomics and transcriptomics with drug testing of patient's primary tumor cells in high throughput drug screening (HTS) and patient‐derived xenograft (PDX) was established to improve identification of therapies in high‐risk pediatric cancer patients. A precision diagnostic platform integrating genomics and transcriptomics with drug testing of patient's primary tumor cells in high throughput drug screening (HTS) and patient‐derived xenograft (PDX) was established to improve identification of therapies in high‐risk pediatric cancer patients.

Blast crisis of chronic myeloid leukemia is associated with poor survival and the accumulation of genomic lesions. Using whole-exome and/or RNA sequencing of patients at chronic phase (CP, n = 49), myeloid blast crisis (MBC, n = 19), and lymphoid blast crisis (LBC, n = 20), we found 25 focal gene deletions and 14 fusions in 24 patients in BC. Deletions predominated in LBC (83% of structural variants). Transcriptional analysis identified the upregulation of genes involved in V(D)J recombination, including RAG1/2 and DNTT in LBC. RAG recombination is a reported mediator of IKZF1 deletion. We investigated the extent of RAG-mediated genomic lesions in BC. Molecular hallmarks of RAG activity; DNTT-mediated nucleotide insertions and a RAG-binding motif at structural variants were exclusively found in patients with high RAG expression. Structural variants in 65% of patients in LBC displayed these hallmarks compared with only 5% in MBC. RAG-mediated events included focal deletion and novel fusion of genes associated with hematologic cancer: IKZF1, RUNX1, CDKN2A/B, and RB1. Importantly, 8/8 patients with elevated DNTT at CP diagnosis progressed to LBC by 12 months, potentially enabling early prediction of LBC. This work confirms the central mutagenic role of RAG in LBC and describes potential clinical utility in CML management.

Background We previously reported a high incidence of mutated cancer-related genes at CML diagnosis in selected chronic phase patients with a poor outcome compared to those with a good outcome. We also found a novel class of variant associated with the formation of the Ph chromosome comprising fusions and/or rearrangement of genes on the translocated chromosomes, with evidence of fragmentation, inversion, and imperfect sequence reassembly. These were termed 'Ph-associated events' and were more frequent in patients with poor outcome. However, the risk attributable to these mutational events at diagnosis has not been defined in unselected cohorts. Aim To assess the impact of genomic events in a cohort of consecutively treated patients at diagnosis of chronic phase CML. Methods A hybridization capture sequencing method targeting genes implicated in myeloid and lymphoid malignancies was applied to diagnostic RNA of patients enrolled in the TIDEL II trial. Patients were treated with upfront imatinib with active intervention, dose escalation or nilotinib switch, primarily for lack of time-dependent molecular milestones. Single base variants, small insertions/deletions, splice variants, gene fusions, and focal gene deletions were assessed with pre-defined criteria for pathogenicity. These were further classified as pathogenic mutations in cancer-related genes or Ph-associated events. Univariate and multivariate analyses were performed to evaluate the influence of mutational events and other key clinical and demographic variables on outcome at 4 years. Failure events were as defined by the ELN 2020 recommendations. Results 160/210 TIDEL II patients have so far been sequenced. 33 relevant mutations with variant allele frequencies ≥5% were identified in 9 genes in 25 patients (16%). ASXL1 was most frequently mutated (10% of all patients) and other recurrently mutated genes at diagnosis were RUNX1, BCORL1, IKZF1 and DNMT3A. Ph-associated events occurred in 25 patients (16%). Most of these (14/25 patients) involved fusions between genes on chromosomes 9 and 22 consistent with deletions adjacent to BCR and ABL1, or fusions between BCR or ABL1 and genes/regions on chromosomes other than 9 or 22. These were consistent with variant translocation and some were cytogenetically cryptic. Among these and other Ph-associated events were complex rearrangements involving inversions and large duplications. These were detected in 14/25 patients. Five patients had both cancer-related gene mutations and Ph-associated events, totalling 45 patients (28%) with at least 1 genomic event. Cancer-related mutations at diagnosis were associated with inferior progression-free survival (PFS) 82% vs 91% P=.03, and failure-free survival (FFS) 55% vs 83% P<.001. Molecular outcomes were also inferior: MMR 71% vs 89% P=.02; and MR4 27% vs 71%; P=.01. Similarly, Ph-associated events predicted inferior outcomes: FFS 63% vs 81% P=.01; MMR 70% vs 88% P=.01; and MR4 47% vs 68% P=.01. Importantly, patients with either or both of these mutational events had an increased likelihood of progression to accelerated phase or blast crisis or development of a BCR-ABL1 kinase domain mutation (Figure). Independent predictors of all survival and molecular outcomes were assessed with univariate and multivariate modelling (Table). Candidate prognostic variables were age at diagnosis, sex, transcript, Sokal and ELTS scores and the genomic variables. The only independent predictor of PFS was mutations in cancer-related genes. Cancer-related gene mutations, Ph-associated events and the ELTS score were independent predictors of FFS, MMR and MR4. We evaluated whether genomic data could be additive to the ELTS risk score. Low risk ELTS patients with any mutational event had inferior outcomes: FFS 76% vs 86%, P=.07; MMR 75% vs 92%, P=.02; MR4 41% vs 75%, P=0.006. Similar findings were observed in Intermediate risk ELTS patients: FFS 22% vs 91%, P<.001; MMR 81% vs 84%, P=.06; MR4 9% vs 72%, P=0.047. The number of high risk ELTS patients were inadequate to perform this analysis. Conclusion Despite a proactive strategy for TKI switch and a higher imatinib starting dose, the presence of cancer-related gene mutations or Ph-associated events conferred inferior outcomes. Combining the ELTS score with any mutational event further differentiated patient outcomes, demonstrating the power of integrating genomic data with current risk stratification. Disclosures Shanmuganathan: Janssen: Other: travel expenses; Novartis: Honoraria, Other: Travel expenses; Gilead: Other: Travel expenses; Amgen: Other: travel expenses. Hughes:BMS: Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding; Novartis: Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding. Branford:Bristol Myers Squibb: Honoraria; Cepheid: Honoraria, Membership on an entity's Board of Directors or advisory committees; Qiagen: Honoraria, Membership on an entity's Board of Directors or advisory committees; Novartis: Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding.

