Table 1 Model selection for the... | American Society of Hematology

Table 1

Model selection for the determination of apparent affinities

Sample*	R² > 0.7		Lower limit of 95% CI > 0		F test	Selected model	Population 1 K_A [M⁻¹]	Population 2 K_A [M⁻¹]
Sample*	M1	M2	M1	M2	F test	Selected model	Population 1 K_A [M⁻¹]	Population 2 K_A [M⁻¹]
A	Yes	Yes	Yes	No	NA	M1	4.9 × 10⁹	NA
B	Yes	Yes	Yes	No	NA	M1	9.1 × 10⁷	NA
C	Yes	Yes	Yes	Yes	M2	M2	5.4 × 10⁹	9.9 × 10⁷

Sample*	R² > 0.7		Lower limit of 95% CI > 0		F test	Selected model	Population 1 K_A [M⁻¹]	Population 2 K_A [M⁻¹]
Sample*	M1	M2	M1	M2	F test	Selected model	Population 1 K_A [M⁻¹]	Population 2 K_A [M⁻¹]
A	Yes	Yes	Yes	No	NA	M1	4.9 × 10⁹	NA
B	Yes	Yes	Yes	No	NA	M1	9.1 × 10⁷	NA
C	Yes	Yes	Yes	Yes	M2	M2	5.4 × 10⁹	9.9 × 10⁷

An explanation of the model-selection strategy can be found in “Materials and methods.” Model 1 (homogenous apparent affinity distribution), assuming 1 major antibody cluster, resulted in the best description of the FVIII competition curves for samples A and B. In contrast, model 2 (bimodal apparent affinity distribution), assuming 2 major antibody clusters with distinct affinities, provided the best fit to describe the competition curve of sample C.

Sample A: monoclonal human FVIII-specific IgG1 antibody spiked into human plasma not containing any FVIII-binding antibodies; sample B: human plasma from a healthy individual with FVIII-specific IgG1, diluted with human plasma not containing FVIII-binding antibodies; and sample C: a mixture of samples A and B.

CI, confidence interval; M1, model 1; M2, model 2; NA, not applicable; R², coefficient of determination; K_A, apparent affinity.

View Large

This Feature Is Available To Subscribers Only