PI3K Inhibition Restores and Amplifies Response to Ruxolitinib in Patients with Myelofibrosis

Primary myelofibrosis (PMF), post-polycythemia myelofibrosis (PPV-MF) and post-essential thrombocythemia myelofibrosis (PET-MF) [collectively referred to here as myelofibrosis (MF)] are clonal myeloid stem cell disorders characterized by megakaryocyte atypia, reticulin fibrosis of the bone marrow, and frequent mutations in Janus Kinase 2 (JAK2), calreticulin (CALR), or myeloproliferative leukemia proto-oncogene/thrombopoietin receptor gene (MPL; refs. 1–6). Common features of MF include extramedullary hematopoiesis, thromboembolic disease, abnormal cytokine expression with associated constitutional symptoms, anemia, thrombocytopenia, and a propensity for leukemic progression with the acquisition of common epigenetic, splicing or DNA damage response gene mutations (e.g., ASXL1, TET2, EZH2, TP53, U2AF1; refs. 7, 8).

While hematopoietic stem cell transplantation (HSCT) may be curative, many patients are not eligible due to comorbidities. Current treatment options for patients with HSCT-ineligible MF are primarily designed to palliate symptoms and are limited both in number and efficacy (9). The JAK1/2 inhibitors, ruxolitinib, fedratinib, and JAK2/IRAK1 inhibitor, pacritinib are FDA approved and have become a mainstay of therapy, improving survival in addition to reducing spleen size and MF-related symptoms (10–12). Follow-up studies have demonstrated long-lasting effects on symptoms and survival for responders to ruxolitinib, but half of spleen responses were lost by 3 years, and only a quarter of study participants remained on ruxolitinib therapy after 5 years of treatment (13). Discontinuation rates may be significantly higher than this in clinical practice, and median survival after discontinuation of ruxolitinib is estimated to range between 13 and 16 months (14–17). In addition to progressive disease and suboptimal response, drug-induced cytopenias, cutaneous malignancies, and risk of infections lead to ruxolitinib dose de-escalation and discontinuation (18). Though patients may not meet criteria for a clinical response at the highest tolerated dose, the drug is often continued to palliate symptoms in clinical practice, so defining ruxolitinib failure is complicated and an active area of investigation (15, 17, 19, 20).

JAK inhibitors were developed following observations that hyperactivated JAK/STAT signaling is central to the pathogenesis of MF. However, resistance to JAK inhibition develops in nearly all cases, and compensatory signaling pathways, including the PI3K/AKT, NFκB, and ERK pathways, have been implicated in this resistance (21–23). Preclinical studies targeting the PI3K/AKT pathway in combination with JAK/STAT inhibition showed significant reductions in proliferation and clonogenicity in JAK2^V617F cell lines (24, 25). Moreover, combined PI3K and JAK inhibition reduced splenomegaly and prolonged survival in a JAK2^V617F-driven murine model (24). The PI3K catalytic subunit, p110, has become a promising target for interruption of this pathway, and its δ isoform has drawn the most attention in hematologic malignancies due to its expression being largely confined to hematopoietic cells. Though inhibitors of PI3Kδ have largely been developed in lymphoid malignancies (26, 27), PI3Kδ appears to be the predominant isoform found in myeloid malignancies, and specifically, ruxolitinib-treated bone marrow samples from patients with MF (28–30).

Umbralisib (TGR-1202), a PI3Kδ isoform-selective and casein kinase 1-ε (CK1ε) inhibitor causes apoptosis in leukemia and lymphoma cell lines and has led to significant responses in clinical trials for lymphoid malignancies (31, 32). While other PI3Kδ inhibitors have been associated with severe inflammation-driven adverse events (AE), such as transaminitis, colitis, and pneumonitis, umbralisib has shown a favorable toxicity profile, thought to be due to its comparatively weaker cytotoxic and inhibitory effect on regulatory T cells (Treg; refs. 31, 33, 34). In this phase I study, we evaluated the safety and preliminary efficacy of the combination of umbralisib and ruxolitinib in untreated patients with MF or patients who had relapsed on or suboptimal response from ruxolitinib monotherapy.

