How I treat transplant-eligible patients with myelofibrosis

Primary myelofibrosis (PMF), post–polycythemia vera myelofibrosis (PV MF), and post–essential thrombocythemia (ET) MF are Philadelphia chromosome–negative myeloproliferative neoplasms with a heterogenous clinical course, characterized by constitutional symptoms, splenomegaly, and progressive bone marrow failure.¹ Despite the approval of Janus kinase (JAK) inhibitors and new developments, hematopoietic stem cell transplantation (HSCT) remains the only curative treatment option. Numbers of HSCT continue to increase worldwide,² but its inherent therapy-related morbidity and mortality hampers its use for many patients. Furthermore, identifying patients who are eligible to receive transplantation and managing various pre-, peri-, and post-HSCT scenarios pose a unique challenge to physicians for patients who are often older and have comorbid MF. For a proper definition, selection, and management of transplant eligibility, several disease-, patient-, and transplant-specific factors must be taken into consideration.³ 

Unfortunately, prospective trials are often lacking, making an evidence-based decision process particularly challenging. To illustrate how we approach patients with MF with respect to HSCT, we present 3 different clinical scenarios to capture relevant aspects that influence our decision making. Concepts and approaches are presented as clinical vignettes with summary points of how we treat transplant-eligible MF.

Over the past 15 years, several prognostic systems have been introduced, informing clinicians about prognosticated outcomes of patients with newly diagnosed PMF and post-PV/ET MF (Table 1). All currently existing prognostic scores have been reviewed in detail recently.⁴ 

We need to acknowledge that the time between diagnosis and HSCT can vary substantially between patients with MF and that risk factors might change during the disease course. In that regard, a comparison of HSCT vs conventional treatment showed an improved survival for patients at intermediate-2 and high risk according to the Dynamic International Prognostic Scoring System (DIPSS), whereas patients with low risk clearly benefited from conventional treatment.⁵ Based on these results, consensus recommends HSCT for patients with intermediate-2– and high-risk DIPSS scores.⁶ The main uncertainty remained for patients at DIPSS intermediate-1 risk for whom the comparison of HSCT vs no HSCT was inconclusive.

In terms of HSCT, various new risk factors and, especially, end points such as treatment-related mortality may interact with established systems and eventually influence the decision for or against HSCT. The Myelofibrosis Transplant Scoring System (MTSS) was developed with a specific aim to predict the outcome after HSCT, showing a 5-year survival and nonrelapse mortality of 90% and 10% for low, 78% and 22% for intermediate, 63% and 36% for high, and 43% and 57% for very high risk.⁷ 

Comparison of prognostic performance in HSCT indicates better performance and significant reclassification of patients using the MTSS, and ∼40% with low transplant-specific risk would be intermediate-1 risk according to other systems.^8-12 Taken together, we consider HSCT for patients with DIPSS intermediate-2/high, MF secondary to polycythemia and thrombocythemia prognostic model (MYSEC-PM) intermediate-2/high, mutation-enhanced International Prognostic Scoring System for patients <70 years (MIPSS70) high, and an MTSS low/intermediate risk profile, whereas patients with intermediate-1 or MIPSS70 intermediate and MTSS low risk are informed about the risks and potential benefits of intensive treatment on an individualized basis.

Another most recently developed prediction model aimed to account for response to ruxolitinib after 6 months and has already been externally validated.^13,14 Such approaches might help navigate the often heterogenous disease course in MF (Figure 1), and future studies with respect to this model and HSCT are warranted.

A 69-year-old man with post-PV MF was referred for HSCT, having received the PV diagnosis 10 years ago and subsequent MF diagnosis 3 years ago. He was initially treated with phlebotomy and acetylsalicylic acid but showed disease progression: bone marrow fibrosis grade 3, beginning fatigue, spleen size 15 cm by ultrasound, and night sweats; his Karnofsky index was 100% without any other comorbidity, with hemoglobin at 9.1 g/dL, leukocytes at 22 × 10⁹/L, platelets at 188 × 10⁹/L, circulating blasts of 9%, trisomy 8, and JAK2V617F and EZH2 mutations. Meanwhile, he became transfusion dependent. His risk was high according to MYSEC-PM (16.65 points). According to the MTSS, he was categorized as being at intermediate (if an HLA compatible donor was available) or high risk (with the availability of a mismatched unrelated donor).

