Patients In Deep Molecular Response Research Articles

Single Cell Analysis of Genes Involved in Asymmetric Hematopoietic Stem Cell (HSC) Division in Chronic Myeloid Leukemia (CML) : Evidence of the Regulation of Tetraspanin CD63 during Leukemia Progression

Tyrosine kinase inhibitor (TKI) therapies have profoundly changed the natural history and survival of CML patients. However, the majority of the patients remain on long-lasting therapies and switch one TKI to another, either because of intolerance or resistance. All TKI cessation trials showed approximately 50% of leukemia recurrence rates even after several years of TKI therapies, suggesting the persistence of residual leukemic stem cells in patients in deep molecular response (DMR). The resistance and persistence of CML HSC are related to intrinsic (BCR-ABL independent survival) or extrinsic (leukemic niche ) mechanisms some of which have been extensively studied. However, there has been no data with regard to the role of asymmetric cell division pathways in the persistence and progression of CML stem cell clone. Indeed, During asymmetric division process of hematopoietic stem cells, some proteins are differently segregated between daughter HSCs such as endosome associated tetraspanin proteins (CD53, CD63) and mitochondria-associated fission regulator Dynamin related protein-1(DRP1). Similarly, it has been shown that asymmetric organelle inheritance can influence HSC decisions towards differentiation versus self-renewal. Interestingly, the tetraspanin CD63 is known to be involved in the quiescence of human and murine HSC via activation of TGF-beta pathway, CD63 hi cells exhibiting increased self-renewal potential after the asymmetric division with reduced differentiation potential. Materials and Methods We have analyzed the transcriptome performed in a cohort of CML patients in chronic phase (CP) and blast phase (BP) (GSE4170) as well as in a validation cohort of RNA-sequencing (GSE100026). Similarly, single cell transcriptome analysis of Lin- CD34+ CD38- CML progenitors (GSE76312) were included in the study evaluating the expression of genes involved in asymmetric cell division. We have then generated an asymmetric hematopoietic stem cell risk score using a panel of genes involved in the asymmetric division of HSCs and correlated these results with the transcriptome of the CML cohorts above. A pseudotime analysis was performed to evaluate single cell trajectories of the genes identified. Results To determine the potential role of asymmetric cell division markers in the progression of CML, we have analyzed the expression of several of these genes during CML progression in two independent transcriptome cohorts. The regulation concordance between the two distinct cohorts was principally observed on asymmetric molecules which were found to be down-regulated during blast crisis as compared to chronic phase. The regulation concordance between the two distinct cohorts was principally observed on asymmetric molecules down regulated during blast crisis as compared to chronic phase suggesting less of asymmetric properties during advanced phase of the disease. The down regulation in patients in blast crisis was confirmed at single cell level for the expression of SELL, CD63, NUMB, HK2 and LAMP2. CD63 expression has also been found to be of interest in CML chronic phase. Indeed, single cell neural network and deep learning in CP samples showed that the CD63 expression was the best asymmetric gene marker predicting a poor prognosis and this appeared to be induced by opposite regulation of HK2 and NUMB. In patients in CP-CML, the single cell trajectory reconstitution revealed the major regulatory role of CD63 highlighting a trajectory cluster implicating HSPB1, PIM2, ANXA5, LAMTOR1, CFL1, CD52, RAD52, MEIS1, PDIA3, all genes known to be involved also in hematopoietic malignancies such as myelodysplasia, AML and ALL. Conclusion We show here for the first time the identification of the expression of genes involved in the asymmetric division of CML HSC. Among these genes we identified the tetraspanin CD63 as being down-regulated during CML progression. The expression of the genes involved in the asymmetric HSC division should require further studies in CML, either to predict persistence of CML or progressive disease. Figure 1 : Single cell transcriptome of Lin- CD34+ CD38- CML progenitors in CP links CD63 expression to progressive disease. A/ Piechart of counts for chronic phase cells analyzed according to prognosis; B/ pseudotime transformation of single cell transcriptome according patient prognosis; C/ Pseudotime expression heatmap; D/ Pseudotime expression of CD63 in CP.

Age-Related Clonal Hematopoiesis Mutations Detected at the Time of Stopping Tyrosine Kinase Inhibitor Therapy Predict the Achievement of Treatment-Free Remission 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.

Minimal Residual Disease Detection at RNA and Leukemic Stem Cell (LSC) Levels: Comparison of RT-qPCR, d-PCR and CD26+ Stem Cell Measurements in Chronic Myeloid Leukemia (CML) Patients in Deep Molecular Response (DMR).

A Deep Molecular Response (DMR), defined as a BCR::ABL1 transcript at levels ≤ 0.01% by RT-qPCR, is the prerequisite for the successful interruption of treatment among patients with Chronic Myeloid Leukemia (CML). However, approximately 50% of patients in Treatment-Free Remission (TFR) studies had to resume therapy after their BCR::ABL1 transcript levels rose above the 0.1% threshold. To improve transcript detection sensitivity and accuracy, transcript levels can be analyzed using digital PCR (dPCR). dPCR increases BCR::ABL1 transcript detection sensitivity 10-100 fold; however, its ability to better select successful TFR patients remains unclear. Beyond the role of the immune system, relapses may be due to the presence of residual leukemic stem cells (LSCs) that are transcriptionally silent. Flow cytometry can be used to identify and quantify circulating bone marrow Ph+ LSCs CD34+/CD38- co-expressing CD26 (dipeptidylpeptidase-IV). To date, the significance of circulating Ph+ LSCs in TFR is unclear. The aim of this work is to compare and examine the values obtained using the three different methods of detecting minimal residual disease (MRD) in CML at RNA (RT-qPCR and dPCR) and LSC (flowcytometry) levels among patients in TFR or exhibiting a DMR. The twenty-seven patients enrolled received treatment with either imatinib (12), dasatinib (6), nilotinib (7), bosutinib (1), or interferon (1). Twelve patients were in TFR, while the rest exhibited a DMR. The TFR patients had stopped therapy for less than 1 year (3), <3 years (2), 6 years (6), and 17 years (1). Blood samples were collected and tested using the three methods at the same time. Both d-PCR and LSCs showed higher sensitivity than RT-qPCR, exhibiting positive results in samples that were undetectable using RT-qPCR (17/27). None of the patients tested negative with d-PCR; however, 23/27 were under the threshold of 0.468 copies/μL, corresponding to a stable DMR. The results were divided into quartiles, and the lowest quartiles defined the lowest MRD. These data were strongly correlated in 15/27 patients, corresponding to almost half of the TFR patients. Indeed, the TFR patients, some lasting up to 17 years, corresponded to the lowest detectable DMR categories. To the best of our knowledge, this is the first attempt to analyze and compare DMRs in a CML population using standard (RT-qPCR) and highly sensitive (dPCR and LSCs) methods.