Splenomegaly is both an important diagnostic feature and a key measure of response in patients with MF. Clinical trials to date have relied on time and labor-intensive methods to determine spleen volume, either by physical examination or carefully measured spleen volumes as determined by CT or MRI. Automated delineation of abdominal anatomic structure, specifically spleen segmentation, from clinical imaging is an emerging technique that offers a potential remedy (35). Recently, an automatic pipeline for spleen segmentation was developed which uses an end-to-end synthesized process, which, for a given CT image, can accurately define spleen volumes within a minute, whereas the manual segmentation requires approximately 30 minutes (36, 37). We validated this technology, and then, as an exploratory aim of this study, utilized this automatized “digital atlas” process for spleen volume assessment for the first time in this clinical trial.

This was an open-label phase I dose-escalation and -expansion study conducted at five clinical sites across the United States. Dose escalation occurred in two stages. A total of 11 subjects were enrolled in escalation stage 1 (ES1), in which patients continued their stable twice daily dose of ruxolitinib, and the umbralisib daily dose was escalated in a modified 3+3 dose-escalation algorithm to determine the MTD of umbralisib irrespective of ruxolitinib dose (Supplementary Table S1). Safety was assessed through patient history, physical examinations, and laboratory parameters. Adverse events were assessed utilizing the NCI Common Terminology Criteria for Adverse Events v4.03. An AE was considered a dose-limiting toxicity (DLT) if it met protocol-specified criteria. The MTD was defined as the highest dose at which ≤1 of 6 patients experienced a DLT during cycle 1 of therapy. After preliminary safety evaluation and finding of MTD of umbralisib in ES1, escalation stage 2 (ES2) tested the combination of umbralisib 600 mg (the previously determined umbralisib MTD from ES1) with escalating dose of ruxolitinib (Supplementary Table S2) to determine the recommended phase 2 dose (RP2D) combination. Patients safely treated with 600 mg umbralisib in ES1 were included in the safety analysis in ES2 at their respective ruxolitinib dose, and additional patients were enrolled with stable ruxolitinib dose in a modified 3+3 dose-escalation algorithm to complete the safety profile (e.g., if 2 patients in ES1 were treated with umbralisib 600 mg daily and ruxolitinib 10 mg twice daily, one additional patient was enrolled in ES2 on a stable ruxolitinib dose of 10 mg twice daily to complete the cohort before any patients on higher dose of ruxolitinib were enrolled). Ruxolitinib 20 mg twice daily and umbralisib 600 mg daily were determined to be the recommended maximum expansion phase doses. In the expansion phase, as in the escalation phase, patients receiving a stable dose of maximally tolerated ruxolitinib (≤20 mg twice daily) for greater than or equal to 8 weeks continued their ruxolitinib, and umbralisib 600 mg was given daily. Treatment cycles were every 4 weeks and continued indefinitely until disease progression with no perceived clinical benefit, unacceptable toxicity, or withdrawal of consent. Response evaluations consisted of physical examination, laboratory evaluation, bone marrow examination, myeloproliferative neoplasm-symptom assessment form total symptom score (MPN-SAF TSS) questionnaire, and CT of the abdomen with digital atlas to determine spleen volume. Response evaluations occurred after the first 8 weeks, on day 1 of cycles 3 and 5, and at the end of study treatment. Additional response evaluations were permitted at the discretion of the treating investigator. Response and progression were evaluated in accordance with revised International Working Group-Myeloproliferative Neoplasms Research and Treatment (IWG-MRT) consensus report (38).

The primary objective was to evaluate the safety of umbralisib in combination with ruxolitinib and to determine the RP2D of each drug in combination. Secondary endpoints were to estimate overall response rate, complete response (CR) rate, partial response rate, clinical improvement (CI), change in the MPN-SAF TSS, and overall survival (OS). A key exploratory objective was to apply the validated automated spleen volume response measurement digital atlas to CT images.

