FDA Approval Summary: Tagraxofusp-erzs For Treatment of Blastic Plasmacytoid Dendritic Cell Neoplasm

Blastic plasmacytoid dendritic cell neoplasm (BPDCN) is an aggressive myeloid malignancy in the dendritic cell lineage. Although BPDCN was initially classified as a subset of acute myeloid leukemia and related neoplasms in the World Health Organization (WHO) 2008 guidelines, it was reclassified as a distinct myeloid malignancy in 2016. Diagnosis is based on morphologic criteria and expression of cell-surface markers, including CD123 (IL3R alpha), CD4, and CD56 (1). Clinically, the disease most commonly (90%) manifests as an indolent cutaneous lesion followed by rapid leukemic dissemination (2). Other extramedullary sites including lymph nodes, spleen, liver, or other organs may also be involved. The incidence of BPDCN is unknown, but the disease is estimated to account for less than 1% of all hematologic malignancies (3). The median age at diagnosis is 53 years (4), but a bimodal distribution has been reported (5).

There are no consensus guidelines on the treatment of BPDCN (6, 7). For adults and children with treatment-naïve BPDCN, treatment with intensive multiagent leukemia or lymphoma chemotherapy regimens has been used followed by allogeneic stem cell transplantation (HSCT), but there have been no prospective trials (6, 7). Complete remission rates have ranged from 25% to 100% in small series reported in the literature, but the reliability of these point estimates is uncertain given small patient numbers and the lack of established response criteria. Median overall survival reported in adults diagnosed with BPDCN ranges from 7 to 24 months (3, 5, 8, 9).

Herein, we provide a summary of the FDA review (10) of the marketing application for tagraxofusp-erzs for the treatment of BPDCN.

Tagraxofusp-erzs, a cytotoxin, is a fusion protein comprised of human IL3 and truncated diphtheria toxin (DT) produced in Escherichia coli by recombinant DNA technology. It targets cells that express CD123, the alpha chain of the IL3 receptor. Intracellularly, DT increases ADP ribosylation of elongation factor 2 (EF2), leading to inhibition of protein synthesis and apoptosis. Tagraxofusp-erzs decreased cell proliferation and increased expression of apoptotic markers in vitro in BPDCN cell lines and in blasts from patients with BPDCN. The IC₅₀ in primary patient BPDCN blast cells was in the femtomolar range. In mouse xenograft models of BPDCN, treatment with tagraxofusp-erzs increased overall survival compared with controls.

Following administration of tagraxofusp-erzs 12 mcg/kg via 15-minute intravenous infusion in patients with BPDCN, the mean area under the plasma drug concentration over time curve (AUC) was 231 (SD 123) hour·mcg/L, the maximum plasma concentration (C_max) was 162 (SD 58.1) mcg/L, volume of distribution (Vd) was 5.1 (SD 1.9) L, clearance was 7.1 (SD 7.2) L/hour, and terminal half-life was 0.7 (SD 0.3) hours. The presence of antidrug antibodies (ADA) had a significant effect on the pharmacokinetics of tagraxofusp-erzs. In patients with preexisting ADA, the mean AUC was 151 (SD 89.2) hour·mcg/L, C_max was 80.0 (SD 82.2) mcg/L, Vd was 21.2 (SD 25.4) L, and clearance was 13.9 (SD 19.4) L/hour after a tagraxofusp-erzs 12 mcg/kg via 15-minute intravenous infusion. An ANOVA showed statistically significant changes in mean AUC and C_max (P < 0.01) between patients with and without preexisting ADA.

No significant differences in the pharmacokinetics of tagraxofusp-erzs were observed based on age, sex, mild to moderate renal impairment (RI), mild to moderate hepatic impairment (HI), or body weight after adjusting dose by body weight. The effect of severe RI or severe HI on tagraxofusp-erzs pharmacokinetics is unknown.