Five-Year Follow-up of the Phase I/II Discontinuation Trial of Imatinib after Pioglitazone (EDI-PIO) in Chronic Myeloid Leukemia Patients in Deep Molecular Response

Treatment-free remission (TFR) is successful in approximately 40-60% of CML patients who achieve a stable and deep molecular response. Relapses may occur due to the persistence of quiescent leukemic stem cells (LSC). Pioglitazone, a drug used for diabetes, is a PPAR gamma agonist and reduces STAT5 activity, and its downstream targets HIF2α and CITED2, key guardians of the quiescent LSC. Residual LSC can be gradually purged from bone marrow niches by pioglitazone. Objectives: to evaluate the efficacy and safety of pioglitazone used with imatinib before treatment discontinuation. Methods: EDI-PIO (from the Portuguese Estudo de Descontinuação de Imatinibe após Pioglitazona) is a prospective, single arm, phase I/II discontinuation trial. Inclusion criteria: CML in chronic phase, treated with IM for at least 3 years, with MR4.5 reported by the International Scale (IS) for 2 years. Patients received pioglitazone 30 mg/day orally for three months before IM discontinuation. After discontinuation, BCR-ABL levels were measured by real-time quantitative PCR monthly for 12 months, every two months in the second year, and then every three months. IM was re-initiated at molecular relapse (single sample with a value of >0.1% or two consecutive samples >0.01%). TFR was calculated from IM discontinuation until molecular relapse, progression, or death to causes related to CML. Overall survival (OS) was calculated from IM discontinuation until the last seen or death date by any cause. This trial is registered with Clinicaltrials.gov, NCT02852486. Results: Between Jun/2016 and Jan/2019, we evaluated 32 CML patients, with a median age of 54 years (29-77), treated with imatinib for a median time of 9.5 years (3-16). The median duration of MR4 and MR4.5 was 106 and 93 months, respectively. The cut-off date for this analysis was July 1st, 2022. One patient left the trial before imatinib discontinuation and was not analyzed for TFR. There were no grade 3 or 4 adverse events related to pioglitazone. The median follow-up of the 31 patients that discontinued IM was 61 months (37-69). As previously reported, 15/31(48%) patients presented symptoms related to withdrawal syndrome. Twelve patients presented molecular relapse after a median time of 5 months (2-30). Nine relapses in the first six months and three at 7, 13, and 30 months after stopping imatinib. All relapsed patients re-achieved MMR in a median time of 3 months (1.8-4.1). One patient developed an anal canal adenocarcinoma in the third year after discontinuation and was treated with surgery and chemotherapy. TFR was 71%, 67%, 61% and 61% at 6,12, 30 and 60 months (Figure 1A), respectively. 60-month OS was 95% (CI 95%: 85-100%)(Figure 1B). There were 5 cases of COVID-19 among the 19 patients in TFR (26%) and two suspects. Four cases were mild, and one patient in MR4.5 died due to severe COVID-19. Sokal score and MR4.5 duration were significant factors for prolonged TFR (P=0.032 and 0.012, respectively). Conclusions: the combination of pioglitazone and IM was feasible and safe, with TFR rates consistent with previous discontinuation trials. The 5-year long-term follow-up demonstrated durable and stable molecular responses. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Discontinuation of Tyrosine Kinase Inhibitor Therapy and Treatment-Free Remission in Chronic-Phase Chronic Myeloid Leukemia: An Observational Retrospective Multi-Center Study

Introduction: Discontinuation of tyrosine kinase inhibitor (TKI) therapy and achievement of treatment-free remission (TFR) in chronic-phase chronic myeloid leukemia (CP-CML) patients is feasible and can be implemented in clinical practice. We conducted a retrospective study at 3 tertiary care centers in Riyadh, Saudi Arabia of all CP-CML patients who discontinued TKI therapy. Methods: All CP-CML patients being followed at participating centers who discontinued TKI, were eligible. TKI discontinuation was carried out as per clinical practice guidelines. Criterion to restart TKI therapy in relapsed patients was loss of major molecular response (MMR; BCR-ABL > 0.1). We collected data on gender, age, prognostic risk-score, type(s) of TKI therapy, type and duration of last TKI before discontinuation, total duration of TKI therapy and response, time to achieve MMR, relapse data including time to relapse and response to re-initiation of TKI therapy, and analyzed the risk factors for relapse. Study was approved by the institutional review boards of all participating centers. Results: Fifty-five patients (35 females) with a median age of 40 years (range 16-74) were included in the study. Median age at TKI stop was 48 years (range 21-80). Majority of the patients (48; 87.27%) received imatinib as first line therapy; 29 (53.74) patients were receiving imatinib at TKI-stop, and 26 (47.27) a second-generation TKI (dasatinib or nilotinib). Fifty-two (94.5%) patients discontinued TKI as per clinical practice (shared decision between treating physician and the patient), and 3 (5.5%) stopped because of pregnancy; 2 patients fulfilled the criteria to stop post-delivery as they remained in deep MR, they were offered to continue stopping. One patient refused to restart TKI (dasatinib) after delivery although she had only 24 months of deep response but continues to have deep remission after 91 months of follow up. Median time to achieve MMR after diagnosis was 12 (range 2-164) months. All the patients had achieved MR4.5 or better (complete molecular response) before stop. Median time from diagnosis to TKI stop was 86 months (IQR 60;132) and median time of last TKI before stop was 66 months (IQR 47;114). After a median follow up of 42 months (IQR 12.5;69) post TKI-discontinuation, 15 (28%) patients relapsed. Median time to relapse was 5 months (range 3-38). Most of the relapses occurred during the first 6 months, except 3 patients who relapsed after 10, 11, and 38 months. Most of the relapsed patients had intermediate or high risk Sokal score. All the relapsed patients achieved MMR after a median of 3 months (range 2-6) after restarting TKI. None of the patients progressed to advanced phase. Conclusion: In real-life practice, TKI discontinuation was possible without relapse for CP-CML patients in deep molecular response, with over two-third of patients continuing in TFR. Majority of the relapsed patients belonged to the intermediate/high-risk score Sokal category at diagnosis. Although this study has the limitation of a retrospective study, our experience in this population confirms that discontinuation of TKI therapy in CP-CML patients is feasible and safe in clinical practice.