Background Epithelioid inflammatory myofibroblastic sarcoma (eIMS) is characterised by perinuclear ALK localisation, CD30 expression and early relapse despite crizotinib treatment. We aimed to identify therapies to prevent and/or treat ALK inhibitor resistance. Methods Malignant ascites, from an eIMS patient at diagnosis and following multiple relapses, were used to generate matched diagnosis and relapse xenografts. Results Xenografts were validated by confirmation of RANBP2-ALK rearrangement, perinuclear ALK localisation and CD30 expression. Although brentuximab-vedotin (BV) demonstrated single-agent activity, tumours regrew during BV therapy. BV resistance was associated with reduced CD30 expression and induction of ABCB1. BV resistance was reversed in vitro by tariquidar, but combination BV and tariquidar treatment only briefly slowed xenograft growth compared with BV alone. Combining BV with either crizotinib or ceritinib resulted in marked tumour shrinkage in both xenograft models, and resulted in prolonged tumour-free survival in the diagnosis compared with the relapse xenograft. Conclusions CD30 is a therapeutic target in eIMS. BV efficacy is limited by the rapid emergence of resistance. Prolonged survival with combination ALK and CD30-targeted-therapy in the diagnosis model provides the rationale to trial this combination in eIMS patients at diagnosis. This combination could also be considered for other CD30-positive, ALK -rearranged malignancies.

Sphingolipid dysregulation is often associated with insulin resistance, while the enzymes controlling sphingolipid metabolism are emerging as therapeutic targets for improving insulin sensitivity. We report herein that sphingosine kinase 2 (SphK2), a key enzyme in sphingolipid catabolism, plays a critical role in the regulation of hepatic insulin signaling and glucose homeostasis both in vitro and in vivo. Hepatocyte-specific Sphk2 knockout mice exhibit pronounced insulin resistance and glucose intolerance. Likewise, SphK2-deficient hepatocytes are resistant to insulin-induced activation of the phosphoinositide 3-kinase (PI3K)-Akt-FoxO1 pathway and elevated hepatic glucose production. Mechanistically, SphK2 deficiency leads to the accumulation of sphingosine that, in turn, suppresses hepatic insulin signaling by inhibiting PI3K activation in hepatocytes. Either reexpressing functional SphK2 or pharmacologically inhibiting sphingosine production restores insulin sensitivity in SphK2-deficient hepatocytes. In conclusion, the current study provides both experimental findings and mechanistic data showing that SphK2 and sphingosine in the liver are critical regulators of insulin sensitivity and glucose homeostasis.