Is there an indication for HSCT in this 69-year-old man? If yes, what is the optimal preparation to reduce morbidity, risk of mortality, and relapse?

Current consensus recommends HSCT for patients with MF with a DIPSS intermediate-2 and high-risk score and age <70 years.⁶ However, a higher chronological age should not be prohibitive for the eligibility decision in general, acknowledging that current life expectancy for the general population aged 70 years is ∼15 years. The European Society for Blood and Marrow Transplantation registry reported an increase in the median recipient age for patients with MF from 49 years before 2006 to 59 years currently, with a significant increase of patients aged >70 years.² Careful selection of patients aged >65 years with a low comorbidity score may result in an excellent 6-year survival of 72%.¹⁵ 

Apart from age, other important factors are comorbidities and overall physical performance of the patient. The Karnofsky index is the most frequently used and a powerful tool to evaluate patients’ performance with respect to post-HSCT outcomes.^7,16,17 In terms of comorbidities, the hematopoietic cell transplantation comorbidity index represented a valid approach to calculate comorbidity-associated risk after HSCT but has been shown to have less utility in the modern era and has not been validated in MF yet.^18,19 

We differentiate between general coexisting dysfunctions and disease-specific comorbidities that will disappear when the disease is treated successfully. For instance, resolution of pulmonary hypertension after HSCT was confirmed in several reports, with repeat measurement of pulmonary hemodynamic characteristics, suggesting a myelopulmonary pathophysiologic link.^20,21 Nevertheless, this remains a serious comorbidity that needs close monitoring and timely action in case of no response after transplantation.²² Outcomes of patients with portal hypertension or portal vein thrombosis, in our experience, seem to be particularly dismal if patients present late with progressive disease, but systematic data are lacking.

Other tools such as geriatric and social assessments together with patient preferences after appropriate consultation are also relevant when considering individual eligibility for HSCT.

Cytogenetics are not included in the current transplant-specific risk stratification process. It has previously been shown that HSCT can overcome the negative effect of poor-risk cytogenetics.^23,24 Additionally, many patients have no analyzable metaphases, and approximately two-thirds have a normal karyotype, making karyotype-based models vastly impractical.^25,26 The only potentially useful cytogenetic information could be the presence or absence of a complex karyotype, because this might help inform patients with concurrent TP53 mutation (discussed subsequently).²⁷ 

Regarding molecular genetics, several studies confirmed that high-risk molecular mutations (ASXL1, EZH2, SRSF2, and IDH1/2) as a whole category have no prognostic utility in the transplant setting,^7,24,28,29 whereas the presence of ASXL1 and CALR/MPL-unmutated driver mutation genotype negatively affected survival.^7,28-30 Of note, no difference was found for different CALR mutations.³¹ Patients carrying MPL mutation showed excellent 5-year survival of >80%.³² Therefore, patients newly diagnosed with MPL mutation should know that this mutation confers improved outcomes when undergoing HSCT.

Mutations in TP53 have been analyzed previously for patients who did not undergo HSCT, showing worse outcomes.^33-35 In the transplant setting, we showed that only patients presenting with a TP53 multihit configuration showed an exceptionally high-risk feature in terms of survival and, importantly, also in terms of leukemic transformation after HSCT. Therefore, in environments in which vast next generation sequencing panel analysis may not be available, clinicians should at least include driver mutation genotypes, ASXL1 and TP53, in their workup at the time of HSCT.³⁶ 

A major concern in our aforementioned first case may be the high number of circulating blasts, which is part of all disease-specific risk scores.³⁷ In contrast, several studies failed to show an impact of circulating blasts on posttransplant outcomes. For example, although an international collaborative study showed that the risk of relapse after HSCT increases when the number of circulating blasts is ≥10% in patients undergoing uniform reduced intensity conditioning (RIC) without specific reductive treatment before HSCT, the overall survival and nonrelapse mortality were similar between chronic- and accelerated-phase MF.³⁸ Thus, we currently do not use any blast reduction therapy before HSCT for chronic-phase MF (<10% blasts), whereas investigational blast reduction can be considered in accelerated-phase MF to reduce risk for relapse.