CML-428 Real World Data of Treatment Free Remission (TFR) in Patients With Chronic Myeloid Leukemia (CML) Treated at a Tertiary Care Oncology Center in India

TFR has emerged as a new treatment goal in treatment of CML. To describe the outcomes of CML patients undergoing treatment free remission (TFR) in clinical practice. Retrospective analysis of CML patients treated from January 2018 to December 2021 who underwent TFR. Tertiary care cancer center Patients: 38 patients who opted for TFR after counselling or tyrosine kinase inhibitor (TKI) stopped due to various reasons in deep molecular response (DMR) were included. 92% were treated with imatinib at diagnosis. Median duration of TKI treatment was 135 months (6-242). 97.5% achieved DMR on TKI and median time from TKI initiation to DMR was 96 months (12-228). Median duration of DMR prior to TKI discontinuation was 41 months (0-67). TKI was discontinued after counselling for TFR in 26 patients (68%) while it was discontinued due to intolerance in 10 patients (29%). Data was analyzed for molecular relapse (MR), survival without molecular relapse (SWMR), TFR duration, and factors affecting MR. At median follow up of 25 months, nine patients (23.7%) had molecular relapse and three patients died. Median SWMR was not reached, and 2-year estimated SWMR was 65.2% (95%CI,47.2-83.2). After censoring for the competing events of death, 2-year cumulative molecular relapse was 26.8% (95%CI, 10.5-43). Median time to MR was 4 months. Of all relapses, 55.5% occurred in the first six months of TFR. Nine patients (23.7%) were restarted on TKI. Two patients with MR were not restarted on TKI as one patient died immediately after MR and another patient was not keen on restarting TKI despite repeated counselling. Median TFR duration was not reached, and 2-year estimated TFR was 73.4% (95%CI, 58.3-88.5). After restart of TKI, median duration to achieve MMR was 6 months (range, 1-17) in patients with molecular relapse. At the time of analysis, molecular disease statuses were DMR, MMR and less than MMR in 23, 10 and 2 patients respectively. TKI can be safely discontinued in real life practice in chronic phase CML patients in DMR and comparable TFR outcomes can be achieved. Grant acknowledgement: None.

Immunomodulatory Effects of IFNα on T and NK Cells in Chronic Myeloid Leukemia Patients in Deep Molecular Response Preparing for Treatment Discontinuation

A deep and stable molecular response (DMR) is a prerequisite for a successful treatment-free remission (TFR) in chronic myeloid leukemia (CML). In order to better identify and analyze potential candidates of successful TFR, we examined the phenotypic and functional host immune compartment in DMR patients who had received TKI treatment only (TKI-only) or had been previously treated with interferon-alpha (IFNα + TKI) or had received IFNα treatment only (IFNα-only). The T/NK-cell subset distribution, NK- and T-cell cytokine production, activation and maturation markers were measured in 44 patients in DMR treated with IFNα only (9), with IFNα + TKI (11) and with TKI-only (24). IFNα + TKI and TKI-only groups were eligible to TKI discontinuation according to the NCCN and ESMO guidelines (stable MR4 for more than two years). In IFNα-treated patients, we documented an increased number of lymphocytes capable of producing IFNγ and TNFα compared to the TKI-only group. In INFα + TKI patients, the percentage of NKG2C expression and its mean fluorescence intensity were significantly higher compared to the TKI-only group and to the INFα-only group in the CD56dim/CD16+ NK cell subsets (INFα + TKI vs. TKI-only p = 0.041, p = 0.037; INFα + TKI vs. INFα-only p = 0.03, p = 0.033, respectively). Furthermore, in INFα-only treated patients, we observed an increase of NKp46 MFI in the CD56bright/CD16- NK cell subset that becomes significant compared to the INFα + TKI group (p = 0.008). Our data indicate that a previous exposure to IFNα substantially and persistently modified the immune system of CML patients in memory T lymphocytes, differentiated NKG2C+ “long-lived” NK cells responses, even years after the last IFNα contact.

Digital PCR can provide improved BCR-ABL1 detection in chronic myeloid leukemia patients in deep molecular response and sensitivity of standard quantitative methods using EAC assays

BCR-ABL1 molecular detection using quantitative PCR (qPCR) methods is the golden standard of chronic myeloid leukemia (CML) monitoring. However, due to variable sensitivity of qPCR assays across laboratories, alternative methods are tested. Digital PCR (dPCR) has been suggested as a robust and reproducible option. Here we present a comparison of droplet dPCR with routinely used reverse-transcription qPCR (RT-qPCR) and automated GeneXpert systems. Detection limit of dPCR was above 3 BCR-ABL1 copies, although due to background amplification the resulting sensitivity was 0.01% BCR-ABL1 (MR4.0). Nevertheless, in comparison with GeneXpert, dPCR categorized more than 50% of the patients into different MR groups, showing a potential for improved BCR-ABL1 detection.

Prospective evaluation of variables affecting platelet function in patients with newly diagnosed chronic myeloid leukemia.

