Hemophilia A (HA) is a rare bleeding disorder caused by a lack of functional coagulation
factor VIII (FVIII). Treatments for patients with HA (PwHA) can be achieved by administration
of the missing functional FVIII. Other coagulation factors called bypassing agents
are also used for PwHA with inhibitors. Emicizumab is a humanized bispecific monoclonal
antibody which mimics the cofactor function of FVIIIa by binding to activated factor
IX (FIXa) and factor X (FX), and it is used for routine prophylaxis in PwHA with or
without inhibitors to prevent bleeding.[1]
[2] Multiple clinical studies have confirmed the efficacy of emicizumab,[3]
[4]
[5]
[6]
[7] although treatment with monoclonal antibodies may result in the development of antidrug
antibodies (ADAs).[8] A recent report revealed that in seven phase 3 trials of emicizumab, 34 of 668 PwHA
developed ADAs against emicizumab, and 4 of them developed neutralizing ADAs with
decreased emicizumab exposure.[9] Decreased emicizumab efficacy due to ADAs has also been reported.[10]
[11]
[12]
[13]
[14] In these cases, decreased efficacy was caused by neutralizing and/or clearing abilities
of the ADAs. However, the characteristics of ADAs have not been well understood. In
this research, we aimed to evaluate the characteristics of ADAs by analyzing neutralizing
activity and epitopes using repository samples from healthy volunteers (HVs) and PwHA
who tested positive for ADAs in clinical studies of emicizumab conducted in Japan.
To assess the neutralizing activity of ADAs in repository samples, we established
a neutralizing activity assay for ADAs against emicizumab. The neutralizing activity
of ADAs was quantified as the inhibited fraction of spiked emicizumab concentration
in a modified one-stage clotting assay in vitro. Animal-derived ADAs neutralized the
spiked emicizumab in a concentration-dependent manner, and differences in sensitivity
were observed in 50, 5, and 0.5 µg/mL emicizumab-spiked conditions ([Fig. 1A]). These results thus indicated that this neutralization assay could detect a broad
range of neutralizing activity by ADAs using a certain amount of spiked emicizumab.
Fig. 1 Measurement of neutralizing activity of animal-derived ADAs against emicizumab and
ADAs from clinical repository samples. (A) The neutralizing activity of the animal-derived monoclonal ADAs against emicizumab
was measured using FVIII-deficient control plasma spiked with 50, 5, and 0.5 μg/mL
emicizumab in one-stage clotting assay. (B) The neutralizing activity of ADAs in plasmas from 6 HVs or 4 PwHA was measured using
FVIII-deficient control plasma spiked with 0.5 μg/mL emicizumab in a one-stage clotting
assay. The neutralizing activity of the representative sample (shown as the sampling
day below) from HVs or PwHA is shown; HV A (day 1), HV B (day 113), HV C (day 112),
HV D (day 113), HV E (day 110), HV F (day 113), PwHA G (day 589), PwHA H (day 505),
PwHA I (day 197), PwHA J (day 1). Among the samples in which plasma emicizumab concentration
is below the limit of quantification, the sample collected at the latest day from
the beginning of emicizumab treatment is shown as the representative. Emicizumab concentration
of HV A, B, C, D, E, F, and PwHA J is below the limit of quantification, and plasma
emicizumab concentration of PwHA G, H, and I is 14.0, 16.9, 0.0853 µg/mL, respectively.
We then analyzed the ADA-positive repository samples collected from 6 HVs (HVs A–F)
and 4 PwHA (PwHA G–J) from 3 clinical studies: 2 HVs from a phase 1 study,[15] 4 HVs from a bioavailability study,[16] and 4 PwHA from a phase 1/2 study[17] ([Table 1]). Informed consent about storing and using the repository samples had been obtained
in the clinical studies, and this research was approved by an internal ethics committee.
In the clinical studies, four HVs showed shorter elimination half-lives of emicizumab
(C: 10.1 days, D: 14.1 days, E: 6.93 days, F: 9.27 days) than those for the other
two HVs (A: 31.1 days, B: 30.0 days), indicating that ADAs of the former four HVs
affected emicizumab pharmacokinetics (PK) ([Table 1]). No PwHA showed decreased emicizumab exposure, indicating that their ADAs did not
affect PK ([Table 1]).