Blast-phase transformation before HSCT represents the most feared complication in MF and its estimated incidence after 20 years is <10%.³⁹ HSCT is the only option for significantly improved outcomes, showing a 3-year survival of ∼30%, whereas only 1% survived in the absence of HSCT.^40,41 The major prognostic factors are time to HSCT and achievement of complete remission to induction chemotherapy before HSCT as well as absence of unfavorable cytogenetics and mutations.^40,42,43 Certain molecular profiles, including RAS or TP53 mutations, might be associated with a progression to blast-phase MF before HSCT.^35,36,38 

Recent small-sample studies investigated combination therapies using venetoclax and hypomethylating agents or liposomal daunorubicin/cytarabine (CPX)-351, either upfront or after failing another induction.^44,45 Median survival was ∼9 months.⁴⁶ This underscores undergoing HSCT as soon as possible to avoid postponing curative treatment for eligible patients.

Like circulating blasts, anemia is a risk factor in all disease-specific risk scores but is not predictive of outcome after HSCT, unless transfusion dependence has resulted in severe iron overload and hemosiderosis.

Deferasirox is used for the management of iron overload and achieved transfusion independence in ∼20% when used in combination with ruxolitinib.⁴⁷ However, because effective chelation takes several weeks, transplantation should not be postponed to avoid disease progression. For patients with high iron overload determined by high serum ferritin at time of HSCT, we give deferasirox during conditioning to suppress the appearance of labile plasma iron in order to decrease free radicals, which may decrease the incidence of infection and organ toxicity (Table 2).⁴⁸ Importantly, therapeutic drug monitoring is the key to facilitate a safe coadministration with busulfan, because deferasirox leads to a considerable increase in busulfan area under the curve (∼40%).⁴⁹ 

Bone marrow fibrosis is a hallmark of the disease, but its prognostic impact is unclear and might significantly differ between different types of MF, whereas evaluation of the degree of fibrosis may be prone to subjectivity. Furthermore, grade of fibrosis has no prognostic role for post-HSCT outcomes, and most patients referred for HSCT already present with fibrosis grade 2 to 3.

Transplantation is capable of reversing fibrosis.⁵⁰ A complete or nearly complete regression could be seen in 33% on day 100 and in 90% on day 180.⁵¹ Earlier complete regression is associated with better outcomes.^52,53 

Splenomegaly is another hallmark of MF but is not part of any disease-specific risk scores. In the setting of HSCT, splenomegaly is associated with considerable burden and higher risk of delayed engraftment, graft failure, and relapse.⁵⁴ This might be associated with significant sequestration of (stem) cells by massive spleens, whereas mortality could be associated with subsequent hepatic injury, portal hypertension, and overall morbidity. Notably, evidence of effects on mortality remains heterogeneous, and many studies rely on simple palpation measures (referred to as “cm below left costal margin”), with high interobserver and intraobserver variability. In our experience, when prospectively measuring spleen volume with computed tomography, we confirmed higher risk of relapse based on spleen size but similar outcome in terms of mortality.⁵⁴ 

To reduce splenomegaly before HSCT, there are 3 options: splenic irradiation, splenectomy, and drug treatment. Neither splenic irradiation nor splenectomy before HSCT have shown an overall survival benefit,⁵⁵ whereas splenectomy is associated with significantly increased risk of relapse, being mainly driven by spleen size before splenectomy, reflecting advanced disease despite removal of the organ.^56,57 Splenectomy of exceptionally large spleens is not a routine surgery and requires experience to minimize complications, which subsequently could delay HSCT.⁵⁸ Therefore, it seems more rational, to us, to aim for best spleen response/reduction in order to reduce disease burden.