: Platelet function in chronic myeloid leukemia (CML) could be affected by either hyperleucocytosis, clonal megakaryopoiesis, or tyrosine kinase inhibitors. However, these variables have never been prospectively evaluated. We conducted a prospective study over a period of 1.5 years in a tertiary care center of north India. Patients with CML in chronic phase, more than 18 years, and treated with imatinib were enrolled (n = 32). Age, and sex-matched controls were also included. Platelet function test was performed using two-channel Chrono-Log aggregometer 490 at four time-points: first, at diagnosis; second, after leucoreduction (total leucocyte count, <10 × 10/l) achieved with hydroxycarbamide; third, on-imatinib at BCR-ABL less than 1%; and fourth, in an independent cohort (off-imatinib) at deep molecular response (DMR) (BCR-ABL < 0.01%). Statistical analysis was performed using IBM SPSS statistics (version 22.0). Median age of patients was 42 years (15-65), and M : F ratio was 1 : 1. At diagnosis, platelet function correlated negatively with total leucocyte count, but not with platelet count. As compared with baseline, platelet aggregation with ADP (2.5 μl), and collagen (2.5 μl) improved significantly after leucoreduction (P = 0.05 and 0.009, respectively). Imatinib further caused significant impairment of aggregation with ADP (2.5 μl), collagen (2.5 μl), and collagen (1 μl) (P = 0.04, 0.008, and 0.02, respectively). Patients in DMR also demonstrated a significant impairment of platelet aggregation with all the agonists as compared with controls. While leucoreduction alone can improve the baseline platelet function derangement in CML, imatinib further impairs it. Residual CML stem cells, or effect of imatinib on normal common myeloid progenitors might account for platelet function derangement at DMR.

Digital PCR for BCR‐ABL1 Quantification in CML: Current Applications in Clinical Practice

Molecular monitoring of the BCR-ABL1 transcript for patients with chronic phase chronic myeloid leukemia (CML) has become increasingly demanding. Real-time quantitative PCR (qPCR) is the routinely used method, but has limitations in quantification accuracy due to its inherent technical variation. Treatment recommendations rely on specific BCR-ABL1 values set at timed response milestones, making precise measurement of BCR-ABL1 a requisite. Furthermore, the sensitivity of qPCR may be insufficient to reliably quantify low levels of residual BCR-ABL1 in patients in deep molecular response (DMR) who could qualify for an attempt to discontinue Tyrosine Kinase Inhibitor (TKI) therapy. We reviewed the current use of digital PCR (dPCR) as a promising alternative for response monitoring in CML. dPCR offers an absolute BCR-ABL1 quantification at various disease levels with remarkable precision and a clinical sensitivity reaching down to at least MR5.0. Moreover, dPCR has been validated in multiple studies as prognostic marker for successful TKI treatment discontinuation, while this could not be achieved using classical qPCR. dPCR may thus prospectively be the preferred method to reliably identify patients achieving treatment milestones after initiation of TKI therapy as well as for the selection and timing for TKI discontinuation.

Point: Is there a best duration of deep molecular response to achieve therapy-free remission in chronic myeloid leukaemia?

Point: Is there a best duration of deep molecular response to achieve therapy-free remission in chronic myeloid leukaemia?

High Level of Successful TKI Discontinuation in Chronic Myeloid Leukemia (CML) Patients: Preliminary Results of AST-Argentina Stop Trial

Introduction: Treatment-free remission (TFR) is an emerging treatment goal for chronic myeloid leukemia (CML) patients in deep molecular response (DMR). Current evidence shows that 40%-60% of patients relapse while in TFR; and nearly all regain response once tyrosine kinase inhibitors (TKIs) treatment are reinitiated. However a robust predictor of prolonged TFR has not been reported yet. Considering real-life setting, 2 key factors may affect TFR outcome if not properly done: Access to serial molecular monitoring at optimal timepoints and quality laboratory terms as accuracy, sensitivity and rapid results. This motivated the creation of the AST study in our region to guarantee adequate molecular monitoring for TFR in Argentina and characterize new prognostic biomarkers helpful to identify more accurately patients who will be able to sustain TFR. We aimed to assess the proportion of patients with sustained major molecular response (MMR) after TKIs discontinuation and define precise conditions for stopping treatment. Methods: This prospective, multicentre Argentina Stop Trial (AST) trial is recruiting chronic phase CML patients under TKI treatment for at least ≥ 4 years, in DMR (≥MR4.0) sustained for ≥ 2 years in standardized laboratory, confirmed typical BCR-ABL1 transcripts b3a2 and/or b2a2 and aged &amp;gt; 18 years. Molecular tests are centralized in 2 harmonized laboratories and performed monthly for the first 6 months, every 2 months until the first year, and every 3 months during the second year. If patients lost MMR, TKI was restarted immediately. Molecular relapse Free Survival was estimated by Kaplan-Meier method. Difference between survival variables was evaluated through log-rank test. Multivariate analysis was performed through Cox proportional hazards model. The cutoffs of the numerical variables were considered according to the log-rank test. Results: Between February 2019 and July 2020, we evaluated 50 CML patients of whom 46 were enrolled from 7 centers in Argentina and 4 were screening failures. Recruitment was interrupted due to COVID-19 pandemic. Patient median age was 57.5 years (range 24-85). Before discontinuation, TKI treatment was as follows: Imatinib 37/46 (80%), Nilotinib 5/46 (11%) and Dasatinib 4/46 (9%), 2G-TKI as 1st line, 11% of the patients received non-branded treatment. Sokal risk score showed to be low in 22 patients (48%), intermediate in 14 (30%) and high in 10 (22%). Median follow-up was 10 months (range 4-17) and the estimated molecular relapse-free survival was 80.2% (95%CI 69-93) at 6 months Fig 1. Longer DMR durations before discontinuation were associated with increased probability of maintaining response at 6 and 12 months: 83.2% for patients who had &amp;gt;54 months in DMR vs 70% with &amp;lt;54 months and 72% vs 23.3% respectively (p=0.0453) Fig 2. Cox multivariate analysis was performed including different variables as age at diagnosis, time in DMR, time in TKI previous to discontinuation and Sokal risk. The only significant variable associated to improved prognosis was time in DMR (HR 2.8 95%CI 1.002-8.07 p=0.0495). Our cohort had a long time on TKI treatment previous to discontinuation, median 10.5 years (4.16-17.5) probably considering it a favorable factor for the high TFR rates described at 6 months. Among the 46 patients included, 15 (33%) lost MMR, all restarted treatment with the same TKI used before discontinuation, 12/15 (80%) regained MMR with a median time of 3 months (range1-8) and 9/15(60%) obtained MR 4.0 with a median time of 3 months (range1-5). Conclusion: This is the first multicenter study of TKI discontinuation in CML patients in Argentina showing that TKI can be safely discontinued in those who achieve and maintain a DMR before discontinuation. We observed high rates of molecular relapse free survival, although longer follow-up is needed. We must continue with this approach for patients participating in TFR trials or TFR programs in order to decrease the risk of relapse and make this goal a fact in our region. This discontinuation study will allow in a near future significant saving of economic resources and might improve patients quality of life specially in those who are currently experiencing treatment adverse events. Disclosures Pavlovsky: Novartis: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; BMS: Speakers Bureau; Pfizer: Speakers Bureau; Pint Pharma: Speakers Bureau. Varela:Novartis: Consultancy, Speakers Bureau. Pavlovsky:Janssen: Membership on an entity's Board of Directors or advisory committees, Other: travel grants, Speakers Bureau; Abbvie: Membership on an entity's Board of Directors or advisory committees, Other: Travel grants, Speakers Bureau; Astra Zeneca: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Varifarma: Speakers Bureau. Moiraghi:BMS: Speakers Bureau; Novartis: Speakers Bureau.