Around 10% of acute leukemias harbor a rearrangement of the MLL/KMT2A gene, and the presence of this translocation results in a highly aggressive, therapy-resistant leukemia subtype with survival rates below 50%. There is a high unmet need to identify safer and more potent therapies for MLL-rearranged (MLL-r) leukemia that can be combined with established chemotherapeutics to decrease treatment-related toxicities. The curaxin, CBL0137, has demonstrated nongenotoxic anticancer and chemopotentiating effects in a number of preclinical cancer models and is currently in adult Phase I clinical trials for solid tumors and hematological malignancies. The aim of our study was to investigate whether CBL0137 has potential as a therapeutic and chemopotentiating compound in MLL-r leukemia through a comprehensive analysis of its efficacy in preclinical models of the disease. CBL0137 decreased the viability of a panel of MLL-r leukemia cell lines (n = 12) and xenograft cells derived from patients with MLL-r acute lymphoblastic leukemia (ALL, n = 3) in vitro with submicromolar IC50s. The small molecule drug was well-tolerated in vivo and significantly reduced leukemia burden in a subcutaneous MV4;11 MLL-r acute myeloid leukemia model and in patient-derived xenograft models of MLL-r ALL (n = 5). The in vivo efficacy of standard of care drugs used in remission induction for pediatric ALL was also potentiated by CBL0137. CBL0137 exerted its anticancer effect by trapping Facilitator of Chromatin Transcription (FACT) into chromatin, activating the p53 pathway and inducing an Interferon response. Our findings support further preclinical evaluation of CBL0137 as a new approach for the treatment of MLL-r leukemia.

