Use of PopPK and E‐R Analyses toward Explaining Causal Link Between ADAMTS13 in Recombinant vs. Plasma‐Based Therapies and Clinical Effects in cTTP

Abstract

Congenital thrombotic thrombocytopenic purpura (cTTP) is an ultra‐rare, potentially life‐threatening condition caused by a deficiency of the blood enzyme ADAMTS13. Until now, ADAMTS13 replacement has been achieved with infusions of plasma or plasma‐based therapies (PBT). However, the quantitative relationship between ADAMTS13 plasma activity and clinical manifestations is poorly understood. We therefore conducted integrated population pharmacokinetics (PopPK) analysis and exposure–response modeling based on three clinical trials of recombinant ADAMTS13 (rADAMTS13, Takeda Pharmaceuticals U.S.A., Inc.) in patients with cTTP. These aim to assess the clinical benefit of rADAMTS13, which at the proposed dose of 40 IU/kg provides ADAMTS13 peak levels of approximately 100% of normal levels. The PK model indicated that, besides body weight–based dosing, no further dose adjustment was required based on age or race. The only extrinsic covariates with a significant impact on ADAMTS13 plasma activity levels were dosing interval and treatment type (rADAMTS13 vs. PBT). The correlation between ADAMTS13 plasma activity levels and cTTP manifestations was investigated with three different exposure–response models. Increasing exposure to ADAMTS13, as measured by average activity over a one‐to‐two‐week period, predicted the probability of disease manifestations, primarily assessed as thrombocytopenia. Model simulations predicted that >90% of patients treated with 40 IU/kg rADAMTS13 achieve an average ADAMTS13 plasma activity >13% of normal, which was shown to be highly protective against thrombocytopenia (>70% lower hazard). Similar results were observed for protection against elevation of lactate dehydrogenase, a marker of microangiopathic hemolytic anemia. Overall, these results support the use of rADAMTS13 treatment for patients with cTTP.

Study Highlights.

• WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC?

Congenital thrombotic thrombocytopenic purpura (cTTP) is an ultra‐rare, potentially life‐threatening condition caused by a deficiency of the blood enzyme ADAMTS13. Current plasma‐based therapies (PBT) provide ADAMTS13 supplementation to only about 20% of normal levels in cTTP. The recent availability of recombinant ADAMTS13 (rADAMTS13) allows for improved ADAMTS13 replacement to approximately 100% of normal levels.

• WHAT QUESTION DID THIS STUDY ADDRESS?

This analysis produced novel and long‐awaited insights into the pharmacokinetic parameters of plasma ADAMTS13 activity that are expected to be protective against ongoing cTTP disease activity. First, the effect of key intrinsic and extrinsic factors affecting the pharmacokinetics of plasma ADAMTS13 activity was quantified. Second, the unknown relationship between ADAMTS13 activity and clinical manifestations following intravenous administration of rADAMTS13 or PBT in patients with cTTP was quantified and characterized.

• WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE?

Higher ADAMTS13 activity exposure achieved with rADAMTS13 treatment (40 IU/kg once every week or once every 2 weeks) correlates with greater protection against thrombocytopenia compared with PBT. Additionally, rADAMTS13 demonstrates consistent clinical efficacy regardless of age, body weight, or ethnicity.

• HOW MIGHT THIS CHANGE CLINICAL PHARMACOLOGY OR TRANSLATIONAL SCIENCE?

Pharmacometrics approaches have for the first time been conducted to supplement and support the assessment of clinical benefit of rADAMTS13 in the cTTP patient population. Thus, by performing this integrated analysis, we were able to establish the causal link between ADAMTS13 activity and the clinical effects of rADAMTS13 compared with PBT, a finding that could substantially impact the treatment and management of cTTP.

Congenital thrombotic thrombocytopenic purpura (cTTP) is a severe, ultra‐rare genetic deficiency of “a disintegrin and metalloproteinase with thrombospondin motifs 13” (ADAMTS13) enzyme. The diagnosis of cTTP is typically confirmed by ADAMTS13 activity <10% of normal in the absence of an acquired inhibitor ¹ , ² and has an estimated prevalence ranging between 0.5 and 2 cases per million population, although the prevalence may be considerably higher in certain regions. ³ While disease onset can occur anytime, two incidence peaks have been reported at birth and during pregnancy. ⁴ , ⁵ ADAMTS13 is secreted into the blood and cleaves von Willebrand factor (VWF) multimers into smaller and less biologically active units. ⁶ Deficiency of ADAMTS13 can cause accumulation of large and ultra‐large VWF multimers with high platelet‐binding activity, leading to platelet aggregation and thrombi formation in the microvasculature. In line with this, acute exacerbations of TTP are often complicated by organ ischemia that can be life‐threatening in the absence of treatment. ⁵ , ⁷