Risk of molecular recurrence after tyrosine kinase inhibitor discontinuation in chronic myeloid leukaemia patients: a systematic review of literature with a meta-analysis of studies over the last ten years.

More than 10years ago, the first pilot observational study of imatinib discontinuation was reported in chronic myeloid leukaemia (CML) patients in deep molecular response (DMR). Several studies have been published since then, in patients treated with frontline imatinib, or second-generation tyrosine kinase inhibitors (TKI) in first or second line but also on second attempt of TKI discontinuation. Our objective was to estimate, through meta-analyses of the literature data, the probability of molecular recurrence (MolRec) in the time periods of 0-6, 6-12, 12-18 and 18-24months after a first and second TKI discontinuation and the probability of re-acquisition of DMR after MolRec. The Medline and Scopus databases were searched up to April 2019. The studies were selected by three independent reviewers. Random-effect meta-analyses were conducted using the MetaXL software. The probability of MolRec in the time periods 0-6, 6-12, 12-18 and 18-24months after the first attempt was respectively 35%, 8%, 3% and 3%, whereas the probability of MolRec in the time periods 0-6, 6-12 and 12-18after the second attempt was 48%, 27% and 12% respectively. Re-acquisition of a DMR was observed in 90% of patients. Most of the MolRec occur during the first six months in case of a first attempt, whereas the second MolRec occurs over a larger window of time.

Accumulation of DNA Damage and Alteration of the DNA Damage Response in Chronic Myeloid Leukemia

The accumulation of DNA damage and the alteration of the DNA damage response (DDR) are critical features of genetic instability that is presumed to be implicated in BCR/ABL1-mediated blastic transformation of chronic myeloid leukemia (CML). The aim of our study was to analyze underlying mechanisms of genetic instability with regard to DNA damage such as DNA double-strand breaks (DSB), DSB repair and DDR signaling during blastic transformation of CML. Immunofluorescence microscopy of γH2AX was performed for quantification of DSB in peripheral blood mononuclear cells (PBMC) of 8 healthy individuals, 24 chronic phase (CP)-CML patients under current/discontinued tyrosine kinase inhibitor (TKI) treatment (21 patients in deep molecular response (DMR), 3 patients in major molecular response (MMR)), 5 CP-CML patients under current/discontinued TKI treatment with loss of MMR, 3 de novo non-treated CP-CML patients and 2 blast phase (BP)-CML patients. In addition, immunofluorescence microscopy of γH2AX/53BP1 was used for semi-quantification of error-prone DSB repair. Furthermore, immunoblotting of p-ATM, p-ATR, p-CHK1, p-CHK2 and p-TP53 was performed in PBMC of CML patients in comparison to PBMC of healthy individuals. Our analysis revealed an increase in numbers of γH2AX foci in PBMC of CP-CML patients under current/discontinued TKI treatment with loss of MMR (1.8 γH2AX foci per PBMC ± 0.4), in PBMC of de novo non-treated CP-CML patients (2.3 γH2AX foci per PBMC ± 0.7) and in PBMC of BP-CML patients (4.9 γH2AX foci per PBMC ± 0.9) as compared to the number of γH2AX foci in PBMC of healthy individuals (1.0 γH2AX foci per PBMC ± 0.1) and in PBMC of CP-CML patients under current/discontinued TKI treatment in DMR/MMR (1.0 γH2AX foci per PBMC ± 0.1) (Figure 1A and B). Analysis of co-localizing γH2AX/53BP1 foci in PBMC suggested progressive activation of error-prone nonhomologous end-joining repair mechanisms during blastic transformation in CML. Signatures of p-ATM, p-ATR, p-CHK1, p-CHK2 and p-TP53 indicated alterations of the DDR. In summary, our data provide evidence for an accumulation of DNA damage in PBMC of CML patients towards BP-CML patients. We hypothesize that ongoing DSB generation, error-prone DSB repair and DDR alterations might be critical mechanisms of blastic transformation in CML. Figure 1 Analysis of γH2AX foci in freshly isolated peripheral blood mononuclear cells (PBMC) of healthy individuals and chronic myeloid leukemia (CML) patients. (A) Exemplary immunofluorescence microscopic images of γH2AX foci (green, Alexa 488) and cell nuclei (blue, DAPI) in PBMC of a healthy individual (HEALTHY#3), a chronic phase CML patient with a deep molecular response to tyrosine kinase inhibitor (CP-CML DMR#16), a de novo non-treated chronic phase CML patient (CP-CML#1) and a blast phase CML patient (BP-CML#2). (B) γH2AX foci levels in PBMC of healthy individuals and in PBMC of CML patients. Figure 1 Disclosures Saussele: Pfizer: Honoraria; Novartis: Honoraria, Research Funding; Incyte: Honoraria, Research Funding; BMS: Honoraria, Research Funding. Fabarius:Novartis: Research Funding.