Results of splenic irradiation are limited to small-sample studies but have been reported more frequently in recent years, but a coherent approach across centers has not yet been determined.^59,60 Notably, splenic irradiation can also result in prolonged myelosuppression, suggesting cautious dosing and an unmet clinical need to predict efficacy and safety of this approach. Thus, in rare cases and until better evidence emerges, we use spleen irradiation starting ∼7 to 10 days before conditioning therapy with moderate radiation dose (3-4 Gy). In clinical practice, we observed a reduction in spleen size during busulfan-based conditioning regimen in the majority of patients.

Most patients in the current era will have received JAK inhibitor treatment before HSCT,^2,61,62 and a thorough review on current treatments has been published recently.⁴ Ruxolitinib is the preferred first-line option in most countries.⁶² 

For select patients, alternative JAK inhibitors can be considered. For example, patients with thrombocytopenia potentially benefit from second-generation inhibitors such as pacritinib or fedratinib.^63,64 This could be of benefit particularly in the HSCT setting, because thrombocytopenia is a consistent factor for worse posttransplant outcomes (Table 2).⁷ Momelotinib, another new option, is a JAK1/JAK2 inhibitor also antagonizing activin A receptor type 1, thereby downregulating hepcidin expression and increasing availability of iron for erythropoiesis.⁶⁵ It demonstrated a unique ability to improve hemoglobin and reduce transfusion burden, leading to spleen size and overall symptom burden reduction.⁶⁶ Although anemia seems to have no significant impact on HSCT-related outcomes across studies, many patients presenting for HSCT are transfusion dependent, and momelotinib as bridging strategy could, at least, ameliorate the clinical burden of patients with MF. All in all, apart from ruxolitinib, evidence on other JAK inhibitors with respect to the HSCT setting must be generated to facilitate decision making and counseling of patients.

The most important factor for a successful combination of pretreatment and HSCT is the right time point, and a large registry study has shown best outcomes for patients with ongoing spleen response to JAK inhibitors.⁶¹ Three months of JAK inhibitor treatment were shown to achieve a spleen response in patients who are treatment naive.^67-69 Therefore, we recommend administering ruxolitinib around 3 months prior to HSCT and continuously assessing the response, taking also a certain risk of clonal evolution into account.^70,71 

If patients need therapy and do not respond to ruxolitinib while preparing for HSCT, we consider new drugs, depending on the clinical profile of each patient. For patients with a high or very high risk according to MTSS, we would consider treatment with new JAK inhibitors as well as other investigational drugs in clinical trials to improve patients’ fitness and hematological features and continuously reevaluate their individual transplant-specific risk.

1. Age, per se, is not a contraindication for HSCT, and we recommend transplantation in patients aged >65 years and even in those aged >70 years, depending on performance status and comorbidities after balancing disease-specific and transplant-specific risks. In general, we recommend transplantation for patients with DIPSS and MYSEC-PM intermediate-2/high and MIPSS70 high-risk scores, when patients have low/intermediate risk per the MTSS.

2. We do not use blast reduction for chronic-phase MF. For accelerated- and blast-phase MF, investigational blast reduction with chemotherapy or with a combination of venetoclax and hypomethylating agents can be considered. Early HSCT is the most important factor for best outcome.

3. We aim for our patients to undergo HSCT at the time of best spleen response to JAK inhibitor treatment. If JAK inhibitors fail and spleen size is extensive, we consider spleen irradiation for a few cases and splenectomy only for selected patients taking high complications of the procedure into account.

A 51-year-old male patient with JAK2-, ASXL1-, and EZH2-mutated PMF, with a DIPSS intermediate-2 score (score 3: hemoglobin level at 8 g/dL and leukocytes at 30 × 10⁹/L) and at high risk per the MIPSS70 (score 6: non-CALR type 1, high molecular risk, hemoglobin level at 8 g/dL, and leukocytes at 30 × 10⁹/L), was referred for HSCT. He had a younger 47-year-old HLA-identical brother. According to MTSS, he was at intermediate risk (score 4: leukocytes at 30 × 10⁹/L, and ASXL1 and JAK2 mutation).