To date, treatment of cTTP has consisted of ADAMTS13 replacement using plasma‐based therapies (PBT; plasma infusions or plasma‐derived factor VIII/von Willebrand factor concentrates [pdFVIII:VWF]), ¹ , ⁷ , ⁸ either for treatment of an acute event or as long‐term prophylaxis. Due to limitations in infusion volume and low ADAMTS13 levels in the preparations, PBT achieves a maximum of only approximately 20% to 30% of normal ADAMTS13 activity (1 IU/mL plasma), with high variability in actual replacement levels. ⁹ , ¹⁰ In addition, PBT carries a high treatment burden, with high infusion volumes (approximately 10 mL/kg) and a high incidence of hypersensitivity reactions that may be severe and require premedication. ¹¹

Due to the rarity of cTTP and the limited treatment options available to date, there is insufficient evidence for the management of cTTP. In particular, there are no quantitative data on ADAMTS13 exposure and its relationship to cTTP manifestations to provide clinical guidance for adequate prophylaxis in cTTP. ³ , ⁵ , ⁹ While endogenous ADAMTS13 levels <10% of normal are generally considered to be severely deficient, ¹² the clinically relevant target for optimal prophylaxis remains unknown. Furthermore, it is not known which pharmacokinetic (PK) metric (e.g., trough, maximum, or average activity levels) is most relevant to guide clinical management. ¹³ Additionally, the clinical goal of prophylactic treatment is evolving beyond just the prevention of acute events toward the prevention of non‐overt symptoms and long‐term organ damage. Recent data suggest that even in the absence of acute events, there can be ongoing disease activity such as chronic, subacute platelet consumption, and recurrent microangiopathic hemolytic anemia (MAHA). ⁵ , ¹⁴

To address these unresolved questions about ADAMTS13 activity, its exposure, and its association with clinical manifestations of cTTP, we conducted an integrated population PK (PopPK) and exposure–response analysis as part of the interim analysis of an ongoing phase III, randomized, controlled, open‐label, crossover study of recombinant ADAMTS13 (rADAMTS13) in patients with cTTP (NCT03393975). ¹⁰ rADAMTS13 (Takeda Pharmaceuticals U.S.A., Inc.) is an enzyme replacement therapy using the mature, fully glycosylated, human ADAMTS13 produced in a CHO cell line. ¹⁵ , ¹⁶ An intravenous dose of 40 IU/kg of rADAMTS13 in cTTP patients ≥12 years old resulted in a C _max of ~1.15 IU/mL, with an AUC₀₋₁₆₈ₕ of 57.2 IU·h/mL and mean (SD) incremental recovery of 0.03 (0.01) (IU/mL)/(IU/kg). The terminal half‐life of rADAMTS13 was ~47.8 hours. ¹⁵ Because much higher replacement levels can be achieved with rADAMTS13 than with PBT, ¹⁰ we could explore the exposure–response relationship between ADAMTS13 activity and the probability of observing decreases in platelet count and increases in lactate dehydrogenase (LDH; as a marker of MAHA). Using a PopPK approach, we also characterized the PK of ADAMTS13 activity in patients of all ages following intravenous administration of rADAMTS13 or PBT to identify potential covariates to explain any variability that might be observed.

METHODS

Patients and data collection

Data for the PopPK analysis were obtained from three clinical trials of rADAMTS13 (TAK‐755; BAX930; SHP655) in patients with cTTP. These included a phase I dose‐escalation PK/safety study in adults and adolescent patients with cTTP (NCT02216084). ¹⁷ The second trial was a phase III randomized crossover, open‐label, efficacy, and safety study in patients with cTTP (NCT03393975). ¹⁰ In that trial, patients received either rADAMTS13 (40 IU/kg once every 1 or 2 weeks) or PBT for the first 6 months (Period 1), and then crossed over to the alternative treatment for a further 6 months (phase II), followed by an additional 6 months (Period 3), in which all patients received rADAMTS13. The third trial was a phase IIIb open‐label, continuation study in patients with cTTP (NCT04683003). ¹⁸ , ¹⁹ Both phase III trials enrolled patients aged 0–70 years. For the PopPK analysis, ADAMTS13 activity was measured using the fluorescence resonance energy transfer substrate (FRETS)‐VWF73 assay. ²⁰ Samples with ADAMTS13 activity below the limit of quantitation were set to missing in the PopPK analysis.

All trials that contributed data to the research reported herein were approved by their respective institutional review boards or independent ethics committees at all participating sites (Table S1), and patients or their legal guardian/parent gave written informed consent. The trials were conducted in accordance with the Declaration of Helsinki.

