Blastic plasmacytoid dendritic cell neoplasm (BPDCN) is an ultra-rare, aggressive malignancy that mostly affects older adults, many of whom can’t tolerate intensive chemotherapy or reach allogeneic hematopoietic stem cell transplantation (allo-HSCT).
Tagraxofusp, a CD123-directed fusion protein, gave these patients a targeted frontline option, but registration trials were small, and the contribution of transplant to long-term outcomes after tagraxofusp remained unclear. A retrospective multicenter study of 32 patients treated across 13 European centers examined both questions (Schroeder et al., 2026).
Background: BPDCN and the Role of CD123 Targeting
BPDCN derives from precursors of plasmacytoid dendritic cells. It typically presents with skin lesions (bruise-like patches, plaques, or nodules) but is systemic and can involve bone marrow, peripheral blood, lymph nodes, and the central nervous system. Diagnosis rests on immunophenotype, with expression of CD4, CD56, and CD123.
Before targeted therapy, patients received intensive chemotherapy adapted from acute lymphoblastic leukemia, acute myeloid leukemia, or lymphoma regimens. These produce initial remissions, but relapse is common. Allo-HSCT offers the best chance of durable control, especially in first remission, but it requires an adequate response, sufficient fitness, and a suitable donor.
Tagraxofusp fuses interleukin-3 to a truncated diphtheria toxin. It binds CD123, the interleukin-3 receptor alpha chain highly expressed on BPDCN cells, and the toxin inhibits protein synthesis and induces cell death. Prospective trials showed activity in untreated BPDCN and supported FDA approval, but patients in routine practice may be older or have more extensive disease. The main toxicity is capillary leak syndrome (CLS), which can cause severe organ dysfunction.
Study Design
The investigators analyzed 32 patients who received frontline tagraxofusp between November 2019 and March 2024 at 13 tertiary centers in Germany, Austria, and Switzerland. Endpoints included response, adverse events, progression-free survival, overall survival, and outcomes after consolidative allo-HSCT.
The comparator was a published historical cohort of 40 patients treated between 2001 and 2022. Because treatment was not randomized, the investigators used propensity-score matching, inverse probability weighting, and multivariable regression. Transplantation was analyzed as a time-dependent event, and a day-90 landmark analysis reduced bias from the survival time needed to reach transplant.
Median age was 71.5 years, and more than half of patients were 70 or older. Skin involvement was present in 93.8%, bone marrow in 84.4%, peripheral blood in 53.1%, and CNS in 15.6%. CNS assessment followed individual center practice, so occult involvement may be undercounted. Intrathecal prophylaxis was given to 40% of patients.
Response and Survival
The overall response rate was 82.1%, with complete response in 60.7% of evaluable patients and partial response in 21.4%, consistent with prospective data. Median duration of response was 7.8 months. Response could not be assessed by compartment, so control of CNS disease in particular is unknown. At a median follow-up of 18.4 months, two-year OS was 40.5% and two-year PFS was 30.4%.
Durable Remission Concentrated in Transplanted Patients
Fourteen of 32 patients (43.8%) proceeded to frontline consolidative allo-HSCT at a median of 80 days, mostly with reduced-intensity conditioning. One more patient was transplanted later. Outcomes for transplanted versus nontransplanted patients:
- Two-year OS: 58.6% versus 24.9%
- One-year PFS: 92.9% versus 5.6%
- In remission at last follow-up: 78.6% versus 22.2%
- Median time to relapse: 504 versus 115 days
The advantage persisted when the analysis was limited to patients in complete remission.
Patients who reach transplant must survive long enough, keep disease control, and meet eligibility criteria, while those who don’t may have more aggressive disease or greater frailty. This selection bias can inflate the apparent benefit.
Adjusted analyses were directionally consistent but not significant: the time-dependent Cox model, adjusted for age and peripheral blood involvement, gave a hazard ratio of 0.31, and the day-90 landmark analysis (29 patients, 13 events) gave an HR of 0.25. The data can’t establish that transplant alone explains the difference, but durable remissions occurred almost exclusively in transplanted patients. Two late CNS relapses occurred among transplanted patients.
Historical Comparison: No Detectable Survival Advantage
The tagraxofusp and historical cohorts were similar in age, sex, CNS involvement, and peripheral blood involvement, but differed in skin, bone marrow, and hepatosplenic involvement. In pooled multivariable regression, tagraxofusp was not significantly associated with OS (HR 0.89).
Propensity-score matching gave two-year OS of 59.4% with tagraxofusp and 60.8% historically, and inverse probability weighting gave a similar result. These matched estimates apply only to the matched population and shouldn’t be compared with the 40.5% of the full tagraxofusp cohort.
The comparison was observational, small, and not designed to test noninferiority. The historical cohort also spanned two decades of changing diagnostics, supportive care, and transplant practice. A high response rate did not produce a detectable OS gain over that cohort.
Frontline allo-HSCT rates differed: 43.8% with tagraxofusp versus 17.5% historically. The investigators suggest that rapid responses with manageable toxicity may help more patients reach transplant. Changes in referral patterns, conditioning intensity, and patient selection could also explain the difference, so this remains a hypothesis.
CLS Was Frequent and Often Severe
CLS occurred in 53.1% of patients, and 47.1% of those had grade 3 or higher events. The investigators attribute the severity partly to the older population and differences in routine management. Management requires close attention to fluid balance, albumin, blood pressure, and organ function, particularly in early cycles. One-year nonrelapse mortality was 15%. No unexpected safety signals emerged.
Prognostic Factors
Peripheral blood involvement was independently associated with worse OS in the pooled multivariable analysis (HR 2.63), including after adjustment for treatment era. It may mark more disseminated disease, but small numbers and limited molecular and clinical data prevent building a validated risk model.
CNS involvement was documented in 15.6% of patients at baseline. Two-year OS was 48.1% without CNS disease and 0% with it, a difference limited by the small number of affected patients. CNS evaluation and intrathecal treatment were not standardized across centers, so the study can’t define optimal screening, prophylaxis, or treatment.
Patients Who Did Not Undergo Transplant
Eighteen patients did not proceed to frontline allo-HSCT, mainly because of advanced age, frailty, or disease progression, and their remissions were short. Subsequent chemotherapy was associated with longer OS than venetoclax-based therapy or no further treatment. The comparison doesn’t establish that chemotherapy outperforms venetoclax-based approaches.
Limitations and Clinical Implications
The retrospective design and small sample limit the reliability of multivariable models and subgroup comparisons. Performance status, transplant comorbidity scores, and molecular data were unavailable for some analyses.
Frontline tagraxofusp remains an effective induction option, including in older patients with extensive systemic disease, and the pattern of durable remissions supports planning allo-HSCT during the first response. For patients who can’t be transplanted, no consolidation or maintenance strategy has yet shown durable control, and optimal CNS management is unresolved.