How should the transplant be performed considering graft-versus-host disease (GVHD) prophylaxis and donor platform in this relatively young patient with high molecular genetic risk?

GVHD prophylaxis is a crucial component of the HSCT platform. The standard GVHD prophylaxis consists of a calcineurin inhibitor (either tacrolimus or cyclosporine) and methotrexate. In Europe, calcineurin inhibition is combined with in vivo T-cell depletion using either anti–T lymphocyte globulin (ATLG) or antithymocyte globulin, mostly for patients undergoing an unrelated donor HSCT, and results for both products appear to be similar across diseases.^72,73 

Our practice is based on a prospective randomized trial,⁷⁴ and we use ATLG also for HSCT from matched sibling donors at a lower dose (Figure 2). Recently, we could show that GVHD-free and relapse-free survival rates in our MF cohort seems to be higher in comparison with those in another recent report from the Center for International Blood and Marrow Transplant Research (40% vs 20%, respectively, at 2 years),⁷⁵ and that the presence of GVHD was associated with reduced risk of relapse.^76,77 

Posttransplant cyclophosphamide has been used successfully as GVHD prophylaxis for patients with MF undergoing haploidentical, matched, or mismatched unrelated HSCT, but more evidence and longer follow-up is needed. Smaller studies on CD34⁺ selection with acceptable GVHD incidence but ∼7% graft failure have been reported.^78,79 

Several studies of patients with MF showed best absolute survival rates for matched related donor HSCT.^17,24,29 The MTSS reflects increased risk of death only for those with HSCT from a mismatched unrelated donor.^56,80,81 The observation of statistically comparable outcomes of matched related and matched unrelated HSCT has been confirmed by several studies.^17,75,82 

Numbers of haploidentical donor with posttransplant cyclophosphamide are steadily increasing, showing better results than, at least, mismatched unrelated donors,⁸¹ but evidence for MF is still limited. Small studies evaluated its feasibility reporting conflicting results in terms of survival (with short follow-up) and high relapse rates.^83-85 Furthermore, rejection rates and poor graft function are not negligible under current limited evidence. Cord blood transplantation in MF is rarely used and is associated with a high risk of graft failure.⁸⁶ 

Peripheral blood is the predominant stem cell source.² No direct comparison with bone marrow grafts has been reported for patients with MF. Peripheral blood has been shown to be associated with a higher probability of engraftment.⁸⁷ Infusion of higher doses might be associated with higher rates of neutrophil and platelet recovery.⁸⁸ 

1. We use ATLG as GVHD prophylaxis for all patients and a reduced dose (30 mg/kg) for patients with a matched related donor.

2. If available, we prefer matched related donor whereas most of our HSCTs are on basis of matched unrelated donors. Haploidentical transplants remain investigational but can be used in environments with limited access to matched unrelated donors.

3. We usually perform transplantation with >5 × 10⁶ CD34⁺/kg cells and use peripheral blood for almost all procedures, if available.

A 53-year-old male patient was diagnosed with PMF 4 years ago. He presented with fibrosis grade 3, hemoglobin level of 11.8 g/dL, leukocytes at 34 × 10⁹/L, 1% circulating blasts, platelets at 480 × 10⁹/L, and a spleen size of 2 fingers below the left costal arch. Molecular analyses revealed JAK2V617F and ASXL1 mutations. He had 1 HLA-identical brother (aged 53 years). Risk was classified as DIPSS intermediate-1, MIPSS70 intermediate, and MTSS intermediate. At that time, balancing risks and benefits were in favor of postponing HSCT. Two years later, he presented with similar blood levels, increased spleen size, and constitutional symptoms. Molecular genetics showed increased variant allele frequency. Now, the patient was classified as being DIPSS intermediate-2 risk, MIPSS70 high risk, and MTSS intermediate risk. He received ruxolitinib and HSCT indication.