PopPK analysis

A PopPK analysis was conducted to assess the influence of covariates on the PK characteristics of ADAMTS13 activity. Key covariates were selected based on typical popPK covariate consideration, ²¹ understanding of extrinsic factors influencing plasma therapies, and knowledge of the disease biology of cTTP and ADAMTS13. Clinical and demographic covariates consisted of age, sex, race/ethnicity, height and weight, estimated body surface area, laboratory parameters (including hematocrit, hemoglobin, alanine aminotransferase, aspartate aminotransferase, albumin, alkaline phosphatase, creatinine, and bilirubin), estimated glomerular filtration rate, and blood group type. Age is an important covariate in this study to ensure appropriate dosing across age groups, including pediatric patients. In addition, the effect of blood group is relevant to rADAMTS13 owing to potential interactions between blood group antigens and VWF. ²² Key extrinsic factors evaluated were active treatment (rADAMTS13 vs. PBT) and also the type of PBT (fresh frozen plasma, solvent/detergent‐treated plasma, or pdFVIII:VWF concentrates) as different PBTs can have highly variable rADAMTS13 content. ¹⁰ Forest plots were created to visualize the predicted covariate effects with uncertainty on PK exposure parameters of ADAMTS13 by changing one covariate at a time while fixing others at the reference values.

In the three clinical trials of rADAMTS13, the dose of rADAMTS13 was administered based on actual body weight, and body weight was incorporated a priori in the model using a power function centered to the median body weight (68.7 kg). Fixed allometric exponents were used to describe the effect of body weight on clearance and central volume of distribution (Figure S1 ). Final parameters estimated from the PopPK model and key model equations are presented in Table 1 . The PopPK model analysis followed best pharmacometrics practice processes and procedures for population modeling. ²³ , ²⁴ Additional details are reported below and in Supplementary Methods S1 . The performance of the final PopPK model was evaluated based on residual‐based diagnostic plots (goodness‐of‐fit), model convergence and precision of parameter estimates, random effect shrinkage, and prediction‐corrected visual predictive checks. Nonlinear mixed‐effects modeling (NONMEM) was performed using the first‐order conditional estimation approach or using the Laplacian method in NONMEM (Version VII; ICON plc, Dublin, Ireland).

Table 1.

Parameter estimates from the population pharmacokinetic model of rADAMTS13

Parameters	Estimate	RSE%	IIV, % (RSE%)a	Shrinkage, %	Bootstrap median	Bootstrap 95% CI
Clearance, L/h	0.0398	10.7	36.3 (45.8)	14.3	0.0402	0.0288–0.0506
	× (WT/68.7)0.75	–	–	–	–	–
Central volume of distribution, L	2.69	4.99	25.4 (20.9)	4.9	2.71	2.35–3.00
	× (WT/68.7)1.0	–	–	–	–	–
Peripheral clearance, L/h	0.0456	12.1	–	–	0.0487	0.0272–0.487
	× (WT/68.7)0.75	–	–	–	–	–
Peripheral volume of distribution, L	3.71	51.4	–	–	3.43	1.04–19.9
	× (WT/68.7)1.0	–	–	–	–	–
Relative ADAMTS13 activity	1, Fixed	–	–	–	–	–
	× (1–0.390) if PBT	6.14	–	–	−0.374	−0.427 to −0.270
	× (1–0.933) if FVIII:VWF concentrates	2.06	–	–	−0.915	−0.986 to −0.828
Error model	Additive (IU/L): 79.9	16.4	–	–	75.9	41.0–108.0
	Proportional (Fraction): 0.204	14.7	–	–	0.203	0.124–0.275

ADAMTS13, a disintegrin and metalloproteinase with thrombospondin motifs 13; CI, confidence interval; FVIII, factor VIII; IIV, interindividual variability; PBT, plasma‐based therapy; rADAMTS13, recombinant ADAMTS13; RSE, relative standard error; VWF, von Willebrand factor; WT, weight.

The reference patient is a 68.7‐kg patient who received rADAMTS13. Objective Function = 27068.413. The condition number (229) was low (typically interpreted as <1,000), suggesting that the model was numerically identifiable.

^a IIV is presented as a % coefficient of variation derived as (100 × [exp(ω²)–1]^0.5).

Exposure–response analyses

The exposure–response analyses were based on data from the cTTP pivotal phase III study (NCT03393975). ¹⁰ The analyses focused on thrombocytopenia (platelet count decreased ≥25% from baseline or a platelet count <150,000/μL) and elevated LDH (to >1.5 times the baseline value or >1.5 times the upper limit of normal) as a marker of MAHA. Further details of the methodology are published elsewhere. ²⁵ , ²⁶ Additional characteristics of the exposure–response analyses are presented in Figure 1 and the Supplementary Methods S1 .