PS1172 DDPCR AT THE TIME OF TKI DISCONTINUATION IN CHRONIC PHASE CHRONIC MYELOID PATIENTS IS PREDICTIVE OF TREATMENT‐FREE REMISSION OUTCOME

Background:Only few prognostic factors ‐ the median duration of treatment or of deep molecular response (DMR) ‐ have so far been associated to a positive outcome after TKI discontinuation in chronic myeloid leukemia (CML) patients. Thus, the identification of candidates that may impact on this primary treatment endpoint remains elusive. Digital droplet PCR (ddPCR) is an accurate molecular tool that could prove advantageous for the quantification of the BCR‐ABL1 fusion transcript copy number compared to classic RQ‐PCR and could help in better identify patients candidate for TKI discontinuation.Aims:To evaluate the predictive role of ddPCR in CML patients in sustained DMR tested at the time of TKI treatment discontinuation.Methods:We studied 50 RNA samples of CML patients who discontinued treatment while in RQ‐PCR undetectable DMR (MR4–4.5) according to IS for more than 2 years. The ddPCR assay was performed in triplicate using the Biorad‐QX200 platform. To test the BCR‐ABL fusion transcript and ABL control gene, we used the same primers and probes used in the standard RQ‐PCR reaction. Each reaction mixture was partitioned into approximately 20,000 droplets and then amplified. Cycled droplets were read in the QX200 droplet‐reader and analysis of the ddPCR data was performed using the QuantaSoft analysis software (Biorad Laboratories GmbH, Munich, Germany). We collected the clinical follow‐up data from the patients’ medical history and correlated them with ddPCR positivity or negativity at the time of discontinuation using the Wilcoxon signed‐rank test.Results:Fifty patients discontinued treatment in DMR after a median duration of response of 2.5 years (range 1.5–7.2). The median age was 59 years (range 32–87.4), 33 were female. According to the Sokal risk at baseline, 25 patients were classified as low risk, 21 as intermediate and 4 as high risk. Forty‐two patients received front‐line imatinib and 8 were treated with second generation TKIs. The median duration of treatment before discontinuation was 11.9 years (range 3.3–17.9). At the time of discontinuation, 28 patients (56%) were ddPCR‐negative and 24 of them (86%) remained in treatment‐free remission (TFR), while 4 patients (14%) relapsed after a median time of 5.5 months (range 4.0–27.0). Twenty‐two patients (44%) were ddPCR positive at the time of discontinuation: 12 of them (55%) remained in TFR after a median time of observation of 13.2 months, while 10 patients (45%) relapsed after a median of 3.5 months (range 1.0–16.0). The difference in TFR rate between the two groups is statistically significant (p = 0.026, Figure 1). We compared the characteristics of ddPCR‐negative versus ddPCR‐positive patients and failed to detect differences in the type of transcript (p = 0.155), Sokal risk score (p = 0.684), previous IFN treatment (p = 0.70), type of TKI received (p = 0.250), median duration of TKI treatment (10.9 years in positive versus 13.1 years in negative patients, p = 0.639) and median duration of DMR (2.4 years in positive versus 2.8 years in negative patients, p = 0.134).Summary/Conclusion:These results suggest that by ddPCR it is possible to predict the TFR outcome of CML patients in DMR (MR4–4.5) at the time of TKI discontinuation. This method, compared to RQ‐PCR, appears to be superior in terms of quantification of low levels of residual disease in patients who have already achieved a DMR and may successfully identify candidates to treatment discontinuation.image

PF414 DNA‐BASED BCR‐ABL1 ANALYSIS PROVIDES A “TRAFFIC LIGHT” STRATIFICATION MODEL FOR CHRONIC MYELOID LEUKEMIA WITH IMPLICATIONS FOR TREATMENT FREE REMISSION

Background:Treatment free remission (TFR) or molecular relapse free survival (MRFS) have become important therapeutic aims for patients with chronic myeloid leukemia (CML). This aim can be realised through achievement of a sustained deep molecular response (DMR).Aims:In our previous work (Machova Polakova K, EHA 2018) we showed that a BCR‐ABL1 DNA‐based approach for residual disease monitoring may be superior to mRNA analyses in a significant number of peripheral blood samples. In this collaborative EUTOS study we aimed to assess whether DNA PCR provided better prediction of TFR.Methods:Consecutive samples before and after TKI cessation were available for 54 CML patients, of whom 16 were enrolled in the EURO‐SKI study. Thirty‐five patients, who fulfilled criteria but could not be involved in the EURO‐SKI trial, gave consent for an intermittent regimen (INTReg) of TKI administration every second month. The MRFS rate of INTReg patients was comparable to EURO‐SKI and other TKI stopping trials in later time points (39% at 20 months and 32% at 70 months after TKI cessation). Patient‐specific genomic BCR‐ABL1 fusions were characterized by next generation sequencing and measurable residual disease was compared using DNA and mRNA based digital droplet (dd) PCR. The sensitivity of standard DNA and mRNA analysis of peripheral blood samples were comparable, varying between MR4‐MR5.Results:DNA and mRNA BCR‐ABL1 levels from consecutive analysis were available for 33/54 patients (altogether 945 samples; median 30 samples per patient; range 11–44). Analysis of these 33 cases enabled a “traffic light” stratification according to BCR‐ABL1 DNA and mRNA status before TKI cessation (Figure 1); “Green” group ‐ continually double negative (n = 4), “Yellow” group ‐ continually DNApos/RNAneg (n = 13) and “Red” group ‐ double positive (n = 16). All 4 patients from the Green group sustained in MRFS with continuous double negative status. Two of these patients gave consent for bone marrow (BM) aspiration at the time of TKI cessation. Both BM samples were double negative when analysed by ddPCR with enhanced sensitivity of MR6‐MR6.5 by performing multiple replicates. Molecular relapse (loss of MMR) occurred in 12/16 patients of the Red group. In the Yellow group, 8/13 patients survive without molecular relapse after TKI cessation (median 39.8 months; range 19.1–82.2); in 5/8 patients, the BCR‐ABL1 expression became detectable, however, only at DMR levels, i.e. MR4 or lower. BCR‐ABL1 expression remained undetectable in 3/8 patients after TKI cessation (median 27.3 months; range 19.1–53.6). BM was collected in 2/3 patients at the time of TKI cessation; in both cases, DNA assays detected residual CML cells despite the absence of BCR‐ABL1 expression.Summary/Conclusion:BCR‐ABL1 DNA analysis allows clinically relevant stratification of CML patients in DMR before complete or intermittent treatment cessation. Our approach identifies truly BCR‐ABL1 negative patients that are likely to benefit from TKI cessation. A biologically interesting Yellow group represents patients with detectable CML cells but with undetectable BCR‐ABL1 expression on TKI therapy. Although BCR‐ABL1 expression became detectable after TKI cessation in many of these cases, a high proportion (62%) achieved MRFS. Molecular relapse occurred in a high proportion (76%) of double positive patients. Currently we are validating the “traffic light” stratification scheme on the remaining 21 patients from our cohort.imageSupport: EUTOS2018, MZCR 00023736, AZV 16–30186A