Study STML-401-0114 (NCT 02113982) was a multicenter, multistage, single-arm trial of tagraxofusp-erzs. Stage I was a dose-escalation cohort that included 23 patients with BPDCN or relapsed or refractory (R/R) acute myeloid leukemia (AML) treated with 7–16 mcg/kg/day for 5 days in a 21-day cycle. On the basis of stage I, the MTD of 12 mcg/kg/day for 5 days was chosen as the recommended phase II dose (RP2D). Stage II was a dose-expansion cohort that included 58 patients with BPDCN or R/R AML treated at the RP2D. Of the 16 patients with treatment-naïve BPDCN treated in stages I and II, 14 achieved complete response (CR) or clinical CR (CRc) as defined in Table 1. A permutation test showed that the durability of CRc was similar to the durability of CR in this population, qualifying CR + CRc as a measure of efficacy.

Stage III was the pivotal cohort to support the indication for first-line treatment of BPDCN. Treatment consisted of tagraxofusp-erzs 12 mcg/kg/day i.v. on days 1–5 of a 21-day cycle and continued indefinitely in patients receiving benefit. The first cycle was given in the inpatient setting, and patients were premedicated to prevent infusion-related reactions. The primary endpoint was CR/CRc (Table 1). The key secondary endpoint was duration of CR/CRc defined as the time from when criteria were first met for CR or CRc until death or the date that criteria for relapse after CR/CRc were met. A sample size of 10 patients was calculated to have 90% power to target a 60% CR + CRc rate and exclude a 10% rate using a two-sided 95% Clopper Exact confidence interval. Because determination of eligibility required central review that could be delayed, the protocol specified that up to 15 patients could be included in the analysis population.

Thirteen patients with treatment-naïve BPDCN were treated in stage III of Study STML-401-0114. At the time of the efficacy analysis, all patients in the pivotal cohort had completed treatment with at least 6 months of follow-up. The demographics and baseline disease characteristics of the pivotal cohort are shown in Table 2. These 13 patients who received first-line treatment with tagraxofusp-erzs had a CR/CRc rate of 54% (95% CI: 25%–81%; Table 3); the median duration of CR/CRc was not reached (range: 3.9–12.2 months) with a median follow-up of 11.5 months (range: 0.2–12.7 months).

Additional exploratory response analyses were performed in the 15 patients with R/R BPDCN treated in stages I and II of Study STML-401-0114. Two (13%, 95% CI: 2%–41%) of these patients achieved CR/CRc, with one response lasting 3.7 months and one lasting 13.9 months.

Toxicology studies were conducted in cynomolgus monkeys. At human equivalent doses (HED) ≥1.6 times the recommended dose based on body surface area, severe kidney tubular necrosis was observed. At HEDs equal to the recommended dose, necrosis and degeneration of the choroid plexus in the brain was observed. This finding was irreversible and became progressively more severe at an HED 1.6 times the recommended dose 3 weeks after dosing stopped.

The safety of the recommended dose and schedule of tagraxofusp-erzs was assessed in 94 patients with active BPDCN (n = 58) or AML (n = 36). Overall, patients received a median of 2 cycles of tagraxofusp-erzs, and patients with BPDCN received a median of 4 cycles (range: 1–43). The most common (≥20%) treatment-emergent adverse events (TEAE) are listed in Table 4. The most common grade ≥3 nonhematologic TEAEs were AST increase, ALT increase, and glucose increase.

Treatment-related mortality in the safety population was 2% and was due solely to capillary leak syndrome (CLS). The specific term CLS was reported by investigators in only 18% of patients, but during the safety review it was noted that individual signs and symptoms of CLS were reported at a much higher frequency. Based on FDA's screening algorithm for patients experiencing the onset of 2 or more signs or symptoms of CLS within 7 days of each other (Supplementary Table S1), 55% of patients were determined to have had CLS. Grade ≥3 CLS was observed in 9% of patients, including the 2 fatal cases. The median time to onset of the first symptom was 4 days (range: 1–96), and most cases occurred during the first two cycles. Three patients required vasopressors and two required ventilator support.