Figure 1.

General framework for repeated time‐to‐event manifestations. ADAMTS13, a disintegrin and metalloproteinase with thrombospondin motifs 13; C _ave, average ADAMTS13 activity; C _trough, trough ADAMTS13 activity; E _max, maximum effect; FOCE, first‐order conditional estimation; LAPLACE, Laplacian method; LDH, lactate dehydrogenase; NONMEM, nonlinear mixed‐effects modeling; PK, pharmacokinetic. ^aTwo patients included in the exploratory data analysis and repeated time‐to‐event modeling were not considered part of the prophylaxis cohort in the count exposure–response analysis.

Exposure–response count modeling

Count modeling was used to determine how the hazard of occurrence of TTP events varied with specific factors (covariates), including treatment effect (rADAMTS13 vs. PBT). In this approach, events were counted within time intervals, so the exact time of occurrence was not taken into account in the analysis. ²⁵ The observed count distributions for thrombocytopenia and elevated LDH events in Periods 1 and 2 of the cTTP phase III study ¹⁰ were described by Poisson, zero‐inflated Poisson, negative binomial, and zero‐inflated negative binomial models. The effect of average ADAMTS13 plasma activity for a given dosing interval (C _ave) was then evaluated as a covariate in the model using linear, exponential, and sigmoidal maximum effect (E _max) drug exposure–response relationships. The C _ave values were derived after each dose in each patient, and a mean C _ave per patient value was derived within Periods 1 and 2, respectively. The mean C _ave is representative of the mean exposure to ADAMTS13 activity over time in each patient in prophylactic Periods 1 and 2. Count modeling was performed initially in adolescents and adults (age ≥ 12 years). ¹⁰ The analysis was then repeated without age restrictions. The final model was evaluated by comparing the model‐predicted count values against the observed counts.

Cox‐proportional‐hazards exposure–response modeling

A semiparametric analysis was performed based on thrombocytopenia and elevated LDH events collected in Periods 1 and 2 of the cTTP phase III study ¹⁰ in patients aged <12 and ≥12 years. This approach has the advantage of avoiding the need to estimate the baseline hazard function, while a key limitation is that the hazard ratio is assumed to remain constant throughout the follow‐up. ²⁷ , ²⁸ The PopPK model was used to predict individual longitudinal ADAMTS13 activity–time profiles based on actual dosing information and individual parameter estimates after accounting for between‐patient variability. The individual longitudinal ADAMTS13 activity profiles were merged with a dataset including individual events of thrombocytopenia and elevated LDH. The C _ave since the start of treatment to the first event and between events was included in the dataset for each patient in Periods 1 and 2.

Longitudinal repeated time‐to‐event exposure–response modeling

A parametric repeated time‐to‐event exposure (RTTE) model driven by time‐varying ADAMTS13 activity was developed to assess the probability of thrombocytopenia and elevated LDH based on predicted ADAMTS13 activity during the phase III study. ¹⁰ This approach allows the time‐varying probability of events to be correlated with longitudinal ADAMTS13 activity predictions. A disadvantage is that it is based on the assumption that the underlying population distribution is correctly specified, and so is vulnerable to misspecification. ²⁹ , ³⁰ The analysis was based on a model with constant baseline hazard combined with a sigmoid E _max model describing the effect of daily C _ave. Longitudinal data were used from all three prophylactic periods to demonstrate ADAMTS13‐dependent reduction in the hazard of thrombocytopenia and elevated LDH. The RTTE model was qualified by visual predictive checks. Simulations were performed to determine the probability of being event‐free at 6 months and 1 year based on typical ADAMTS13 activity for once every week (Q1W) and once every 2 weeks (Q2W) dosing with rADAMTS13 and PBT.

RESULTS

PopPK analysis

The PopPK analysis included 65 unique patients across the three cTTP trials (NCT02216084, NCT03393975, and NCT04683003) who received at least one dose of either rADAMTS13 or PBT, from whom 2,462 samples with measurable ADAMTS13 activity were obtained. The ADAMTS13 content across various PBTs, including pdFVIII:VWF concentrates, was estimated to be 40% to 94% lower than that measured in rADAMTS13. The baseline characteristics of the 65 patients included in the PK analysis are shown in Table 2.

Table 2.