Results of the European survey on the assessment of deep molecular response in chronic phase CML patients during tyrosine kinase inhibitor therapy (EUREKA registry).

The advent of tyrosine kinase inhibitor (TKI) therapies has revolutionized the treatment of chronic myeloid leukemia (CML). The European LeukemiaNet (ELN) recommends quantification of BCR-ABL1 transcripts by real-time quantitative PCR every 3months during TKI treatment. Since a proportion of patients in deep molecular response (DMR: MR4, MR4.5, MR5) maintain remission after treatment stop, assessment of DMR is crucial. However, systematically collected molecular data, monitored with sensitive standardized assays, are not available outside clinical trials. Data were collected on the standardized assessment of molecular response in the context of real-life practice. BCR-ABL1 transcript levels after > 2years of TKI therapy were evaluated for DMRby local laboratories as well as standardized EUTOS laboratories. Since standardized molecular monitoring is a prerequisite for treatment discontinuation, central surveillance of the performance of the participating laboratories was carried out. Between 2014 and 2017, 3377 peripheral blood samples from 1117 CML patients were shipped to 11 standardized reference laboratories in six European countries. BCR-ABL1 transcript types were b3a2 (41.63%), b2a2 (29.99%), b2a2/b3a2 (3.58%) and atypical (0.54%). For 23.72% of the patients, the initial transcript type had not been reported. Response levels (EUTOS laboratory) were: no MMR, n = 197 (6.51%); MMR, n = 496 (16.40%); MR4, n = 685 (22.64%); MR4.5, n = 937 (30.98%); MR5, n = 710 (23.47%). With a Cohen's kappa coefficient of 0.708, a substantial agreement between EUTOS-certified and local laboratories was shown. Multicenter DMR assessment is feasible in the context of real-life clinical practice in Europe. Information on the BCR-ABL1 transcript type at diagnosis is crucial to accurately monitor patients' molecular response during or after TKI therapy.

Minimal Residual Disease Detection at RNA and Leukemic Stem Cell (LSC) Level. Comparison of Qpcr, d-PCR and CD26 Stem Cell Measurements in Chronic Myeloid Leukemia (CML) Patients in Deep Molecular Response (DMR)

Background. A Deep Molecular Response (DMR), defined as BCR-ABL1 p210 transcript at levels <= 0.01% by QPCR, is the prerequisite for a successful interruption of treatment in patients with CML. However, approximately 50% of patients in Treatment Free Remission (TFR) studies had to resume therapy after BCR-ABL1 p210 transcript levels rose above the 0.01% DMR threshold. To improve transcript detection sensitivity, transcript levels were analyzed using D-PCR. D-PCR increased BCR-ABL1 transcript detection sensitivity by 10-100 fold, however its ability to better select successful TFR patients remains unclear. Beyond the role of the immune system, relapse may be due to the presence of residual leukemic stem cells (LSCs) that are transcriptionally low or silent, insensitivity to tyrosine kinase inhibitors (TKIs) or other CML-causing transcripts not easily detectable by BCR-ABL1 PCR techniques Another approach is to use flow cytometry to detect and quantify bone marrow Ph+ LSCs CD34+/CD38− that co-express CD26 (dipeptidylpeptidase-IV) although its meaning in TFR is unclear.Aim. To compare and examine values of the three different methods of detecting minimal residual disease (MRD) in CML at RNA and LSC levels in patients in TFR or DMR.Patients and Methods. The twenty-seven patients in this study received treatment with either Imatinib (12), Dasatinib (6), Nilotinib (7), Bosutinib (1) or Interferon (1). Twelve patients were in TFR, while the rest were in DMR. TFR patients had stopped therapy for less than 1 year (3), <3 years (2), 6 years (6) and 17 years (1). Blood samples were collected and tested 3 times. BCR-ABL1 transcript quantification was performed using the automated Xpert Ultra BCR-ABL1 MonitorTM Cepheid method and calibrated for the ABL reference gene and BCR-ABL1 target gene. Samples were analyzed according to the manufacurer with results expressed in BCR-ABL1/ABL %IS, or undetectable transcript (U), with PCR sensitivity of 5.0 (>250,000 ABL copy numbers).D-PCR analysis used TaqMan-MGB probes targeting the BCR-ABL1 transcript. A custom assay was designed and produced with a FAM-label, basing on the sequence of routinely-used probes. BCR-ABL1 quantifications were performed analyzing 50ng of cDNA on a QuantStudio 3D Digital PCR System (ThermoFisher Scientific). BCR-ABL1 transcript values using dPCR were expressed as number of copies/ul. The secondary analysis were carried out by AnalysisSuite Cloud Software (ThermoFisher Scientific).To detect peripheral blood circulating CD34+/CD38-/CD26+ LSCs, cells were incubated with anti-CD45 (BD Biosciences), anti-CD34 (581), anti-CD38 (HIT2) and anti-CD26 (M-A261) (BD Pharmigen). Acquisition and analysis were performed by FACSCanto II flow cytometer (BD Biosciences, NR Nannini). CD45+ cells acquired for each sample ranged from 500,000 to 1,000,000. Isotype controls were included in each staining. Median absolute number of CD26+ cells/μL were calculated as follows: (# WBCs/μL) × (% CD34+/CD38−/CD26+ stained CD45+ cells).Results. TFR status, the type of TKI treatment, QPCR results, dPCR data and LSCs quantification are summarized in Table 1.Both d-PCR and LSCs showed higher sensitivity than QPCR, exhibiting positive results when transcript levels using QPCR were undetectable (16). None of the patients tested negative with d-PCR, however 14/16 were under the threshold of 0.468 copies/uL, corresponding to a stable DMR. LSC levels were negative in 5 patients, 3 of which also tested negative with QPCR. In all patients QPCR ranged from 0-0.0068, d-PCR from 0.073 to 0.943, and LSCs from 0 to 0.156.Results were divided in quartiles, depending on molecular response for QPCR, the 0.468 threshold for dPCR and the distribution for the LSCs. The 2 lowest quartiles defined the lowest detectable DMR. A strong correlation of these data in TFR patients was noted (10/12 concordant) while 8/15 DMR patients were discordant.Conclusions To our knowledge this is the first attempt to analyze and compare DMR in a CML population using standard (QPCR) and highly sensitive (dPCR and LSCs) methods. TFR patients, some lasting up to 17 years, were in the lowest detectable DMR categories. Very little is known about the biology of CML implications of circulating LSCs. Larger studies and dynamic scoring will help define their informative and predictive value. [Display omitted] DisclosuresAbruzzese:Pfizer: Consultancy; Ariad: Consultancy; Novartis: Research Funding; BMS: Consultancy.