Because HSCT indication is given, what type of conditioning should we use, and how can the risk of relapse and complications be reduced?

The rationale behind choosing a certain conditioning intensity before HSCT is to balance the risk of relapse against the risk of nonrelapse mortality.⁸⁹ Several large international retrospective studies showed that outcomes are at least comparable between RIC and higher intensity conditioning.^29,82 This could be confirmed in the molecular era, showing no clear benefit of higher intensity conditioning for patients with higher molecular risk.²⁹ Digging deeper into certain conditioning regimens, all analyses suggest slightly better outcomes for a RIC regimen incorporating busulfan and fludarabine, with lower rates of acute GVHD and subsequently better overall survival; this regimen was recently shown to be the best option even in the higher intensity setting.^75,82 Adding low-dose total-body irradiation to busulfan + fludarabine regimen or replacing busulfan with thiotepa might mitigate against graft failure in patients with MF without increased rates of complications,⁹⁰ but larger studies are needed. Table 3 summarizes selected key studies evaluating conditioning regimens and intensities in MF.

In terms of the peritransplant management of patients on JAK inhibition, ruxolitinib dose tapering 3 to 5 days before, or stop dosing at, day of conditioning has been suggested to avoid withdrawal syndrome,^61,69,91 In case of withdrawal syndrome, corticosteroids or JAK inhibitor rechallenge can be used to reduce symptom rebound. The continued use of JAK inhibitors during the peritransplant period until hematological recovery is currently under investigation and has been shown to reduce GVHD rates (Figure 2).^92-95 

Primary graft failure and delayed hematopoietic constitution after HSCT are major challenges in MF management. Patients with MF, because of fibrosis, inherent inflammation, or splenomegaly, often need more time to recover than patients with other hematological diseases.⁹⁶ Primary graft failure indicates failure to achieve a neutrophil count on day 28 and is commonly accompanied by low or absent donor chimerism.⁹⁷ Primary graft failure has been reported in ∼14% of patients.¹⁷ To prevent graft failure, reduction of spleen size or adding low-dose total body irradiation (TBI) to the conditioning regimen may be options.⁹⁸ In our ATLG-based reduced conditioning regimen, we observed only 10 primary graft failures in 475 patients (2.1%). After we encounter graft failure, our patients undergo a second HSCT from the same donor, if possible, as soon as possible using minimal intensity conditioning (fludarabine and low-dose TBI).

Poor graft function is a condition of bilineage or trilineage cytopenia, with complete donor chimerism occurring earlier (transfusions dependent but incomplete donor chimerism) or being delayed (after a period of transfusion independence). We see poor graft function in ∼17% of the patients with MF, and the main risk factor is spleen size before HSCT. Growth factors often fail to translate into long-term benefits.

We face the choice between CD34⁺-selected boost without further conditioning or posttransplant splenectomy. For most patients, with hypocellular bone marrow, a CD34⁺ cell–selected stem cell boost is the treatment of choice. In patients with hypercellularity in the bone marrow and persistent spleen size, posttransplant splenectomy may improve peripheral blood counts more likely than a stem cell boost (Figure 3A).⁹⁹ 

Hepatic adverse events occurring early after HSCT are frequently observed and are usually transient, mostly related to AT(L)G.¹⁰⁰ A serious complication constitutes veno-occlusive disease or sinusoidal obstruction syndrome (VOD/SOS), thought to occur more frequently in MF than in other diseases.¹⁰¹ However, frequencies of VOD/SOS differ significantly across reports, ranging from ∼30% to <5%.^102,103 A recent European Society for Blood and Marrow Transplantation study showed rates of 12%, with no significant impact on mortality.¹⁰² Our experience and reports from others show significantly lower rates (∼5%), even with a busulfan-based conditioning regimen.¹⁰⁴ Nevertheless, close monitoring with ultrasound and regular measurement of hepatic enzymes and bilirubin is of utmost importance, especially in the early phase after HSCT.