Eligible patients were ≥18 years old and had pathologically confirmed PMF, PPV-MF or PET-MF (as per World Health Organization diagnostic criteria) with grade ≥1 marrow fibrosis according to European Consensus on Grading of Bone Marrow Fibrosis, and were categorized as having intermediate-1–risk, intermediate-2–risk or high-risk disease using the Dynamic International Prognostic Scoring System (DIPSS) Plus (39). Patients entered the trial on a stable, maximally tolerated therapeutic dose of ruxolitinib for at least 8 weeks prior to study enrollment and either failed to achieve a response in spleen size, blood counts, or disease-related symptoms (according to the MPN-SAF TSS score); or experienced disease progression after an initial response. Other eligibility criteria are included in the Supplementary Table S2.

This study was conducted in accordance with the International Conference on Harmonization Good Clinical Practice guidelines and complied with the ethical principles outlined in the Declaration of Helsinki. The study protocol was approved by institutional review boards at each participating site prior to study initiation. All patients provided written informed consent.

Safety and pharmacokinetic data were examined on an ongoing basis while the study was being conducted. Continuous variables were summarized using the 25th, 50th (median), and 75th percentiles along with the mean and SD for completeness. Categorical variables were summarized as frequencies and percentages of group. The highest-grade AE per category per patient was tabulated. The best improvement in hemoglobin or spleen volume (or worst change in cases where no improvement occurred) was captured for every patient prior to progression. Change from baseline (e.g., hemoglobin, spleen volume) was assessed using the Wilcoxon signed-rank test. Progression-free survival (PFS) was defined as the time from start of therapy to progression or death for any reason. Progression is defined on the basis of IWG-MRT and European LeukemiaNet response criteria for MF (38). Patients progression-free and alive at last follow-up were censored for PFS. OS was defined as the time from the start of therapy to death for any reason. Patients alive at last follow-up were censored for OS. Distributions of PFS and OS were estimated using the method of Kaplan and Meier. Analyses were conducted using open-source R software (R Core Team, 2018).

Abdominal CT imaging was processed centrally with an automated encapsulation of a validated deep learning spleen segmentation approach (36, 37). Results were manually reviewed to ensure that automatic detection of three-dimensional (3-D) spleen boundaries aligned with visual inspection of the tissues and that the imaging field of view included the full spleen with at least one slice on each boundary of the spleen. Volumes were calculated in 3-D and recorded.

The data generated in this study are available upon request from the corresponding author.

Between July 13, 2015 and July 23, 2019 a total of 37 patients with previously treated MF were enrolled—11 patients in ES1, 4 in ES2, and 22 in the expansion cohort (Supplementary Fig. S1). Table 1 and Supplementary Table S3 includes baseline patient characteristics. Among the 37 patients with ruxolitinib-experienced MF who were assessed for response, median age at study entry was 67 years; 62% were male; 89% had Eastern Cooperative Oncology Group performance status <2. Median duration of ruxolitinib prior to study enrollment was 54.7 weeks (range, 8–536 weeks). Patients in ES1 and ES2 began treatment at their stable dose of ruxolitinib [median dose of ruxolitinib at study enrollment was 15 mg twice a day (range, 5–20)], and umbralisib 400–800 mg (600 mg = MTD) daily was added. Treatment was continued until disease progression with no perceived clinical benefit, unacceptable toxicity, or withdrawal of consent. The median number of treatment cycles was 8 (range, 1–34).

Among 37 patients with ruxolitinib-experienced MF, MPN driver mutations were seen in 31 of 32 tested patients (97%): 16 (50%) JAK2^V617F, 10 (31%) CALR (5-bp insertion or 52-bp deletion), and 5 (16%) MPL^515W mutation. One (3%) patient was “triple-negative.” At the time of study enrollment, approximately 24% of the patients had high-risk disease, 32% had intermediate-2–risk disease, and 38% had intermediate-1–risk disease, according to the DIPSS Plus. Six (16.2%) had platelet counts <100 × 10⁹/L, 22 (59.4%) had hemoglobin <10 g/dL, and 34 (91.9%) had less than a complete spleen response to their prior ruxolitinib monotherapy. Of the 36 patients assessed, 13 (36.1%) had abnormal karyotypes, with 7 of those qualifying as unfavorable karyotypes [including complex, del(7q), and 11q23 abnormalities].

