Etcelera Launches AI Bet on Myelofibrosis Without Target Trial or Precedent

Flagship Pioneering Launches AI-Aided Drugmaker Etcelera

Flagship Pioneering has merged three of its portfolio companies—Etiome, Cellarity, and Empress Therapeutics—to form Etcelera, a new AI-aided drugmaker. Led by Adrian “Ad” Rawcliffe, Etcelera aims to leverage AI and diverse drug discovery tools to predict which experimental medicines will succeed in the clinic, focusing on novel disease drivers and small molecule therapies. The company will initially develop treatments for hematological conditions, immune diseases, and metabolic-fibrotic disorders, with specific programs targeting myelofibrosis and essential thrombocythemia.

• Strategic Consolidation for Enhanced Discovery: Flagship Pioneering consolidated three distinct biotech startups, Etiome, Cellarity, and Empress Therapeutics, into Etcelera. This strategic merger, led by former Adaptimmune CEO Adrian “Ad” Rawcliffe, is designed to integrate their unique technological pieces to redefine biological cause-and-effect and accelerate the discovery of new medicines.
• AI-Driven Platform for Predictive Drug Development: Etcelera's core innovation lies in combining the advanced technologies and datasets from its constituent companies with Flagship’s in-house AI capabilities and "perturbation data." This integrated platform is intended to precisely identify novel disease drivers and develop small molecule medicines with a significantly higher probability of clinical success compared to traditional drug discovery methods.
• Focused Therapeutic Pipeline and Collaborative Strategy: The newly formed company is initiating its pipeline with programs targeting hematological conditions, immune diseases, and metabolic-fibrotic disorders, specifically mentioning myelofibrosis and essential thrombocythemia. Etcelera also maintains an existing 2022 partnership with Novo for metabolic dysfunction-associated steatohepatitis (MASH) and actively seeks further collaborations to expand its platform's reach.

Etcelera's AI to Unravel Myelofibrosis's Core Drivers

Myelofibrosis is driven by canonical driver mutations in JAK2, MPL and CALR, present in about 80% of patients, with the remainder (≈10%) lacking identifiable driver mutations. The most prevalent mutation, JAK2 V617F, is associated with polycythemia vera, essential thrombocythemia and myelofibrosis, while CALR and MPL mutations are associated only with essential thrombocythemia and myelofibrosis. All three drivers converge on enhanced thrombopoietin receptor (MPL) signaling, establishing the thrombopoietin/MPL axis as the central molecular hub, leading to pathologic and persistent JAK2-STAT5 activation with activation of the MAPK and PI3K/Akt signaling axes. MPLW515L was identified in 9% (4/45) of JAKV617F-negative myelofibrosis, conferred cytokine-independent growth and thrombopoietin hypersensitivity in 32D, UT7, or Ba/F3 cells, and resulted in constitutive phosphorylation of JAK2, STAT3, STAT5, AKT, and ERK. CALR frameshift mutations produce proteins with a novel C-terminus that aberrantly traffics to the cell surface, activates MPL, and leads to constitutive Janus kinase/signal transducer and activator of transcription signaling, promoting megakaryocytic proliferation and hematopoietic stem cell fitness.

Disease heterogeneity, prognosis and progression to blast phase are shaped by additional non-driver mutations and cytogenetic evolution. Additional mutations affect genes involved in DNA methylation (TET2, DNMT3A, IDH1, IDH2), histone modification (ASXL1, EZH2), mRNA splicing (SF3B1, SRSF2, U2AF1, ZRSR2), signaling pathways (CBL, NRAS, KRAS), and key transcription factors (RUNX1, NFE2, TP53), with paired chronic-blast phase analysis revealing blast phase acquisition of ASXL1, EZH2, LNK, TET2, TP53, and PTPN11 mutations/variants. Among 187 patients screened, 32 (17%) harbored SRSF2 monoallelic mutations affecting residue P95, with significant associations with IDH mutations and shortened overall and leukemia-free survival independent of DIPSS-plus (HR = 1.9; 95% CI, 1.1-3.0) and IDH mutations (HR = 2.3; 95% CI, 1.4-3.8), while RUNX1 and PTPN11 mutations/variants were associated with shorter survival duration with hazard ratios of 2.1 (95% CI, 1.1-3.8) and 3.0 (95% CI, 1.1-6.6). Rapid leukemic transformation was linked to a EWSR1-MYB fusion with expression of MYB and of its target BCL2 respectively ≥4.7 and ≥2.8 fold higher at leukemic transformation than after chemotherapy, and to cytogenetic lesions including gain(1q), del(5q), idic (5), monosomy 17, and/or del(17p), with favorable abnormalities defined as sole deletions in 13q or 20q, or trisomy 9, and unfavorable abnormalities defined as rearrangement of chromosome 5 or 7, or ≥3 abnormalities, while patients with abnormalities of chromosome 17 had a median survival of only 5 months.