Approximately 10% to 30% of patients experience relapse after HSCT within a median of 7 months.^96,97 However, also late relapse can occur, and we observed a 14% cumulative incidence of relapse after 5 years.¹⁰⁵ Molecular monitoring by sensitive polymerase chain reaction analysis for 1 of the driver mutations (JAK2, CALR, or MPL) or highly sensitive chimerism for triple-negative MF after HSCT are able to detect minimal measurable disease, which has been shown to precede hematological relapse. This enables early intervention.^32,106 

If patients still have detectable mutation or mixed chimerism, we start reducing immunosuppression carefully and rapidly, even by day 70 after transplantation if early molecular relapse is detected. If discontinuation of immunosuppression does not lead to complete molecular response, we start targeted donor lymphocyte infusion (DLI) in a dose-escalated manner. Administering DLI for molecular relapse resulted in 88% complete molecular remission in contrast to only 60% if DLI was given in hematological relapse.¹⁰⁷ Furthermore, excellent response to DLI can avoid a second HSCT, and ∼50% of the patients achieved a molecular remission without any GVHD. In selected cases after DLI failure, a second HSCT (for which we used treosulfan-based conditioning) could be considered, which resulted in long-term freedom from disease in ∼40% to 50% of patients (Figure 3B).^108-110 

Ruxolitinib can lead to a reduction in spleen size, an improvement of constitutional symptoms, and a reduction of the blood transfusion interval in patients with relapse after HSCT.¹¹¹ Ruxolitinib does not lead to an increase in donor chimerism. However, ruxolitinib is also used for treatment of steroid-refractory GVHD.¹¹² Most of our patients with MF who received ruxolitinib for GVHD treatment were already exposed to ruxolitinib, and most of them showed GVHD response; we also saw an increase in JAK2 burden and subsequent relapse in a few patients while on treatment.

Thus, ruxolitinib is helpful in treating GVHD, but we do not use ruxolitinib for maintenance or treatment of relapse unless relapsed patients still suffer from splenomegaly or constitutional symptoms.

1. All our patients who receive first HSCT, receive RIC with busulfan (area under the curve targeted) and fludarabine.

2. Currently, we invesitgate to continue JAK inhibition throughout transplantation and taper up to day 28 or to engraftment in patients with spleen response and/or on ruxolitinib before HSCT.

3. CD34⁺-selected boost is a valid option for poor graft function if the bone marrow is hypocellular, whereas in case of hypercellularity and persistent splenomegaly, posttransplant splenectomy can be considered.

4. Long-term molecular monitoring using polymerase chain reaction analysis for driver mutations and/or high-sensitivity chimerism analysis is crucial to adequately monitor patients for relapse.

5. DLI should be administered for molecular persistence and target molecular remission. A second HSCT can be considered for fit patients after DLI failure.

The steady increase in numbers of HSCT for MF worldwide signify the interest in a real disease-modifying treatment. With the introduction of JAK inhibitors and promising new developments, it becomes even more important to identify the right time point and to design the right platform for curative treatment. MF remains a challenging disease, especially in this intensive treatment setting, and therefore complex networks of patient-, disease-, and transplant-related factors need to be considered. Patients should be carefully assessed for treatment response and ideally undergo HSCT at time of response to JAK inhibition. Prospective trials with novel agents should be designed. Last, molecular monitoring and DLI are the standard of care to achieve best outcomes for relapsed MF after HSCT.

Contribution: N.K., C.W., and N.G. wrote the manuscript, and gave final approval for the manuscript.

Conflict-of-interest disclosure: N.K. received honoraria Kite/Gilead, Jazz, MSD, Neovii Biotech, Novartis, Riemser, Pfizer, Bristol Myers Squibb; and research support from Neovii, Riemser, Novartis, and Deutsche Knochenmarkspender Datei (DKMS). The remaining authors declare no competing financial interests.

Correspondence: Nicolaus Kröger, Department for Stem Cell Transplantation, University Medical Center Hamburg-Eppendorf, Martinistr 52, 20246 Hamburg, Germany; e-mail: nkroeger@uke.uni-hamburg.de.