Feasibility of treatment discontinuation in chronic myeloid leukemia in clinical practice: results from a nationwide series of 236 patients

Over half of chronic myeloid leukemia (CML) patients in deep molecular response do not lose the major molecular response (MMR) after stopping treatment with tyrosine kinase inhibitors (TKI). This strategy is safe in clinical trials, but its applicability in the real-life setting remains unsettled. We describe the outcomes after TKI discontinuation in a nationwide series of 236 CML patients. Median follow-up from treatment discontinuation was 21.5 months and 5 patients died from CML-unrelated causes. TKI therapy was reinitiated due to MMR loss (n = 52), increase ≥ 1 log in BCR-ABL transcript level without losing MMR (n = 12), patient preference (n = 2), and withdrawal syndrome (n = 1). Treatment-free remission rate at 4 years was 64% (95% confidence interval, CI: 55%–72%). Cumulative incidence of molecular recurrence at 3 years was 33% (95% CI: 26%–38%). TKI treatment for < 5 years and MR4.5 duration shorter than 4 years were both associated with higher incidence of molecular recurrence. No patient had disease progression. Response status at last control was: MR4.5 (n = 196), MR4 (n = 15), MMR (n = 14), complete cytogenetic response (n = 10), and other (n = 1). A significant increase in Hb and cholesterol levels was observed after imatinib withdrawal. Our results demonstrate that TKI treatment discontinuation is feasible in real-life clinical practice.

1031P - The EUROSKI biomarker study: Analyzing the mechanisms of treatment-free remission in chronic myeloid leukemia

Background: A substantial proportion of chronic myeloid leukemia (CML) patients in deep molecular response (DMR) reach treatment-free remission (TFR) after tyrosine kinase inhibitors (TKI) cessation. The aim of this study is to identify a gene signature predictive for TFR using whole transcriptome expression analyses. Methods: RNA from peripheral blood (PB) leukocytes of 60 CML patients who stopped TKI therapy within the EUROSKI study and 10 healthy controls were isolated. CML patients were divided into two groups of whom n = 30 patients had ongoing TFR, while 30 patients encountered molecular recurrence. RNA was isolated at the last day of TKI intake. In order to investigate differentially expressed genes, whole transcriptome arrays (Clariom D, Affymetrix) were analyzed. Candidate biomarkers were tested in multivariate analyses and gene set enrichment analyses (GSEA). Results: CML patients in DMR compared to healthy controls showed 16000 differentially expressed genes (p < 0.05). The natural killer cell marker CD69 showed overexpression with highest fold change (> 8-fold) for CML patients versus healthy controls. Significant enrichment of NFκB mediated TNFα and TGFβ pathways (FDR<7%, p < 0.03) were found in CML patients. Comparing the CML TFR versus relapse cohort, we found 2600 differentially expressed genes. Most notably the toll-like receptors TLR1, TLR6 and TLR8 were upregulated in the relapse cohort (p < 0.03). Activated downstream signatures of NOD-like, TLR and TNFα pathways, known for their promotion of a protective CML microenvironment, were significantly enriched (p < 0.03, FDR<2%). In contrast, patients in TFR were characterized by an upregulation of T-cell receptor and granzyme gene family members (p < 0.03). Genes of interest showed distinct cut-offs predictive for TFR over a period of 12 months. Conclusions: CML patients in DMR present a considerable inflammatory gene expression pattern in PB leukocytes in contrast to healthy controls. Alike previous studies, genes involved in immune exhaustion and immune surveillance were differentially expressed between patients with TFR and relapse. The specific inflammatory gene signature of the relapse cohort suggests ‘stemness’ as third mechanism and driver for relapse. Legal entity responsible for the study: University Heidelberg Funding: ELN Foundation Disclosure: R. Sébastien: Novartis research fund. S. Saussele: Advisory board: Novartis, Bristol-Myers Squib, Pfizer and ARIAD Fees: Novartis, Bristol-Myers Squib, Pfizer and ARIAD Research grant: Novartis and Bristol-Myers Squib Travel grant: Novartis and Bristol-Myers Squib. All other authors have declared no conflicts of interest.