Bone marrow fibrosis is mediated by the myelofibrosis hematopoietic stem/progenitor cell through fibrogenic cytokines and an impaired microenvironment favoring malignant over normal hematopoiesis. Increased TGF-beta and bFGF expression was demonstrated in circulating megakaryocytic cells and platelets together with alterations of expression of these cytokines and their receptors in hematopoietic CD34+ progenitor cells, while fibroblasts participating in myelofibrosis were shown to be polyclonal, indicating a reactive process. Megakaryocytes, most of which are found in emperipolesis with neutrophils, display an altered secretome containing increased levels of interleukin-8 (IL-8) and P-selectin, with emperipolesis driven by P-selectin causing transforming growth factor-β to be released into the bone marrow microenvironment, and marrow expression levels of TGF-beta, VEGF, osteocalcin, PDGF-alpha, and SDF-1 remaining abnormally high. Expression of both types I and III collagens and fibrillin 1 by bone marrow stromal cells reveals extracellular matrix scaffolding in line with myelofibrosis progression, with increased expression of inflammatory cytokines, lysyl oxidase and transforming growth factor-β, impaired megakaryocyte function, abnormal deposition of excess collagen and reticulin fibers, osteosclerosis and pathologic angiogenesis, extramedullary hematopoiesis in spleen and liver, marked splenomegaly, marked thrombocytosis with platelet count 1.9-4.0 x 10(12)/L in murine bone marrow transplant models, and potential progression to a blast phase defined by 10-19% blasts in accelerated phase and ≥20% blasts in blast phase.

Addressing Myelofibrosis Treatment Gaps with Novel AI Approaches

Current treatment for myelofibrosis remains largely palliative, with allogeneic hematopoietic stem cell transplantation as the only curative option and Janus kinase inhibitors as the cornerstone for symptomatic or intermediate to high risk disease. Durability is limited by resistance, intolerance and cytopenias, and outcomes after ruxolitinib failure remain poor with no clear management standards.

• Transplant eligibility and toxicity restrict curative treatment: Transplant is the only curative option for myelofibrosis, but high rates of morbidity and mortality limit eligibility, and its restricted applicability due to toxicity highlights the importance of ongoing clinical trials. Posttransplant challenges include graft-versus-host disease, infectious complications, relapse, graft failure, and non-relapse mortality, with treatment-related toxicities remaining major problems posttransplant.

• Ruxolitinib resistance, intolerance and discontinuation: Many patients cannot tolerate ruxolitinib due to dose-dependent drug-related cytopenias, and even patients with a good initial response often develop resistance to ruxolitinib after 2-3 years of therapy. Over half of the patients discontinue treatment within 2-3 years, and after three years of ruxolitinib therapy, approximately half of all patients with myelofibrosis will likely have stopped treatment. There is no consensus definition of ruxolitinib failure, and defining which remains challenging.

• Cytopenias limit JAK inhibitor use, especially in cytopenic disease: Early JAK inhibitors (ruxolitinib and fedratinib) are often associated with cytopenias, particularly thrombocytopenia and anemia, which limit their tolerability. Patients with myelofibrosis with thrombocytopenia have few therapeutic options, and the presence of thrombocytopenia frequently precludes the use of JAK2 inhibitors. Cytopenic myelofibrosis is characterized by lower blood counts, more frequent additional somatic mutations outside the Jak/STAT pathway, and a worse prognosis.