Baseline characteristics of the population pharmacokinetic analysis set

Demographics	Phase Ia (n = 15)	Phase IIIb (n = 48)	Phase IIIbc (n = 36)	Overall (N = 65)
Age group
<12 years	0	8 (16.7)	0	8 (12.3)
<6 years	0	4 (8.3)	0	4 (6.2)
6 to <12 years	0	4 (8.3)	0	4 (6.2)
≥12 years	15 (100)	40 (83.3)	36 (100)	57 (87.7)
12 to <18 years	2 (13.3)	4 (8.3)	1 (2.8)	6 (9.2)
≥18 years	13 (86.7)	36 (75.0)	35 (97.2)	51 (78.5)
Sex
Female	8 (53.3)	29 (60.4)	23 (63.9)	39 (60)
Male	7 (46.7)	19 (39.6)	13 (36.1)	26 (40.0)
Race
White	12 (80.0)	31 (64.6)	23 (63.9)	40 (61.5)
Black/African American	1 (6.7)	1 (2.1)	0	2 (3.1)
Asian	2 (13.3)	6 (12.5)	8 (22.2)	11 (16.9)
Multiple	0	1 (2.1)	0	1 (1.5)
Missing	0	9 (18.8)	5 (13.9)	11 (16.9)
Chinesed
Yes	0	0	3 (8.3)	3 (4.6)
No	15 (100)	48 (100)	33 (91.7)	62 (95.4)
Japanesed
Yes	2 (13.3)	5 (10.4)	5 (13.9)	7 (10.8)
No	13 (86.7)	43 (89.6)	31 (86.1)	58 (89.2)
Hispanic/Latino ethnicity
Yes	0	1 (2.1)	1 (2.8)	1 (1.5)
No	15 (100)	39 (81.3)	30 (83.3)	54 (83.1)
Missing	0	8 (16.7)	5 (13.9)	10 (15.4)
Blood group
O	2 (13.3)	20 (41.7)	12 (33.3)	22 (33.8)
A	9 (60.0)	13 (27.1)	13 (36.1)	24 (36.9)
B	3 (20.0)	8 (16.7)	5 (13.9)	10 (15.4)
AB	1 (6.7)	7 (14.6)	5 (13.9)	8 (12.3)
Missing	0	0	1 (2.8)	1 (1.5)
Platelets (109/L)
Mean (SD)	205 (67.4)	228 (93.1)	199 (84.4)	218 (88.0)
Median (range)	203 (30.0–315)	214 (25.0–442)	199 (20.0–344]	205 (20.0–442)
LDH (U/L)
Mean (SD)	160 (75.3)	217 (96.4)	228 (168)	222 (141)
Median (range)	137 (106, 405)	187 (106, 598)	178 (106, 1,030)	178 (106, 1,030)
Body weight, kg
Mean (SD)	75.0 (28.1)	65.8 (21.7)	76.0 (18.3)	70.0 (24.7)
Median (range)	69.0 (43.0–127.0)	68.6 (18.3–104.0)	75.0 (43.0–130.0)	68.7 (18.3–130.0)

LDH, lactate dehydrogenase; SD, standard deviation.

Data are n (%) unless otherwise stated.

^a ClinicalTrials.gov: NCT02216084.

^b ClinicalTrials.gov: NCT03393975.

^c ClinicalTrials.gov: NCT04683003.

^d Based on the country of origin.

The PK of ADAMTS13 activity in patients with cTTP was described using a two‐compartment model with zero‐order infusion and first‐order linear elimination from the central compartment. The base model included allometric scaling factors on clearance and volume of distribution parameters, with covariate evaluation showing that further inclusion of age and body weight had no impact on maximum ADAMTS13 activity (C _max; Figure 2 a ) and ADAMTS13 C _ave activity (Figure 2 b ). Although there was a slight trend for an increase in C _ave with increasing age and body weight, this was not statistically significant (Figure 2 b ). Following covariate evaluation, dosing interval (Q1W vs. Q2W) and treatment type (rADAMTS13 vs. PBT) were the only extrinsic factors with a significant impact on the PK of ADAMTS13 (Figure 2 b ).

Figure 2.

Effect of key covariates on (a) maximum and (b) average ADAMTS13 activity. All data are based on Q2W dosing with the Q1W dosing interval shown for comparison and are derived from 1,000 virtual subjects. ADAMTS13, a disintegrin and metalloproteinase with thrombospondin motifs 13; CI, confidence interval; PBT, plasma‐based therapy; Q1W, once every week; Q2W, once every 2 weeks; rADAMTS13, recombinant ADAMTS13. ^aReference patient is a 68.7‐kg adult receiving rADAMTS13 40 IU/kg Q2W.

The observed median, 5^th, and 95^th percentiles of ADAMTS13 activity of rADAMTS13 or PBT were consistent with the PopPK model‐predicted values in the cTTP phase III study (NCT03393975) ¹⁰ (Figure 3 ). In addition, there was good agreement between observed ADAMTS13 activities and both individual and population model–predicted activities (Figure  S2 ). A summary of the final PopPK model–based steady‐state PK parameters for ADAMTS13 activity following intravenous administration of rADAMTS13 and PBT is provided in Table  S2 , with additional details presented in Tables  S3 and S4 . Overall, these PK parameters indicate that rADAMTS13 is expected to deliver consistently higher exposures compared with PBT.