Other significant adverse reactions included hepatotoxicity and hypersensitivity reactions. Elevated transaminases were reported in 88% of patients treated with tagraxofusp-erzs and were seen most commonly during cycle 1. Grade ≥3 aspartate aminotransferase and alanine aminotransferase elevations were reported in 37% and 30% of patients, respectively. Most cases were transient and resolved without intervention or with temporary dose interruption. Hypersensitivity reactions were reported in 46% of patients treated with tagraxofusp-erzs, and 10% had grade ≥3 events. Manifestations reported in ≥5% of patients included rash, pruritis, stomatitis, and wheezing.

Treatment discontinuations due to adverse reactions occurred in 11% of patients and were most commonly due to CLS (4%) or hepatotoxicity (2%).

Safety data were available for 3 pediatric patients (1 child and 2 adolescents) with BPDCN treated with tagraxofusp-erzs under compassionate use. The toxicity profile in the pediatric patients was similar to that seen in the adults. The toxicity profile of tagraxofusp-erzs in patients ages 75 years and older was notable for an increase in grade ≥3 altered mental status (including confusional state, delirium, mental status changes, dementia, and encephalopathy) compared with younger patients (45% vs. 6%).

Across the drug development program, 95% of evaluable patients had preexisting ADA (likely due to prior diphtheria vaccination), 99% developed treatment-emergent ADA by the end of cycle 2, and 85% had neutralizing antibodies. By cycle 3, exposure to drug (AUC) was markedly decreased, with many patients having measurements below the limit of quantification of the assay (LLOQ). In addition, treatment-emergent anti-IL3 antibodies were detected in 68% of patients.

Tagraxofusp-erzs is the first drug approved by FDA for the treatment of patients with BPDCN. The lack of established response criteria and the rarity of the disease were unique challenges that needed to be addressed during clinical development and the application review process.

The first issue was determining an appropriate efficacy endpoint for this disease. In the absence of established response criteria, CR with durability was initially considered the acceptable measure of clinical benefit in a single-arm trial. However, FDA worked with the applicant to qualify the clinical meaningfulness of CRc by evaluating data from an independent cohort of patients with treatment-naïve BPDCN treated in stages I and II of Study STML-401-0114. This qualification was affirmed by FDA prior to analysis of the pivotal cohort, and based on these data, FDA agreed that CR + CRc as defined in Table 1 with durability were acceptable criteria for demonstrating the clinical benefit of tagraxofusp-erzs in patients with BPDCN.

The second issue was designing a pivotal cohort with hypothesis testing for a very rare disease. Study STML-401-0114 prespecified the objective of excluding a 10% CR + CRc rate in patients with treatment-naïve BPDCN in stage III. Although statistically valid based on this assumption, the sample size of 10 patients is quite small for an assessment of efficacy. However, considering the rarity of the disease, the lack of available therapy, the projected size of the treatment effect, and the supporting data from the exploratory cohorts, the design was considered acceptable to support a marketing application.

In addition, although FDA does not usually approve new drugs based on outcomes from exploratory studies, the cohort with R/R BPDCN represents a distinct population whose response results are significant for patients and healthcare providers. This cohort alone would not be enough to support an approval on its own, but in the context of the positive results in the pivotal cohort and lack of alternative treatments, the results for the patients with R/R BPDCN were considered credible and allowed for the broadened indication for tagraxofusp-erzs.