Table 1
Summary of ADA characteristics in ADA-positive subjects from three clinical studies
|
Subjects
|
Clinical study
|
ADA response
|
Detectable neutralizing activity
|
t
1/2 (d)
|
PK affecting
|
Epitope recognition (predominant)
|
|
FIXa arm
|
FX arm
|
Fc
|
|
HV A
|
Bioavailability
|
Treatment-unaffected
|
No
|
31.1
|
No
|
Not analyzed
|
|
HV B
|
Phase 1
|
Treatment-unaffected
|
No
|
30.0
|
No
|
Fab region
|
–
|
–
|
|
HV C
|
Bioavailability
|
Treatment-induced
|
No
|
10.1
|
Yes
|
Fab region
|
Fab region
|
–
|
|
HV D
|
Bioavailability
|
Treatment-induced
|
Yes
|
14.1
|
Yes
|
CDR1 and 3 of light chain
|
CDR1 and 3 of light chain
|
–
|
|
HV E
|
Bioavailability
|
Treatment-induced
|
Yes
|
6.93
|
Yes
|
CDR1 and 3 of light chain
|
CDR1 and 3 of light chain
|
–
|
|
HV F
|
Phase 1
|
Treatment-induced
|
Yes
|
9.27
|
Yes
|
CDR1 and 3 of light chain
|
CDR1 and 3 of light chain
|
–
|
|
PwHA G
|
Phase 1/2
|
Treatment-induced
|
No
|
N/A
|
No
|
Fab region
|
Fab region
|
–
|
|
PwHA H
|
Phase 1/2
|
Treatment-induced
|
No
|
N/A
|
No
|
CDR1 and 3 of light chain
|
CDR1 and 3 of light chain
|
–
|
|
PwHA I
|
Phase 1/2
|
Treatment-induced
|
No
|
N/A
|
No
|
Not analyzed
|
|
PwHA J
|
Phase 1/2
|
Treatment-boosted
|
Yes
|
N/A
|
No
|
Fab region
|
Fab region
|
–
|
Abbreviations: ADAs, antidrug antibodies; CDR, complement-determining region; Fab,
fragment antigen-binding; Fc, fragment crystallizable; FIXa, activated factor IX;
FX, factor X; HV, healthy volunteer; PK, pharmacokinetics; PwHA, patients with hemophilia
A; t
1/2, elimination half-life; N/A, not available.
Note: ADA response was determined based on the harmonized common terminology for immunogenicity.[8]
To measure the neutralizing activity of ADAs, FVIII activity in plasmas from HVs was
neutralized by adding two anti-FVIII monoclonal antibodies to mimic the FVIII-deficient
condition in vitro.[18] In [Fig. 1B], we show the results of neutralizing activity in the most sensitive 0.5 µg/mL emicizumab-spiked
condition. Neutralizing activity was detected in HVs D, E, and F and PwHA J; among
them, PK was affected in three cases (HVs D, E, and F) but not in PwHA J. On the other
hand, neutralizing activity was not detectable in HVs A, B, and C and PwHA G, H, and
I; among them, PK was affected in HV C, but not in the others (HVs A and B and PwHA
G, H, and I). The lack of detectable neutralizing activity in these six cases might
be accounted for by low titers of ADAs, the interference of the emicizumab remaining
in the samples, and the possibility that their ADAs were really not neutralizing.
We further characterized the ADAs by performing epitope analysis with a previously
reported electrochemiluminescence (ECL) immunoassay.[12] Predominant epitopes of each ADA on the emicizumab molecule were identified when
the ECL signal reduction ratio was above the confirmatory assay cut point. For HV
A and PwHA I, epitope analysis was not performed since their ADA titers were too low.
In the other eight cases, various epitopes recognized by ADAs were detected, but none
were in the fragment crystallizable (Fc) region ([Table 1]). Epitopes of ADAs in HVs D, E, and F, whose ADAs were PK-affecting and neutralizing,
were commonly complement-determining region (CDR) 1 and 3 of the common light chain
(cLC). Epitopes of ADAs in PwHA H were also CDR1 and three of cLC, although ADAs of
this case were not PK-affecting nor neutralizing. ADAs of HV B and PwHA G were also
not PK-affecting nor neutralizing, yet they had different epitopes; fragment antigen-binding
(Fab) region of the FIXa arm only in HV B and Fab regions of the FIXa and FX arms
in PwHA G. In HV C, ADA epitopes were Fab regions of the FIXa and FX arms, although
the ADAs of this case were PK-affecting but not neutralizing. On the other hand, ADAs
in PwHA J also recognized the Fab regions of the FIXa and FX arms, yet the ADAs were
not PK-affecting but neutralizing. Overall, neutralizing activity of ADAs was detected
in HVs D, E, and F and PwHA J. ADAs of HVs D, E, and F were PK-affecting and their
epitopes were commonly cLC.
So far, several reports about emicizumab ADAs have been published. Neutralizing ADAs
that affect PK have been reported in several clinical cases.[10]
[12]
[19] There have been reports of cases of ADAs with undetectable neutralizing activity
that do not affect PK,[9] cases of neutralizing ADAs that do not decrease emicizumab exposure,[9] and one case of ADAs that affect PK but lack detectable neutralizing activity.[13] Although our results were derived from a limited number of subjects, we confirmed
all these reported patterns of characteristics in ADAs against emicizumab.
In conclusion, we characterized ADAs against emicizumab and elucidated various patterns
of ADAs using repository samples obtained from six HVs and four PwHA. We hope that
our results will promote the clearer understanding of ADAs and their characteristics.
