Dosing of factor VIII concentrate by ideal body weight is more accurate in overweight and obese haemophilia A patients

A Generative and Causal Pharmacokinetic Model for Factor VIII in Hemophilia A: A Machine Learning Framework for Continuous Model Refinement

In rare diseases, such as hemophilia A, the development of accurate population pharmacokinetic (PK) models is often hindered by the limited availability of data. Most PK models are specific to a single recombinant factor VIII (rFVIII) concentrate or measurement assay, and are generally unsuited for answering counterfactual ("what-if") queries. Ideally, data from multiple hemophilia treatment centers are combined but this is generally difficult as patient data are kept private. In this work, we utilize causal inference techniques to produce a hybrid machine learning (ML) PK model that corrects for differences between rFVIII concentrates and measurement assays. Next, we augment this model with a generative model that can simulate realistic virtual patients as well as impute missing data. This model can be shared instead of actual patient data, resolving privacy issues. The hybrid ML-PK model was trained on chromogenic assay data of lonoctocog alfa and predictive performance was then evaluated on an external data set of patients who received octocog alfa with FVIII levels measured using the one-stage assay. The model presented higher accuracy compared with three previous PK models developed on data similar to the external data set (root mean squared error = 14.6 IU/dL vs. mean of 17.7 IU/dL). Finally, we show that the generative model can be used to accurately impute missing data (< 18% error). In conclusion, the proposed approach introduces interesting new possibilities for model development. In the context of rare disease, the introduction of generative models facilitates sharing of synthetic data, enabling the iterative improvement of population PK models.

Application of SHAP values for inferring the optimal functional form of covariates in pharmacokinetic modeling

In population pharmacokinetic (PK) models, interindividual variability is explained by implementation of covariates in the model. The widely used forward stepwise selection method is sensitive to bias, which may lead to an incorrect inclusion of covariates. Alternatives, such as the full fixed effects model, reduce this bias but are dependent on the chosen implementation of each covariate. As the correct functional forms are unknown, this may still lead to an inaccurate selection of covariates. Machine learning (ML) techniques can potentially be used to learn the optimal functional forms for implementing covariates directly from data. A recent study suggested that using ML resulted in an improved selection of influential covariates. However, how do we select the appropriate functional form for including these covariates? In this work, we use SHapley Additive exPlanations (SHAP) to infer the relationship between covariates and PK parameters from ML models. As a case-study, we use data from 119 patients with hemophilia A receiving clotting factor VIII concentrate peri-operatively. We fit both a random forest and a XGBoost model to predict empirical Bayes estimated clearance and central volume from a base nonlinear mixed effects model. Next, we show that SHAP reveals covariate relationships which match previous findings. In addition, we can reveal subtle effects arising from combinations of covariates difficult to obtain using other methods of covariate analysis. We conclude that the proposed method can be used to extend ML-based covariate selection, and holds potential as a complete full model alternative to classical covariate analyses.

Population pharmacokinetic modeling of factor concentrates in hemophilia: an overview and evaluation of best practice

The accuracy of pharmacokinetic (PK)-guided dosing depends on the clinical and laboratory data used to construct a population PK model, as well as the patient’s individual PK profile. This review provides a detailed overview of data used for published population PK models for factor VIII (FVIII) and factor IX (FIX) concentrates, to support physicians in their choices of which model best suits each patient. Furthermore, to enhance detailed data collection and documentation, we do suggestions for best practice. A literature search was performed; publications describing prophylactic population PK models for FVIII and FIX concentrates based on original patient data and constructed using nonlinear mixed-effect modeling were included. The following data were collected: detailed demographics, type of product, assessed and included covariates, laboratory specifications, and validation of models. Included models were scored according to our recommendations for best practice, specifically scoring the quality of data documentation as reported. Respectively, 20 models for FVIII and 7 for FIX concentrates were retrieved. Although most models (22/27) included pediatric patients, only 4 reported detailed demographics. The wide range of body weights suggested that overweight and obese adults were represented. Twenty-six models reported the assay applied to measure factor levels, whereas only 16 models named reagents used. Eight models were internally validated using a data subset. This overview presents detailed information on clinical and laboratory data used for published population PK models. We provide recommendations on data collection and documentation to increase the reliability of PK-guided prophylactic dosing of factor concentrates in hemophilia A and B.

Impact of extreme weight loss on factor VIII concentrate pharmacokinetics in haemophilia

We explored the effects of extreme weight loss after gastric bypass surgery on factor VIII concentrate pharmacokinetic (PK) parameters in a patient with haemophilia A. We present a 32-year-old man with severe haemophilia A, with a body mass index (BMI) of 42.6 kg/m² who underwent laparoscopic sleeve gastrectomy. We showed that a population PK model with ideal body weight as morphometric variable instead of bodyweight led to an adequate description of the individual PKs in this patient with a variable BMI. Strikingly, no differences were observed in the individual PK parameters after extreme weight loss. Therefore, the resulting extreme weight loss after surgery did not lead to prophylactic dose changes in this patient with severe haemophilia. We carefully conclude that population PK-pharmacodynamic models are still obligatory to give more insight into functional effects of significant weight loss on the haemostatic balance.

Dosing of factor VIII concentrate by ideal body weight is more accurate in overweight and obese haemophilia A patients

Aims

Under‐ and, especially, overdosing of replacement therapy in haemophilia A patients may be prevented by application of other morphometric variables than body weight (BW) to dose factor VIII (FVIII) concentrates. Therefore, we aimed to investigate which morphometric variables best describe interindividual variability (IIV) of FVIII concentrate pharmacokinetic (PK) parameters.

Methods

PK profiling was performed by measuring 3 FVIII levels after a standardized dose of 50 IU kg⁻¹ FVIII concentrate. A population PK model was constructed, in which IIV for clearance (CL) and central volume of distribution (V1) was quantified. Relationships between CL, V1 and 5 morphometric variables (BW, ideal BW [IBW], lean BW, adjusted BW, and body mass index [BMI]) were evaluated in normal weight (BMI < 25 kg m⁻²), overweight (BMI 25–30 kg m⁻²) and obese haemophilia A patients (BMI > 30 kg m⁻²).

Results

In total, 57 haemophilia A patients (FVIII≤0.05 IU mL⁻¹) were included with median BW of 83 kg (range: 53–133) and median age of 48 years (range: 18–77). IBW best explained observed variability between patients, as IIV for CL and V1 was reduced from 45.1 to 37.6 and 26.% to 14.1%, respectively. CL, V1 and half‐life were similar for all BMI categories. The national recommended dosing schedule did not result in adequate trough levels, both in case of dosing based on BW and IBW. However, dosing based on IBW prevented unnecessary high FVIII peaks.

Conclusion

IBW is the most suitable morphometric variable to explain interindividual FVIII PK variability and is more appropriate to dose overweight and obese patients.