In ES1, the umbralisib starting dose was 400 mg daily. The dose was escalated by increments of 200 mg (with two dose de-escalation levels) in a 3+3 design to determine the MTD of umbralisib irrespective of chronic ruxolitinib dose. Two patients treated with 800 mg umbralisib (together with ruxolitinib 15 and 10 mg, respectively) had asymptomatic grade 3 elevations in serum amylase and lipase during cycle 1 of therapy that persisted after drug was held, meeting DLT criteria. Both patients had peak plasma umbralisib concentrations 1.5–2× higher than the other patients who received 800 mg of umbralisib, although steady-state levels were equivalent. No DLTs were seen in patients treated with lower dose of umbralisib in ES1 or ES2. Thus, the MTD determined for umbralisib was 600 mg daily, and all subsequent patients were treated with 600 mg umbralisib in combination with ruxolitinib.

The most common AEs (regardless of attribution) in the patients with ruxolitinib-experienced MF (n = 37) during treatment were hematologic (Table 2). Fourteen (38%) and 10 (27%) patients experienced grade 1/2 anemia and thrombocytopenia, respectively, whereas 10 (27%) and 4 (11%) patients had grade 3/4 anemia and thrombocytopenia. Two patients (5%) experienced grade 3/4 diarrhea, with one case of colitis in a patient with chronic intermittent diarrhea at enrollment, both of which were self-limiting. Eight patients (22%) experienced grade 1/2 alanine aminotransferase (ALT) elevation and 12 patients (32%) had grade 1/2 aspartate aminotransferase (AST) elevation, but there were no grade 3/4 AST/ALT elevations. No patients experienced autoimmune pneumonitis. A total of 14 patients experienced infectious complications (any grade). Two patients expired during the study due to infectious complications commonly seen in patients with later stage MF. Three patients withdrew from the study. The 23 patients not included in the relapsed and/or refractory MF cohort presented here were evaluated for AEs and there was no clinically meaningful difference between safety (Supplementary Tables S4–S8).

We compared the mean umbralisib concentration in peripheral blood when given concurrently with ruxolitinib. Pharmacokinetic data were consistent with those generated with single-agent umbralisib (31), indicating that ruxolitinib does not alter the absorption or metabolism of umbralisib and vice versa (Supplementary Fig. S2).

A total of 37 patients with ruxolitinib-experienced MF were evaluable for response. Response evaluations included bone marrow examination, assessment of hematologic indices and symptoms, and calculation of spleen volume from CT scans using an automated synthetic segmentation network. Two patients achieved CR per revised IWG-MRT response criteria (38). Prior to the enrollment, Patient #1 had splenomegaly, anemia, and thrombocytosis on the maximal dose of ruxolitinib she tolerated. On study, she achieved CR after 15 cycles of combination therapy and proceeded with matched unrelated allogeneic HSCT (Fig. 1A and C). Patient #2 had an excellent prior response to ruxolitinib [resolution of symptoms, greater than 35% spleen size reduction] but developed rapid onset of symptoms and spleen growth after 15 months of monotherapy. She was enrolled on study and achieved CR after five cycles of combination therapy; she remains on therapy, in CR, for over 56 cycles (Fig. 1A–C). An additional 12 of 37 patients (32.4%) met revised IWG-MRT criteria for CI based on peripheral blood count, symptoms and/or spleen responses. The mean maximal reduction in MPN-SAF TSS was 34.6%, with 11 (30%) patients reaching IWG-MRT criteria for a symptom response (Fig. 2A). These responses were largely durable, with 7 of 11 symptom responders maintaining a greater than 50% improvement in MPN-SAF TSS until last assessment. Per revised IWG-MRT, spleen response can be determined by physical examination or calculation of spleen volume, as assessed by MRI or CT. As documentation of spleen measurements by physical examination is inconsistent and highly subjective, spleen volume historically has been estimated from MRI or CT using three linear dimensions, which provides only a rough estimate of spleen volume. We utilized a novel automated spleen segmentation digital atlas algorithm to calculate precise spleen volumes (36, 37). Using these volumetric measurements, 27 of 32 subjects studied had decreased spleen volumes after initiating treatment with umbralisib + ruxolitinib, with a median change in spleen volume of 105.6 cm³ (33.8–349.4 cm³; Fig. 2B). In addition, 75.7% (n = 28) of patients experienced improvement in symptoms at a level below the threshold of IWG-MRT criteria. Mean improvement in hemoglobin was 1.4 g/dL (range, 0.5–4.3 g/dL). Most significant hemoglobin responses (>2 g/dL) occurred with longer treatment duration (Fig. 2C).

