Pharmacokinetic Variability in Anti‐TFPI Therapies: Why Absolute Metrics Matter for Plasma Monitoring

1.

To the Editors:

Monoclonal antibodies directed against tissue factor inhibitor (TFPI) constitute a new therapeutic class to perform prophylaxis in haemophilia A and B (HA/B), with functional activity independent of FVIII and/or FIX. Now that these therapies are being incorporated into routine clinical practice, there seems to be a need for a greater understanding of their pharmacokinetic (PK) behaviour—and what this means for individual patients. These therapies are characterized by a target‐mediated pharmacological disposition (TMDD), a mechanism known for its nonlinear kinetics and characterized by considerable interindividual variability [1, 2].

Typically, PK variability is expressed as a coefficient of variation (CV%), which turns out to be a relative metric. This, in fact, could pose a problem: when two treatments differ substantially in absolute exposure, CV% could be misleading. It could mask a clinically meaningful dispersion that has value for individual patients. In this context, our opinion is that absolute metrics such as standard deviation (SD) could offer additional fundamental insight into the risk of underexposure [3, 4].

In this line, we´ve examined published data from phase 3 trials and regulatory documents of marstacimab and concizumab, the first two available Anti‐TFPI, in people with non‐inhibitor HA/B [1, 5, 6, 7]. We must emphasize that this isn´t a comparison of efficacy between these treatments. Our objective has been to explore how different variability metrics and dose adjustment strategies could condition our interpretation of PK heterogeneity and what potential implications it might have in terms of plasma monitoring.

As hypothesized, the findings reveal that although marstacimab shows a lower CV% than concizumab, when examining absolute SD, marstacimab shows a considerably greater dispersion in steady‐state plasma concentrations (Table 1). We calculated these SD values from reported means and CV% (not from primary patient‐level data), which represents a methodological limitation [1, 5, 7]

TABLE 1.

Pharmacokinetic variability metrics and dosing strategies of anti‐TFPI therapies.

Parameter	Marstacimab	Concizumab	Reference
Mean Cmin, ss (ng/mL) c	13,700 ng/mL (13.7 µg/mL)	724.4 ng/mL (HA) / 554.9 ng/mL (HB)	[1, 5, 7]
CV% (Relative variability) c	90.4%	153% (HA) / 216% (HB)	[1, 5, 7]
SD (Absolute variability, ng/mL) a , c	∼12,385 ng/mL	∼1,108 ng/mL (HA) / ∼1,199 ng/mL (HB)	Calculated a
Dose Adjustment Strategy	Reactive (bleed‐based)	Proactive (PK‐guided)	[1, 5, 7]
Patients maintaining initial dose	∼59.5% b	∼70%	[1, 5, 7]
Patients meeting escalation criteria	40.5%	Not applicable	[1, 5, 7]
Patients receiving dose escalation	29.8%	Not applicable	[1, 5, 7]

^a SD calculated from reported mean and CV% values (SD = CV%/100 × mean Cmin, ss); not derived from primary patient‐level data.

^b Calculated as 100% minus 40.5% that met escalation criteria. Values from phase 3 trials and regulatory documents presented as approximate ranges. The table is descriptive, not for comparison of efficacy between trials.

^c Corrected values derived from approved Summaries of Product Characteristics: Hympavzi (EMA 2024, Section 5.2) and Alhemo (EMA 2024, Section 5.2, Table 8).

[Correction added on 15 June 2026, after first online publication: Table 1 has been updated]

This may underestimate or overestimate dispersion if distributions are skewed or if variability differs across exposure ranges; therefore, our SD values should be interpreted as approximate and hypothesis‐generating rather than definitive. I This is a clear example of how relative metrics alone could mask clinically meaningful variability.

It is known that the clinical development programs of both products presented differences in terms of, for example, the criteria for dose adjustment. In the case of concizumab, a proactive PK‐guided approach was chosen, with early evaluations defined by protocol (4 weeks after initiation of therapy); 70% of people—combined data of EXPLORER7/8 – maintained the starting dose [7]. In contrast, in the case of marstacimab (BASIS trial), a reactive bleeding/time‐based escalation strategy was followed. Finally, while 40.5% of patients met the protocol‐defined escalation criteria, only 29.8% were given a dose increase [5].

This heterogeneity is even more apparent when data related to annualized bleeding rate (ABR) are analysed. For example, when we reviewed the results of haemophiliac patients on routine marstacimab prophylaxis, the 95% upper confidence interval for ABR was 6.78 [5]. This bleeding outcome variability could partly reflect underlying PK heterogeneity, underscoring the potential value of exposure monitoring in patients with suboptimal control.

Taken together, we understand from these observations that a more nuanced framework may be required to interpret PK variability in anti‐TFPI therapies. Both metrics—relative and absolute—must be present, especially when there are such important differences in exposures. In this sense, it could be that proactive adjustment ‐guided by PK‐ allows an early identification of underexposure, while reactive escalation based on bleeding could delay optimization in some patients. Any of these strategies has different implications for how variability translates into clinical outcomes. It should be noted that, unlike concizumab, there is currently no established clinical assay for marstacimab plasma monitoring, nor are there known plans to implement such monitoring in routine practice.

What´s important to note is that we don´t advocate routine plasma monitoring in all people. What we do want to suggest is a justification—generating hypotheses—for performing PK monitoring in selected situations: for example, unexplained bleeding, people in extreme ranges of variability, or in cases of uncertainty about actual exposure. Prospective studies with exposure‐response analysis will be required to establish the clinical utility, significant thresholds, and cost‐effectiveness of PK‐guided strategies.

Funding

The authors have nothing to report.

Ethics Statement

This correspondence analyses published data from clinical trials that received appropriate ethics approvals as reported in the original publications.

Conflicts of Interest

Manuel Rodríguez López has participated in Advisory Boards for Amgen, CSL Behring, Rovi, Sobi, Novo Nordisk, and Octapharma; and as speaker for Pfizer, CSL Behring, Sobi, Octapharma, Novo Nordisk, and Amgen.

Maria Teresa Alvarez Roman has participated in Advisory Boards for Amgen, Pfizer, CSL Behring, Rovi, Sobi, Novo Nordisk, Takeda, and Octapharma; and as speaker for Pfizer, CSL Behring, Sobi, Octapharma, Novo Nordisk, Amgen, and Takeda.

Michael Calviño Suarez has participated in Advisory Boards for Amgen, Rovi, Sobi, and Novo Nordisk; and as speaker for Sobi, Novo Nordisk, Amgen, and Rovi.

Ramiro Núñez Vázquez has participated in Advisory Boards for Amgen, Pfizer, CSL Behring, Rovi, Sobi, Novo Nordisk, Takeda, and Octapharma; and as speaker for Pfizer, CSL Behring, Sobi, Octapharma, Novo Nordisk, Amgen, and Takeda.

Carmen Albo López declares no conflicts of interest.

Data Availability Statement

Data Availability Statement: All data relevant to this correspondence are included in the manuscript and cited references.