Figure 3.

Prediction‐corrected visual predictive checks: (a) PBT; (b) rADAMTS13 (phase III study). The dashed black lines represent the 5^th and 95^th percentiles of ADAMTS13 activity, and the solid black line represents the median (50^th percentile) of ADAMTS13 activity. The red solid lines are the model‐predicted 5^th and 95^th percentiles, and the shaded red area represents their respective 95% PI. The blue solid line is the model‐predicted median 50^th percentile, and the shaded blue area represents the 95% PI. “Other” is the remaining PBT types in the PopPK methods dataset. ADAMTS13, a disintegrin and metalloproteinase with thrombospondin motifs 13; FFP, fresh frozen plasma; PBT, plasma‐based therapy; pd FVIII/VWF, plasma‐derived factor VIII/ von Willebrand factor; PI, percentile interval; PK‐I, pharmacokinetic infusion I; PopPK, population pharmacokinetics; rADAMTS13, recombinant ADAMTS13; S/D, solvent/detergent.

Exposure–response analyses

The estimated parameters of the three exposure–response models are presented in Tables S5 and S6 .

Exposure–response count modeling

The Poisson model with random effect and sigmoidal E _max drug effect exposure–response relationship for the effect of ADAMTS13 C _ave resulted in an adequate characterization of count zero as well as the long‐tailed distribution of PBT (Figure  S3 ).

The C _ave of ADAMTS13 activity was associated with a significant reduction of the thrombocytopenia hazard (risk) in an activity‐dependent manner (Figure 4 ). For patients ≥12 years, the exposure–response relationship for the C _ave of ADAMTS13 activity was steep (sigmoidicity factor, gamma of 5.36) with an E _max corresponding to a 86.4% reduction of thrombocytopenia counts; the estimated average ADAMTS13 activity associated with a 50% reduction in thrombocytopenia counts (EC_ave50) was 0.0296 IU/mL (or 3% activity). When including patients of all ages, a similar relationship between ADAMTS13 C _ave and thrombocytopenia was observed (EC_ave50: 0.0149 IU/mL; E _max: 91.8% reduction in thrombocytopenia event counts; gamma: 2.58). Over 90% of patients treated with 40 IU/kg rADAMTS13 (Q2W or Q1W) were predicted to have ADAMTS13 C _ave >0.13 IU/mL (13% activity), and this was associated with >70% protection against thrombocytopenia (regardless of Q2W or Q1W) (Figure 5 ). A similar level of protection was expected to be observed in only about <10% of cTTP patients treated with PBT (administered as 10 IU/kg).

Figure 4.

Exposure (average ADAMTS13 activity)‐response (hazard for thrombocytopenia count) relationship for thrombocytopenia count. The final Poisson model with random effect and sigmoidal E _max drug effect adequately described the observed thrombocytopenia counts. The x‐axis represents the distribution (histogram) of the average ADAMTS13 activity values from Periods 1 and 2 of the phase III study. ¹⁰ The left y‐axis represents the model‐predicted hazard (solid blue line) or risk of thrombocytopenia. The right y‐axis represents the total number of patients falling under each bin of exposure (highlighted on the x‐axis). Data are from patients of all ages. ADAMTS13, a disintegrin and metalloproteinase with thrombospondin motifs 13; C _ave, average ADAMTS13 activity; E _max, maximum effect; PBT, plasma‐based therapy; Q1W, once every week; Q2W, once every 2 weeks; rADAMTS13, recombinant ADAMTS13.

Figure 5.

C _ave exposure distribution and model‐predicted probability of zero thrombocytopenia events by percentiles of C _ave (N = 41). ADAMTS13, a disintegrin and metalloproteinase with thrombospondin motifs 13; C _ave, average ADAMTS13 activity; P, percentile; Q1W, once every week; Q2W, once every 2 weeks; rADAMTS13, recombinant ADAMTS13.

The relationship between the C _ave of ADAMTS13 and the count or hazard of elevated LDH is presented in Figure  S4 . Over 90% of patients treated with 40 IU/kg rADAMTS13 (Q2W or Q1W) were predicted to have ≥80% probability of MAHA reduction (Table  S7 ). Taken together, these results suggest that the dosing of rADAMTS13 at 40 IU/kg (Q2W or Q1W) consistently resulted in average ADAMTS13 activity levels that are predicted to saturate the activity‐effect relationship in this model.