• Limited disease-modifying activity and leukemic progression: While ruxolitinib and other JAK inhibitors are effective in controlling splenomegaly and alleviating constitutional symptoms, their benefit in terms of reversing bone marrow fibrosis or inducing complete or partial remissions appears to be limited, with evidence for a robust anti-clonal effect lacking. Ruxolitinib does not appear to affect the risk of leukemic transformation, and progression to myelogenous leukemia occurs in about 20% of patients within 10 years of diagnosis and is often fatal.

• Poor prognosis after JAK inhibitor failure and blast-phase transformation: Patients with myelofibrosis who are resistant to or relapse after Janus kinase inhibitor therapy have limited treatment options and typically poor prognoses, and no phase III trials or clear treatment guidelines exist to guide management strategies in this setting. The transformation of myelofibrosis to accelerated or blast phase is an infrequent form of failure with an extremely poor prognosis, with median survival typically under six months, whereby patients who are ineligible for transplantation have limited treatment options.

Etcelera's Role in Shaping Myelofibrosis's Evolving Treatment Landscape

Ruxolitinib remains first-line standard of care for symptomatic myelofibrosis, but resistance and intolerance develops in most patients, with approximately half of treated patients discontinue ruxolitinib treatment within 3 years and up to approximately 75% within 5 years, with median survival in several series is approximately 12-14 months after discontinuation. Four JAK inhibitors — ruxolitinib, fedratinib, pacritinib, and momelotinib — are currently approved by regulatory agencies, each exhibiting distinct pharmacological profiles suited to different clinical phenotypes. Fedratinib can be used front-line and has demonstrated impressive responses as a salvage option after ruxolitinib loss of response, pacritinib offers an option to address splenomegaly and/or symptoms in patients with severe thrombocytopenia, and momelotinib has demonstrated spleen, symptom, and anemia responses with transfusion independence achieved with momelotinib in severely anemic patients. Pooled ≥35% spleen volume reduction at week 24 was 32.7% (95% confidence intervals 27.3-38.6) for fedratinib, 15.8% (11.7-21.0) for momelotinib and 21.6% (13.8-32.3) for pacritinib 200 mg bis in die, while pooled ≥50% Total Symptom Score reduction rates were 32.8% (27.4-38.7), 25.3% (20.2-31.3) and 32.4% (22.9-43.7) respectively. Gecacitinib (also known as jaktinib) 100 mg twice a day demonstrated SVR35 in 64.8% (46/71) compared to 26.5% (9/34) on hydroxyurea 500 mg BID at 24 weeks (P=0.0002), with TSS50 rates of 62.0% versus 50%, and in ruxolitinib-intolerant patients jaktinib 100 mg bid achieved SVR35 at week 24 in 43.2% (19/44, 95% CI 29.7%-57.8%) with TSS50 in 61.8% (21/34).

Combination of JAK inhibition with non-JAK small molecules has defined recent frontline development. Pelabresib (CPI-0610), an oral, small-molecule BET1,2 inhibitor, in combination with ruxolitinib showed improvements in spleen volume reduction (SVR35) and total symptom score reduction (TSS50) in the phase 2 MANIFEST study (NCT02158858), where 68% (57 of 84) achieved SVR35 at 24 weeks and 56% (46 of 82) achieved TSS50, with 36% (29 of 84) with improved hemoglobin levels (mean, 1.3 g/dL; median, 0.8 g/dL), 28% (16 of 57) with ≥1 grade improvement in fibrosis, and 29.5% (13 of 44) with >25% reduction in JAK2V617F-mutant allele fraction. In the phase 3 MANIFEST-2 study (NCT04603495), pelabresib 125 mg once daily for 14 days followed by a 7-day break plus ruxolitinib 10 or 15 mg twice daily met the primary endpoint in 65.9% versus 35.2% to placebo-ruxolitinib (difference, 30.4%; 95% confidence interval (CI), 21.6, 39.3; P<0.001), with TSS50 in 52.3% versus 46.3% (difference, 6.0%; 95 CI, -3.5, 15.5). Navitoclax, a potent oral B-cell lymphoma (BCL)-X/BCL-2 inhibitor, plus ruxolitinib in JAKi-naïve patients in Cohort 3 of the Phase 2 REFINE study (NCT03222609) achieved SVR at week 24 in 63% (20/32), TSS response in 11 of 27 evaluable patients (41%), BMF improvement by ≥1 grade in 13/27 patients (48%), and ≥50% reduction in driver gene variant allele frequency in 21% of 24 evaluable patients. In the Phase 3 TRANSFORM-1 study (NCT04472598), navitoclax plus ruxolitinib achieved SVR35 at Week 24 in 63.2% versus 31.5% with placebo-ruxolitinib (P<0.0001), with SVR35 at any time in 76.8% versus 44.1% (nominal P<0.0001), while mean change in TSS at Week 24 was -10.2 vs -11.6 (P=0.2852).