Background Mutation of genes linked to hematologic cancer have recently been reported in CML and are associated with early progression and resistance (Reviewed in Branford, Kim Leuk 2019). The mutations comprise single nucleotide variants (SNVs) and small insertions/deletions (indels), plus gene fusions and large focal gene deletions. In 39 patients (pts) in blast crisis (BC), all had at least 1 cancer gene mutation, including fusions in 33%: partner genes MLL, RUNX1, IKZF1, MECOM and CBFB. 50% of the fusions were novel and some were present at chronic phase diagnosis. BCR-ABL1 mutations rarely occurred as the sole mutant. NGS offers critical information for resistance assessment. For many clinical purposes, targeted DNA sequencing (seq) using panels of specific disease related genes is the most cost effective screening choice. However, this strategy could miss relevant fusions and deletions. Aim To determine whether an RNA based approach is more informative than DNA for detecting a broad range of mutations. Method A hybridization capture NGS gene panel was developed to target 126 genes relevant for myeloid/lymphoid leukemia. In a pilot study, DNA and RNA derived from 5 leukemia cell lines with well characterized mutations, including fusions and deletions, were panel sequenced. An additional 6 cell lines were sequenced using RNA, plus 49 pt samples with RNA stored for up to 14.6 years: 45 at diagnosis and 4 at BC/resistance. Six of these pt samples had prior whole exome and/or whole transcriptome seq. We used total RNA that detected intronic splice region variants from pre-spliced RNA. SNVs/indels were called from DNA/RNA with FreeBayes. Manta called focal deletions from DNA. Known and novel RNA fusions and novel splice junctions were detected using the STAR aligner. Gene expression used edgeR. Results For the 5 cell lines with DNA versus (v) RNA seq, SNVs/indels were reliably called in RNA, with a strong positive correlation of mutant allele frequency: DNA v RNA, r = 0.93. Two TP53 small deletions of 26 and 46 bp were not called in RNA, but were instead detected as novel RNA splice junctions. Read counts were 5.2 fold higher for RNA than DNA at sites of clinically relevant mutants, consistent with enrichment of seq read depth proportional to expression. Overall, RNA revealed a higher number of relevant mutants than DNA: RNA = 49 v DNA = 37, Fig A-C. Notably, the functional effect of splice region disrupting mutants and large focal deletions were evident by novel RNA splicing, Fig D-F. In the total 11 cell lines tested with RNA, all 13 reported fusions were called, including BCR-ABL1 and RUNX1, MLL, ETV6 and CBFB fusions. For 7 cell lines with variants described in the COSMIC Cell Lines Project, 23/23 cancer gene SNVs/indels were called, plus 7 cancer gene SNVs/indels not reported. These were verified by DNA seq. 15 gene deletions were evident by atypical RNA splicing and verified by DNA seq: IKZF1, CDKN2A/B, PAX5, BTG1, RB1 and NCOR1. Five other cell lines had verified CDKN2A deletions that were evident by loss of gene expression, Fig G. Two BTG1 deletions were not detected. For the 6 pt samples re-sequenced by the RNA panel, 8/8 verified fusion transcripts were detected with a 31 fold enrichment of read counts. 11/11 cancer gene SNVs/indels were called and 3/4 gene deletions. The exception was a CDKN2A deletion not detected by novel splicing but evident as loss of expression, Fig G. Seven other cancer gene SNVs were found at low allele frequency, including a resistant BCR-ABL1 mutation at 1.7% in the oldest sample. Of the 43 diagnosis samples without prior NGS, BCR-ABL1 transcripts were detected in all. BCR-ABL1 genomic breakpoints were called at base pair resolution in 39, 91%. Two pts had mutated ASXL1 at diagnosis and both failed imatinib by 9 months with mutant BCR-ABL1. By gene expression analysis, all but 1 of the total 45 diagnosis samples clustered together. The exception was a pt who transformed to lymphoid BC at 6 months that clustered with the lymphoid cell lines and lymphoid BC pts, Fig H. Conclusion RNA gene panel seq demonstrated enhanced sensitivity and an increased yield of clinically relevant mutations compared with DNA panel seq. A single RNA assay has the capacity to detect SNV/indels, known and novel gene fusions, focal deletions and the likely functional effect of splice disrupting mutations. RNA panel seq is a valuable tool for the comprehensive assessment of mutations that drive CML treatment failure and drug resistance. Disclosures Branford: Novartis: Consultancy, Honoraria, Research Funding, Speakers Bureau; Bristol-Myers Squibb: Honoraria, Speakers Bureau; Qiagen: Consultancy, Honoraria; Cepheid: Consultancy, Honoraria. Shanmuganathan:Gilead: Other: Travel Support; Janssen: Other: Travel Support; Amgen: Other: Travel Support; Bristol-Myers Squibb: Honoraria, Other: Travel Support; Novartis: Honoraria, Other: Travel Support. Scott:Celgene: Honoraria. Hughes:Novartis, Bristol-Myers Squibb: Consultancy, Other: Travel; Novartis, Bristol-Myers Squibb, Celgene: Research Funding.

Myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML) are diseases caused by an ineffective myelopoiesis.AML is an aggressive blood cancer with poor prognosis for patients despite treatment with intensive chemotherapy, hence alternative therapies are needed.MDS and AML patients with low expression of the transcriptional repressor GFI1 (Growth factor independence 1) or expression of the GFI1 variant GFI1-36N (asparagine instead of serine at amino acid position 36) in their blast cells have an even poorer prognosis.GFI1 recruits amongst others HDAC1 and 2 (histone deacetylase 1 and 2) to its target genes.On a molecular level, expression of GFI1 thus leads to the removal of acetyl groups at H3K9.Low GFI1 (GFI1-KD) or GFI1-36N expression in MDS and AML blasts resulted in increased H3K9ac at its target genes, causing elevated target gene expression.Some of these target genes were oncogenes, explaining why GFI1-KD and GFI1-36N promote AML development.We hypothesized that administration of HAT (histone acetyltransferase) inhibitors could be beneficial for MDS/AML patients with reduced GFI1 or GFI1-36N expression as it could reverse the increased acetylation of H3K9.Curcumin is a HAT inhibitor which is used as a spice with so far no known toxic side effects.To study the effect of Curcumin on MDS/AML development, we crossed the well-established murine model of human MDS/AML, NUP98-HOXD13, with GFI1-WT, GFI1-KD or GFI1-36N mice.The different groups were treated with either curcumin or were left untreated.Curcumin effectively prevented the development of AML in mice with low GFI1 or GFI1-36N expression, but not in GFI1-WT mice.Overall, our data suggest that GFI1 functions as a prognostic marker in MDS/ AML patients, resulting in a personalized treatment approach using Curcumin in patients with low GFI1 or GFI1-36N expression.