Cox proportional hazards exposure–response modeling

A statistically significant exposure–response was observed (P < 0.001), whereby higher exposure to ADAMTS13 C _ave was associated with a lower risk of thrombocytopenia (Figure  S5 ). rADAMTS13 was predicted to have >70% protection against thrombocytopenia relative to PBT, regardless of dosing interval. Similar results were observed for elevated LDH (Table  S8a ).

Longitudinal repeated time‐to‐event exposure–response modeling

Using data from Periods 1 to 3 and all age groups, a parametric RTTE analysis also identified a significant prophylactic treatment effect of rADAMTS13 over PBT for thrombocytopenia events (Figure  S6 ). The exposure–response relationship predicted an almost complete maximal inhibition of hazard with an E _max corresponding to a maximum 97.5% reduction in the hazard of thrombocytopenia and an EC_ave50 of 0.0113 IU/mL. Similar results were observed for elevated LDH (Table  S8b ).

DISCUSSION

Data presented here and in the original phase III study article ¹⁰ show that rADAMTS13 at a dose of 40 IU/kg (Q2W or Q1W) consistently results in ADAMTS13 activity levels reaching approximately 100% of normal at maximum and remaining above ≥10% for approximately 5 days. ¹⁵ , ¹⁶ The integrated analysis presented here demonstrates for the first time a clear significant relationship between the ADAMTS13 plasma activity and clinical outcomes, with a higher probability of protection against thrombocytopenia and MAHA associated with higher ADAMTS13 plasma activity levels. With the approvals of rADAMTS13 for on‐demand and prophylactic treatment of cTTP in the United States, Europe, and Japan, an ADAMTS13 enzyme replacement therapy allowing for improved control and prevention of cTTP manifestations is now available for the first time. ¹⁵ , ¹⁶ , ³¹

In the PopPK analysis, parameter estimates were generally consistent with previously reported PK parameters. For example, Scully et al. (2024) ¹⁰ reported a mean C_max of 1.01 IU/mL and a mean time above 10% of 5.2 days for rADAMTS13, as well as a mean C_max of 0.19 IU/mL and a mean time above 10% of 1.7 days for PBT. Similarly, Taylor et al. (2019) observed a C_max of 0.241 IU/mL for PBT. ⁹ Three complementary exposure–response approaches (count modeling, Cox proportional hazards modeling, and longitudinal RTTE) were leveraged in this study. Regardless of the exposure–response approach used, a significant ADAMTS13 activity–response relationship was identified in which rADAMTS13 reduced the hazard of objective laboratory‐based TTP manifestations (thrombocytopenia and increased LDH) in an activity‐dependent manner. The significant treatment effect between rADAMTS13 and PBT and the ADAMTS13 activity‐dependent response was generally similar and consistent across the three exposure–response approaches.

This study represents the first attempt to develop and understand the exposure–response relationship between ADAMTS13 and platelet count in cTTP. Exposure–response modeling was important for supporting clinical dosing considerations in cTTP. Given that cTTP is an ultra‐rare disease, conducting a clinical efficacy study with multiple dose levels would not have been feasible. Although there is currently no consensus on the minimal or optimal ADAMTS13 threshold levels for prophylaxis in cTTP, an rADAMTS13 dose of 40 IU/kg achieved normal peak ADAMTS13 levels (~100%) in this analysis. In comparison, in a single‐dose phase I study, 20 IU/kg resulted in peak levels of ~40%, but the associated efficacy was not evaluated in that study. ¹⁷ Therefore, while the phase III pivotal trial evaluated the efficacy of a 40 IU/kg prophylactic dose, ¹⁰ it alone did not enable the identification of an optimal dosage. The present analysis leveraged the similar mechanism of action of rADAMTS13 and PBT (i.e., replacement of ADAMTS13 in patients with cTTP) to estimate the probability of protection against thrombocytopenia and MAHA event rates over a range of ADAMTS13 levels.

The findings from this analysis suggest that increasing average ADAMTS13 activity reaches a plateau in reducing the hazard of thrombocytopenia and thus provides insights into the target ADAMTS13 level required to prevent subclinical disease. Model simulations indicated that >90% of patients treated with rADAMTS13 are predicted to achieve high protection against thrombocytopenia (>70% lower hazard), which corresponds to an average ADAMTS13 activity >13% within a dosing interval. Trough levels were likely undetectable, indicating that a trough level > 13% is not required for a high level of protection against thrombocytopenia with 1‐ or 2‐week dosing intervals. It also suggests that in the actual scenario of the detectable trough level, the threshold can possibly be in the range of <10%. Nevertheless, although no statistically significant differences were observed between Q1W and Q2W dosing regimens, count modeling and Cox proportional hazard analyses (but not RTTE analysis) showed numerical differences favoring Q1W dosing, which were attributed to higher ADAMTS13 activity exposures.