Beyond spleen and symptom endpoints, assessment now encompasses anemia response, bone marrow fibrosis change, clonal suppression, and overall survival as measures of disease modification. Imetelstat 9.4 mg/kg intravenous once every 3 weeks in intermediate-2 or high-risk myelofibrosis relapsed or refractory to JAK inhibitors led to a median OS of 29.9 months with spleen and symptom response rates of 10.2% and 32.2% at week 24, and bone marrow fibrosis improvement in 40.5% and variant allele frequency reduction of driver mutations in 42.1% of evaluable patients, where fibrosis improvement and variant allele frequency reduction correlated with OS. In REFINE cohort 1a in patients with disease progression or suboptimal response on stable ruxolitinib monotherapy, patients with improvements in fibrosis of one grade or more and a reduction of 20% or more in variant allele frequency had improved overall survival (median overall survival not reached) compared with those who did not achieve fibrosis improvement or a reduction in variant allele frequency (median overall survival 28.5 months [95% CI 19.6-not estimable] for both). Novel candidates exploit multiple molecular pathways targeting telomerase (imetelstat), epigenetic regulation via BET inhibition (pelabresib/CPI-0610), the MDM2-p53 axis (navtemadlin/KRT-232), PI3K signaling (parsaclisib), and PIM inhibitors (nuvisertib), with potential to improve management of MF-associated cytopenia and recover bone marrow fibrosis.

AI's Bold Bet: Redefining Treatment in Complex Blood and Fibrotic Disorders

The emergence of Etcelera, a new AI-aided drugmaker, signals a significant evolution in the pharmaceutical industry's approach to tackling some of the most challenging diseases. By merging three Flagship Pioneering companies, Etcelera aims to harness artificial intelligence and diverse drug discovery tools to predict the clinical success of experimental medicines, focusing on novel disease drivers and small molecule therapies. This strategy is particularly pertinent for conditions like myelofibrosis (MF) and essential thrombocythemia (ET), where current treatments, while beneficial, often fall short of providing curative or truly disease-modifying outcomes.

For instance, in MF, JAK inhibitors like ruxolitinib have transformed symptom management and spleen volume reduction, yet they do not consistently impact the underlying abnormal hematopoietic clone, leading to disease persistence and a risk of leukemic transformation. Patients with high-molecular risk mutations (e.g., ASXL1, EZH2) or those with triple-negative MF (lacking JAK2, CALR, or MPL mutations) face particularly poor prognoses and shorter responses to existing therapies. This highlights a critical unmet need for agents that can address these specific patient populations and achieve more profound disease alteration.

Etcelera's AI-driven approach, which can integrate multi-omics data and single-cell atlases, holds the promise of identifying previously unrecognized somatic mutations or epigenetic dysregulations that drive disease progression and resistance. This could lead to the development of therapies that:

• Target novel pathways beyond the JAK/STAT cascade, especially for triple-negative MF or those with HMR mutations.

• Overcome the 'persistence' phenomenon observed with current JAK inhibitors, potentially by identifying drugs that induce complete hematological remission or reverse bone marrow fibrosis.

• De-risk drug development by more accurately predicting clinical success, thereby accelerating the pipeline for small molecule therapies in complex indications.

However, the path is not without its challenges. The biological complexity of these diseases, with their heterogeneous genetic landscapes and varied clinical presentations, demands highly sophisticated and validated AI models. Furthermore, the competitive landscape, particularly in MF with multiple approved JAK inhibitors and a robust pipeline of new agents, means any novel therapy must demonstrate clear superiority or address distinct unmet needs to achieve significant clinical and commercial impact. Ultimately, Etcelera's success will hinge on its ability to translate its AI-powered insights into tangible, disease-modifying treatments that move beyond palliative care and offer a new paradigm for patients.