Combination therapy with ruxolitinib and umbralisib in patients with prior ruxolitinib experience led to an estimated median PFS of 15.3 months (95% confidence interval: 8–24.7 months; Fig. 3A); median OS was not reached with a median follow-up of censored patients was 50.3 months (Fig. 3B).

Here, we report the results of an escalation and expansion study of the novel PI3Kδ inhibitor umbralisib in patients with at least 8 weeks of ruxolitinib exposure and failure to achieve or maintain clinical goals with ruxolitinib alone. The rationale for this study was the observation across multiple research groups of the possible synergistic relationship between inhibiting both the PI3K/AKT and JAK/STAT pathways in myeloid disease (22, 29, 40, 41), and the observation that ruxolitinib treatment may increase MF hematopoietic stem cell dependence on the PI3Kδ isoform for AKT signaling (17). The clinical use of PI3Kδ inhibition after failure to achieve and maintain clinical response with ruxolitinib is particularly relevant, given that this population of patients has poor outcomes and limited treatment options. Survival of patients with MF who develop intolerance, resistance, or progression of disease on ruxolitinib therapy is poor (estimated median OS 13 to 16 months; refs. 15, 42).

Umbralisib, in combination with ruxolitinib, is safe and well tolerated in patients with MF. Colitis and transaminitis are well-documented class effects associated with PI3Kδ inhibitors; however, in our study, these were relatively rare (26, 27, 43, 44). Inflammation induced by PI3Kδ inhibitors is thought to be attributed to enhanced inhibition of the function and proliferation of Tregs, specifically, creating a proinflammatory imbalance in T-cell subsets (34). Umbralisib uniquely inhibits CK1ε and has been shown to preserve Treg function in vivo (30), which may explain a decreased immune-mediated toxicity profile in our current study. The addition of ruxolitinib may further impact Treg function, though that was not explored here.

In this study, 57% of the patients were DIPSS intermediate-2 risk or high risk which carries an estimated median OS of 16–35 months (39). Here, the sample size was small, and it was not clear whether the rate of expected infections exceeded expectations in advanced MPNs. While umbralisib was approved by the FDA in 2021, long-term follow-up data from the phase III UNITY-CLL trial (NCT026112311), led to the voluntary withdrawal of umbralisib and revised scrutiny on infectious risk with PI3Kδ inhibitors. The UNITY-CLL trial revealed the possibility of an increase of death from infections, especially from COVID-19, in patients with chronic lymphocytic leukemia (CLL) being treated with umbralisib (45, 46). There is considerable speculation that patients with MPNs receiving ruxolitinib, however, may be protected from postinfection inflammatory reactions associated with increase in mortality from COVID-19 after a prospective observational study of ruxolitinib in an aged population of hospitalized patients revealed improvement in pulmonary function and recovery from COVID-19 in over 85% (47). Likewise, unlike patients with CLL, patients with MPNs often have functional cellular immunity. Taken together, with the reasonable toxicity profile in our current study of ruxolitinib together with umbralisib at 600 mg, it is possible umbralisib can be reconsidered in myeloid disease.

