Prognostic Factors Suggest Benefit for Some With Myeloablative Busulfan/Fludarabine Conditioning in Myelofibrosis

September 10, 2020

Prognostic factors in patients with myelofibrosis treated with a myeloablative busulfan/fludarabine conditioning regimen prior to allogeneic stem cell transplantation in a single-institution analysis showed that a time from diagnosis to transplant of more than 12 months was associated with poorer overall survival and displayed a trend toward higher non-relapse mortality.

Prognostic factors in patients with myelofibrosis (MF) treated with a myeloablative busulfan/fludarabine (Bu-Flu) conditioning regimen prior to allogeneic stem cell transplantation (alloSCT) in a single-institution analysis showed that a time from diagnosis to transplant of more than 12 months was associated with poorer overall survival (OS) and displayed a trend toward higher non-relapse mortality (NRM).

The purpose of the analysis of Bu-Flu in patients with MF was to define an optimal condition regimen. It was hypothesized that intensifying Bu-Flu using pharmacokinetic (PK) monitoring would reduce relapse without increasing the rate of NRM.

Outcomes are typically determined using prognostic scoring systems, such as Dynamic International Prognostic Scoring System (DIPSS)-Plus and Mutation-Enhanced International Prognostic Scoring System for Transplant-Age Patients (MIPSS70), that often include spleen size >22 cm, presence of severe marrow fibrosis, >20 red cell transfusions, low hemoglobin <10 g/dL, matched-unrelated donor (MUD), matched-related donor (MRD), and lack of clinical improvement or progression on a JAK inhibitor. These factors all indicate poor prognosis in patients with MF.

The analysis included 65 consecutive adult patients with primary or secondary MF and used the traditional prognostic scoring factors to determine outcomes. Patients were treated with Bu-Flu from August 2007 to September 2019. Data from the study were presented during the eighth annual Society of Hematologic Oncology (SOHO) virtual meeting by Jacinth Joseph, MD.

At a median follow-up of 36 months, the 1-year OS rate following alloSCT was 78% (95% CI, 65%-86%), and at 3 years, the OS rate was 65% (95% CI, 51%-76%). In terms of NRM, the rate at 1 year was 16% (95% CI, 8%-26%), and the NRM rate at year 3 was 20% (95% CI, 11%-30%). The cumulative incidence of relapse at 1 year was 21% (95% CI, 12%-31%), and at 3 years, the rate was 27% (95% CI, 16%-30%).

The key baseline patient characteristics showed a median age of 61 years (range, 27-73). At the time of transplant, 77% of patients were ≤65 years old and 23% were >65 years of age. The majority of the patient population was female (52%). The Karnofsky performance score at baseline was 90 for the majority of patients (68%), followed by 80 for 17% of patients, 100 for 8% of patients, and 70 for 7% of patients. Performance status information was missing for 5 patients. Most patients (54%) had received 1 prior line of therapy; however, 22% had 2 prior lines, 12% had 3 prior lines, and another 12% had 4 or more prior lines of treatment. Fifty-one percent of the population did not have prior treatment with a JAK inhibitor, but 49% of patients did.

No high-molecular risk (HMR) mutations (ASXL1, SRSF2, U2AF 1Q157, EZH2, IDH1/2, TP53, and TET2) were found in most patients, but 40% of patients did have 1 HMR mutation. In addition, HMR mutational information was missing for 30 patients. JAK2 mutations specifically were present in 57% of patients and CALR mutations were found in 36%; CALR mutational information was missing for 23 patients. Per DIPSS-Plus, only 5% of the study population was intermediate-1 risk at baseline, while 53% were intermediate-2 risk and 42% were high risk.

Risk scores and molecular and cytogenetic characteristics were also observed at baseline. Few of the patients enrolled (8%) had an unfavorable karyotype per DIPSS-Plus. On the other hand, karyotype per MIPSS70 was favorable for 78% of patients and unfavorable for 22%. The key disease characteristics also assessed at baseline showed a hemoglobin level <10 g/dL in 85% of patients and ≥10 g/dL in 15% of patients. The donor type was MRD for 41% of patients, MUD for 44%, and mismatched unrelated donor transplant (MMUD) for 16%.

The first multivariable analysis looked at OS based on baseline characteristics. Regarding this analysis, Joseph stated, “the multivariable analysis of OS showed that time from diagnosis to transplant of greater than 12 months was associated with worse OS, with a hazard ratio (HR) of 3.01.” The P value for this OS result of time from diagnosis to transplant of ≤12 months versus >12 months was .040.