In line with the linear PK characteristics for ADAMTS13 activity following body weight–based dosing, the dosing interval and treatment type were the only significant extrinsic factors found to affect ADAMTS13 activity. When compared with PBT, the higher and more consistent delivery of ADAMTS13 activity following administration of rADAMTS13 resulted in a significant reduction in the probability of thrombocytopenia and elevated LDH. This was supported by generally comparable estimated EC_ave50 in a count analysis based on data from Periods 1 and 2 regardless of age (<12 or ≥ 12 years). Furthermore, the RTTE analysis showed that a persistent and similar ADAMTS13 activity–hazard reduction relationship was observed in the three efficacy prophylaxis periods, encompassing an additional 6 months of rADAMTS13 treatment exposure. The exposure–response relationship predicted an almost complete maximal inhibition of hazard with an E _max corresponding to a maximum 97.5% reduction in the hazard of thrombocytopenia.

The PopPK analysis predicted that body weight–based dosing would result in approximately 20% to 30% lower average exposures in the youngest patients with body weight <10–15 kg vs. typical adults. However, the reduction in the predicted risk of thrombocytopenia or elevated LDH remained on the plateau of the overall exposure–response relationship curve, which suggests that the pediatric C _ave exposures compared with adults will still result in a comparable clinical response (protection against thrombocytopenia and MAHA).

The main limitation of this study was the small sample size, particularly for longitudinal exposure–response modeling. In addition, the number of PK samples was very limited for patients aged <12 years. The lack of dose‐ranging efficacy studies with rADAMTS13, although common for ultra‐rare diseases, is another limitation of this study. Pooling of data from patients after dosing with rADAMTS13 and PBT represents an attempt to overcome this limitation. This approach assumes that no component of plasma other than ADAMTS13 contributes to its efficacy. As the only known mechanism of action of any PBT in the treatment of cTTP is the replacement of ADAMTS13, this assumption is considered reasonable. However, other contributions cannot be excluded.

While the data included in this analysis may not precisely establish the lower PK threshold of ADAMTS13 plasma activity associated with protection against cTTP manifestations, the data clearly show that increased average ADAMTS13 levels (for a given dosing interval of Q2W or Q1W) are favorable and lead to a decreased risk for thrombocytopenia and LDH increase. Furthermore, the data demonstrate that highly protective plasma levels are unlikely to be achieved using PBT, consistent with the phase III study analysis. ¹⁰

Finally, it may be noted that there is an interindividual variability component both in ADAMTS13 activity exposure and the degree to which patients respond clinically. Thus, from a clinical point of view, these findings support that a majority of patients can start treatment with rADAMTS13 at a dose of 40 IU/kg given Q2W. Depending on the clinical response, assessed as cTTP manifestations including thrombocytopenia events, dose frequency may be adjusted to Q1W to take an additional advantage of the higher average ADAMTS13 activity exposures. The availability of rADAMTS13 represents a major milestone in the management of patients with cTTP suffering from this devastating and potentially life‐threatening condition.

In conclusion, the findings of these PopPK and three different sets of exposure–response analyses suggest that across age cohorts (adults, adolescents, and pediatrics), rADAMTS13 at 40 IU/kg Q2W or Q1W has a favorable prophylactic treatment effect compared with PBT, which is the result of higher ADAMTS13 content and plasma activity exposures.

FUNDING

The study was funded by Takeda Development Center Americas, Inc., Cambridge, Massachusetts, USA.

CONFLICT OF INTEREST

H.X., P.P., A.Z.X.Z., L.T.W., W.W., and I.B. are employees of Takeda Development Center Americas, Inc. and Takeda stockholders. M.P. was an employee of Takeda Development Center Americas, Inc. at the time this work was done. B.M. was an employee at the time this work was done and is currently a consultant at Takeda. P.D., J.F.M., and T.P. are employees of Certara Strategic Consulting and Certara stockholders. O.B. was an employee of Certara at the time of the analysis. Certara received research funding to perform this pharmacometrics analysis from Takeda Development Center Americas, Inc.

AUTHOR CONTRIBUTIONS

All authors wrote the manuscript and analyzed the data. M.P., A.Z., I.B., W.W., and B.M. designed the research.

Supporting information

Data S1.

DATA AVAILABILITY STATEMENT

The datasets, including the redacted study protocol, redacted statistical analysis plan, and individual participant data supporting the results reported in this article, will be made available within 3 months from the initial request to researchers who provide a methodologically sound proposal. The data will be provided after its deidentification, in compliance with applicable privacy laws, data protection, and requirements for consent and anonymization. The study sponsor analyzed the data and conducted the statistical analyses. Data outputs were shared and discussed with all authors.