In this article, 15 of 37 (41%) patients experienced, at least, a clinical benefit by IWG-MRT criteria, and 2 patients reached a CR, an outcome not commonly reported in the absence of HSCT. Likewise, the median OS in this study was not reached, with a median follow-up of 50.3 months, which is two to three times longer than estimated survival of patients with MF after ruxolitinib experience, based on retrospective studies (15, 17, 42, 48). Our study population of patients with ruxolitinib-experienced MF was comparable with these retrospective populations in average age, hemoglobin, and spleen size, but appeared to include patients with higher platelet counts and more participants with normal karyotypes. Interestingly, platelet count of <100 × 10⁹/L has been previously identified as a predictor of poor outcome after ruxolitinib failure, with an estimated survival of 12 months (17). However, the 6 patients with ruxolitinib-experienced MF with low platelets (<100 × 10⁹/L) at the time of study treatment had a median follow-up of 36.7 months (range, 26.3–40.6 months), and at the data cutoff (November 1, 2021) 3 of the 6 patients were deceased, but each of these patients had lived more than 24 months (range, 26.3–40.6 months).

In the treatment of MF, disease progression is often nonbinary. Ruxolitinib had become a backbone of therapy, and often ruxolitinib has been continued with dose de-escalations to balance control of constitutional symptoms or splenomegaly at the expense of worsening anemia and thrombocytopenia, as the marrow becomes more scarred and less capable of normal hematopoiesis. While there have been efforts to standardize “ruxolitinib-refractoriness” in clinical trials, this remains a moving target (49). Here, we purposely chose to treat patients on study who were “ruxolitinib-experienced” for greater than 8 weeks at the same dose, as long as this therapy underperformed with respect to normalization of symptoms, spleen size, or blood counts; or if they lost response previously achieved on ruxolitinib. Granted, this more liberal inclusion does make the independent effect of umbralisib more difficult to assess, and a randomized study of continued ruxolitinib with umbralisib versus a standard therapy is necessary to illustrate definitive improvement in risk-benefit. Still, the addition of umbralisib to ruxolitinib in patients who began to lose ruxolitinib response, or who never achieved a response in the first place, fits well into the use of such an agent in clinical practice, and the responses in this initial phase I dose-escalation and -expansion study exceed expectations from ruxolitinib alone in historical controls (10, 11).

Spleen measurement via MRI or CT requires investment of time from radiologist, and thus can become a cost-limitation in investigator-initiated clinical trials. The digital atlas created via the spleen segmentation algorithm is a validated option for body organ measurement (36, 37) and for the first time, this technique was effectively used in a clinical trial to measure response to therapy. This should provide an effective and efficient resource in future clinical trials requiring organ volumetric measurements.

In conclusion, the current therapeutic options for patients who do not benefit or lose benefit from ruxolitinib are limited. The addition of umbralisib to ruxolitinib in these patients was well tolerated and resulted in clinically significant reductions in symptom burden and spleen volume and increases in hemoglobin. CRs were seen in a minority of patients with long-term benefit both on and off therapy, and OS for the entire population was meaningfully greater than previously reported in this limited sample size. Enhancing suppression of pathologic JAK-STAT signaling with PI3Kδ inhibition can augment and induce a response in patients with MF with suboptimal or lost response to ruxolitinib monotherapy, and the use of umbralisib in this context should be further considered.