“HCT-CI [hematopoietic cell transplant comorbidity index] score greater than or equal to 3 was also associated with a worse OS, with a hazard ratio of 3.41. Donor type, being a mismatched unrelated donor compared to MRD was associated with an HR of 7.16 and a worse OS,” he said. The difference between patients with an HCT-CI score of 0 to 2 and ≥3 was significant (P = .016).

Joseph also noted that the multivariable analysis data show that patients who had peripheral blasts had a worse overall survival. The HR was 2.83 (95% CI, 1.07-7.48; P = .036). The final baseline characteristic determined to be associated with a worse OS was requiring a splenectomy prior to transplant. The HR for this comparison was 3.69 (95% CI, 1.17-11.64; P = .026).

The second multivariable analysis assessed how baseline characteristics impacted NRM. It was observed that an HCT-CI score ≥3 was associated with worse NRM. The HR for HCT-CI score of 0 to 2 versus ≥3 was 10.09 (95% CI, 2.09-48.76; P = .004). It was also determined that having an MUD donor was associated with a worse NRM compared with having MRD donor type with an HR of 5.38 (95% CI, 1.14-25.30; P = .033). Having an MMUD donor compared with MRD was also associated with worse NRM at an HR of 10.73 (95% CI, 1.05-109.4; P = .045).

The time from diagnosis to transplant was not statically significant in the study, but Joseph noted that having a >12-month time to transplant showed a trend toward worse NRM. The HR for this comparison was 7.20 (95% CI, 0.96-53.94; P = .055). Having a splenectomy prior to transplant was solely associated with worse overall outcomes with an HR of 12.50 (95% CI, 4.75-32.88; P <.001).

It was also noted that prior use of JAK inhibitors was associated with better outcomes in the multivariable analysis for cumulative incidence of relapse. The HR for patients who had prior JAK inhibition compared with those who did not was 0.29 (95% CI, 0.08-1.07; P = .06).

The analysis of myeloablative Bu-Flu as treatment of patients with MF was conducted at The University of Texas MD Anderson Cancer Center. During conditioning, patients received intravenous (IV) fludarabine 40 mg/m2 daily for 4 days, IV busulfan, which was PK-guided to a daily average AUC of 4000 mM/min, or a total course AUC of 16,000 mM/min. Patients also received graft-versus-host disease (GVHD) prophylaxis of tacrolimus (Prograf) and methotrexate (Rheumatrex). Patients identified as MUD also received Rabbit-antithymocyte globulin at 4 to 7.5 mg/kg.

Looking further into baseline characteristics, the assessment showed that white blood cell counts were ≤25 x 109 cells/L in 63% of patients and >25 in 37% of patients. Platelet counts were <100 x 109 cell/L in 51% of the study population and ≥100 in 49%. The majority of patients (91%) had bone marrow blasts ≤5%, but 6 patients (9%) had bone marrow blasts >5% and the laboratory results were missing for 1 patient. Peripheral blasts were 0 for 57% of patients and more than 0 for 43%.

Outside of laboratory disease characteristics, spleen size at baseline was a median of 17 cm (range, 0-46). Overall, the baseline spleen size was ≤20 cm for 89% of patients and >20 for 11%. Most patients (86%) did not undergo splenectomy, but the remaining 14% did have the surgery. The majority of the study population (81%) showed a normal liver size. The liver was enlarged for 19% of patients and 2 patients did not have baseline data regarding liver size.

Transfusion dependence and transfusion independence at baseline were almost even with 58% of patients found to independent, and 42% were found to be dependent on transfusion. B-symptoms were observed in 60% of patients. Sixty-nine percent of patients had primary MF at baseline and 31% had secondary MF. CD34 was at 52.5 cells/μL (range, 0.0-16428) with data missing for 11 patients. Overall, baseline CD34 was ≤126 in 65% of patients and >126 in 35%. Finally, the fibrosis grade was 1 for 12% of patients, 2 for 42%, and 3 for the remaining 46%.

The median number of months from diagnosis of MF to allogeneic stem cell transplant was 17.4 months (range, 2.6-297.5). Forty-three percent of the study population had a time from diagnosis to stem cell transplant of ≤12 months and 57% had a median time of >12 months. For 14 patients (22%), the HCT-CI score at baseline was 0. Among those who did have a score above 0, the score was 1 in 15%, 2 in 20%, 3 in 14%, 29 in ≥4%, 0 to 2 in 57%, and ≥3 in 43% of patients. The cell source for most patients was the peripheral blood stem cell, but bone marrow was the cell source for the remaining 9%.


Joseph J, Miltom D, Al-Atrash G, et al. Myeloablative fludarabine and busulfan regimen in myelofibrosis: long term outcomes and analysis of prognostic factors. Presented at: Society of Hematologic Oncology Virtual Annual Meeting; September 9-12, 2020. Abstract MPN-188.