Several notable safety concerns were observed during review of this application. First, 55% of patients experienced CLS, with grade ≥3 events occurring in 9% of patients, including 2% with fatal events. Although most patients could be managed by close monitoring and simple medical interventions, 3 patients required more advanced supportive care. The severity of this adverse reaction warranted a boxed warning for CLS and inclusion of monitoring and CLS sign/symptom-based management guidelines in the prescribing information (PI). In addition, the PI states that cycle 1 should be administered in the inpatient setting with observation through at least 24 hours after the last infusion as well as observation for a minimum of 4 hours following each infusion during subsequent cycles. Elevated transaminases were common with tagraxofusp-erzs treatment, occurring in 88% of patients, including grade 3 in 36% and grade 4 in 4%. Hypersensitivity reactions occurred in 46% of patients, including grade ≥3 events in 10%. The severity of both adverse reactions merited warnings and inclusion of instructions for monitoring and early intervention in the PI.

Patients with BPDCN in the safety population received a median of 4 cycles of treatment with tagraxofusp-erzs, but nearly all patients had ADAs by the end of cycle 2 of therapy, the majority had neutralizing antibodies, and measurable exposure was reduced. In vitro, tagraxofusp-erzs has been shown to be highly potent with an IC₅₀ approximately 6 logs lower than the LLOQ of the drug assay; therefore, the limit of detection of the assay might not be a reliable indicator of potential activity in vivo. Given the small sample size, no conclusions can be drawn regarding significant differences in response based on preexisting or treatment-emergent ADA. Because no safety issue was identified in patients with ADA, there was no basis for restricting treatment duration. However, the treatment-emergent anti-IL3 antibodies raised concerns about the effect of ADA on hematopoiesis. Although prolonged cytopenias were not observed in the trial, this finding is concerning because patients may proceed to hematopoietic stem cell transplantation and the effect of such antibodies on posttransplantation hematopoietic recovery is unclear. To determine whether such a risk exists, the approval requires enhanced pharmacovigilance for reports of graft failure or delayed engraftment in the postmarketing period.

Finally, although no pediatric patients were included in the clinical trial, safety data were available for 3 pediatric patients with BPDCN treated with tagraxofusp-erzs. Based on the biological similarity of BPDCN in adult and pediatric patients, and the lack of new safety signals in the children treated with tagraxofusp-erzs, the indication was extended to include pediatric patients 2 years and older.

Patients with BPDCN have limited treatment options and poor outcomes. In Study STML-401-0114, patients with treatment-naïve BPDCN treated with tagraxofusp-erzs had a CR/CRc rate of 54% (95% CI: 25%–81%), and the median duration of CR/CRc was not reached with a median follow-up of 11.5 months (range: 0.2–12.7). Responses were also seen in two patients with R/R disease. Capillary leak syndrome, hepatotoxicity, and hypersensitivity reactions were the major safety issues related to treatment with tagraxofusp-erzs. Development of ADAs was common and enhanced pharmacovigilance for graft failure or delayed engraftment in the postmarketing setting is warranted. Overall, with careful monitoring and appropriate risk mitigation strategies in place, the clinical benefit from treatment with tagraxofusp-erzs outweighs the expected risks for patients 2 years and older with BPDCN.

No potential conflicts of interest were disclosed.

The Editor handling the peer review and decision-making process for this article has no relevant employment associations to disclose.

Conception and design: E.Y. Jen, D. Przepiorka, G.M. Blumenthal, R. Pazdur

Development of methodology: E.Y. Jen, D. Przepiorka, C. Liu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E.Y. Jen, X. Gao, L. Li, L. Zhuang, D. Przepiorka, Y.L. Shen, C. Liu, S. Bowen, A.T. Farrell, G.M. Blumenthal, R. Pazdur

Writing, review, and/or revision of the manuscript: E.Y. Jen, X. Gao, L. Li, N.E. Simpson, B. Aryal, R. Wang, D. Przepiorka, R. Leong, C.M. Sheth, S. Bowen, K.B. Goldberg, A.T. Farrell, G.M. Blumenthal, R. Pazdur

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E.Y. Jen, D. Przepiorka, C. Liu, K.B. Goldberg, R. Pazdur

Study supervision: R. Pazdur

The authors thank Dr. Kristopher Kolibab for expert project management.