T.K. Moyo reports personal fees from TG Therapeutics during the conduct of the study as well as personal fees from Kite Pharmaceuticals outside the submitted work. A. Kishtagari reports personal fees from CTI Biopharma and Geron outside the submitted work. B. McMahon reports other support from CTI Biopharma outside the submitted work. S.R. Mohan reports grants from Incyte during the conduct of the study. M. Childress reports grants from TG Therapeutics during the conduct of the study as well as other support from Foundation Medicine outside the submitted work. P.B. Ferrell reports grants from Novartis outside the submitted work. S.A. Strickland reports grants from TG Therapeutics during the conduct of the study as well as personal fees from AbbVie, BerGen Bio, Genentech, Ellipses, Nkarta, Novartis, SentiBio, Kura Oncology, and Syros outside the submitted work. B.A. Landman reports grants from Federal government and Vanderbilt University Medical Center during the conduct of the study. R.A. Mesa reports grants and personal fees from Incyte and BMS and personal fees from Novartis, CTI, Telios, Geron, and Morphosys outside the submitted work. J.M. Palmer reports other support from Sierra Oncology, CTI, Incyte, and Pharmessentia outside the submitted work. L.C. Michaelis reports personal fees from AbbVie, NKarta, Incyte, Novartis, and Sierra Oncology and grants and personal fees from Jazz Pharmaceuticals outside the submitted work. M.R. Savona reports grants from TG Therapeutics during the conduct of the study as well as personal fees from AbbVie, BMS, CTI, Forma, Geron, Karyopharm, Novartis, Ryvu, Sierra Oncology, and Taiho; grants from ALX Oncology, Incyte, TG Therapeutics; and grants and personal fees from Takeda outside the submitted work. In addition, M.R. Savona has a patent for VU16112A issued, licensed, and with royalties paid from Boehringer-Ingelheim and a patent for PCT/US22/31403 issued. No disclosures were reported by the other authors.

T.K. Moyo: Conceptualization, data curation, software, formal analysis, investigation, visualization, methodology, writing–original draft. A. Kishtagari: Data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft. M.T. Villaume: Data curation, software, formal analysis, investigation, visualization, methodology, writing–review and editing. B. McMahon: Data curation, software, investigation, visualization, methodology, writing–review and editing. S.R. Mohan: Data curation, software, investigation, visualization, methodology, writing–review and editing. T. Stopczynski: Resources, software, formal analysis, investigation, visualization, methodology, writing–review and editing. S.-C. Chen: Software, formal analysis, visualization, methodology, writing–review and editing. R. Fan: Software, formal analysis, visualization, methodology, writing–review and editing. Y. Huo: Software, formal analysis, visualization, methodology, writing–review and editing. H. Moon: Software, formal analysis, visualization, methodology, writing–review and editing. Y. Tang: Data curation, formal analysis, visualization, methodology, writing–review and editing. C.A. Bejan: Software, visualization, methodology, writing–review and editing. M. Childress: Data curation, formal analysis, visualization, methodology, writing–review and editing. I. Anderson: Data curation, writing–review and editing. K. Rawling: Data curation, writing–review and editing. R.M. Simons: Data curation, writing–review and editing. A. Moncrief: Data curation, writing–review and editing. R. Caza: Data curation, writing–review and editing. L. Dugger: Data curation, writing–review and editing. A. Collins: Data curation, writing–review and editing. C.V. Dudley: Data curation, writing–review and editing. P.B. Ferrell: Data curation, formal analysis, investigation, visualization, writing–review and editing. M. Byrne: Data curation, formal analysis, investigation, visualization, writing–review and editing. S.A. Strickland: Data curation, formal analysis, investigation, visualization, writing–review and editing. G.D. Ayers: Conceptualization, software, formal analysis, supervision, investigation, visualization, methodology, writing–original draft. B.A. Landman: Software, formal analysis, visualization, methodology, writing–review and editing. E.F. Mason: Formal analysis, visualization, methodology, writing–review and editing. R.A. Mesa: Resources, supervision, investigation, writing–review and editing. J.M. Palmer: Resources, supervision, investigation, writing–review and editing. L.C. Michaelis: Resources, supervision, investigation, writing–review and editing. M.R. Savona: Conceptualization, data curation, formal analysis, supervision, funding acquisition, investigation, visualization, methodology, writing–original draft.

This study was a multicenter investigator-initiated study funded in part by TG Therapeutics.

M.R. Savona is a Leukemia and Lymphoma Society Clinical Scholar and is supported by the E.P. Evans Foundation, the Biff Ruttenberg Foundation, the Adventure Alle Fund, the Beverly and George Rawlings Directorship, and the NCI. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.