Type A Botulinum Toxin–Induced Antibody Production: A Murine Model of Antibody Response

Methods 

Following study review and approval from the Institutional Animal Care and Use Committee (IACUC) of the University of Tennessee, 40 female CD-1 mice in a 27- to 30-g weight range were purchased from Charles River Laboratories (Wilmington, MA). The mice were randomly divided into four groups of 10 animals each. Each mouse was injected with BCB2024 BTA (Botox Cosmetic, Allergan Inc., Irvine, CA) according to the following regimen: group A received 0.12 U every two months; group B received 0.12 U once a month; group C received 0.24 U every two months; and group D received 0.24 U once a month. Injections were administered to the quadriceps muscle of the hind limbs of the animals in an alternating fashion over the span of four months on a 28-day cycle by the principal investigator (DDS) and the project veterinarian (WAH). Blood was collected by mandibular bleed at day zero (four days before administration of the first dose) per the project laboratory technician (WP) and every month thereafter for four months (27–28 days apart). The sera were separated and stored at −20°C until the completion of all scheduled injections. The immune response to BCB2024 BTA was then determined by measuring the level of serum immunoglobulin G (IgG) using a direct horseradish peroxidase enzyme-linked immunosorbent assay (ELISA; ELISA Reagents Kit; Chemicon, Billerica, MA). The test wells of U-bottom MaxiSorp polystyrene microtiter plates (Nunc, Roskilde, Denmark) were coated with 50 μL of BTA antigen (0.25 U/mL in bicarbonate/carbonate coating buffer [100 mM; pH 9.6]) and incubated overnight at 4°C. The plates were washed four times with wash buffer solution (Chemicon), blocked with conjugate diluent (Chemicon), and incubated overnight at 4°C. The supernatant was then removed, and duplicates of 1:100 serum dilutions in conjugate diluent (Chemicon) were added to the appropriate wells and incubated for two hours at room temperature (RT). Four washes with wash buffer solution (Chemicon) were followed by peroxidase-conjugated AffiniPure goat antimouse IgG heary and light (H+L) secondary antibody (Jackson ImmunoResearch Laboratories, West Grove, PA) diluted 1:5000 in conjugate diluent (Chemicon) for two hours at RT and washed five times with wash buffer solution (Chemicon). TMB substrate plus substrate buffer solution (Chemicon) was added to each well and, while covered, it was incubated for 10 minutes at RT. The reaction was stopped with stop solution (Chemicon) added to each well. Color intensity was read at a 450-nm wavelength with a SpectraMax Plus384 microplate reader (Molecular Devices, Sunnyvale, CA). Negative controls were the sera collected before any dose of antigen. Positive controls were a known primary and secondary antibody pair that were found to react with the antigen (1° Ab was rabbit polyclonal antibody to Clostridium botulinum toxin A [Acris Antibodies, Herford, Germany]; 2° Ab was goat polyclonal antirabbit IgG (H + L), conjugated to horseradish peroxidase (HRP) [Novus Biologicals, Littleton, CO]).

Data were analyzed using a mixed model, repeated measures analysis of variance. Several of the most common covariance matrix structures were examined for best fit using Schwartz's Bayesian information criterion.

Results 

Residuals were examined using a histogram and were found to be normally distributed. The frequency of drug administration had significant effects, while dose failed to reach significance (P = .056).

Because the P value for dose was just over the .05 cutoff, an independent samples t test was run on the mean of the dose levels across time and frequency groups, after subtracting out the baseline measure. This yielded a mean ABA of 1.27 in the low-dose group and a mean 1.77 in the high-dose group. However, this also failed to reach statistical significance (t[33.78] = 1.81; P = .08).

Because the effect of time depended upon the level of frequency and vice versa, the main effects of each factor were not examined in favor of a careful analysis of the interaction itself. The photospectrometric absorbance of sera obtained from monthly and bimonthly mouse cohorts were plotted over time, revealing an increase in the level of ABA with each trial. Of note, although both treatment regimens showed an increase in ABA over time, the higher frequency treatment did so to a greater degree (Figure 1).

View Large Image Download to PowerPoint

Figure 1. Plot of mean antibotox antibody (ABA) levels (as measured using photospectrometry) across time for each treatment group. Note that while both treatment regimens demonstrate an increase in ABA over time, the higher frequency treatment does so to a greater degree. (*Time in months; ABA measured by photospectrometric absorbance.)

For both frequency groups, the mean photospectrometric absorbance at each injection point was compared to their respective baseline measurements using paired t tests whose P values were corrected for multiple testing using the sequential Bonferroni method. Each time differed significantly from baseline and the mean differences tended to increase across time (Table 1).

Table 1. Repeated measures analysis of variance results

	Cohort	Serum acquisition	t	df	Significance (two-tailed)	Mean difference	95% CI of the difference
	Bimonthly	Postinjection 1	4.136	19	.001	0.92985	.4593–1.4004
		Postinjection 2	4.060	19	.001	0.94498	.4579–1.4321
		Postinjection 3	7.388	19	.000	1.48235	1.0624–1.9023
		Postinjection 4	6.936	19	.000	1.43215	1.0000–1.8643
	Monthly	Postinjection 1	5.943	19	.000	1.39123	.9013–1.8812
		Postinjection 2	9.314	19	.000	1.93338	1.4989–2.3678
		Postinjection 3	10.586	19	.000	2.01078	1.6132–2.4084
		Postinjection 4	11.563	19	.000	2.03660	1.6679–2.4053

CI, confidence interval.

To obtain another perspective on the time by frequency interaction, the frequency groups were compared at each time point. Independent samples t tests showed the higher frequency treatment group had significantly more ABA than did the lower frequency group at all except the first measurement time (Table 2).

Table 2. Statistically significant differences are noted at each postinjection interval except after the first injection

	Point of comparison	t	df	P	Mean difference	Standard error difference	95% CI of the difference
	Postinjection 1	1.645	37.928	.108	.54128	.32896	−1.20726-0.12471
	Postinjection 2	3.359	37.659	.002	1.06830	.31801	−1.71226–0.42434
	Postinjection 3	2.157	37.948	.037	.60833	.28202	−1.17928- −0.03737
	Postinjection 4	2.476	37.255	.018	.68435	.27636	−1.24418- −0.12452

Each time point differed significantly from baseline, with a tendency for the mean differences to increase across time.

In summary, nascent antibody production did occur in response to BCB2024 BTA administration and was observed in all subgroups. While the impact of dosage on antibody response neared but failed to reach statistical significance (P = .056), the effect of frequency of administration was statistically significant both within subgroups and in comparison of monthly versus bimonthly cohorts (P = .00 and P = .018, respectively).

Discussion 

When reviewing the findings of our investigation and considering the potential antigenicity of BTA, it is important to recall some basic immunologic principles as they pertain to the drug's molecular structure in its presently marketed form. Although both size and epitopic complexity are important factors in antibody formation, carrier proteins used for various reasons (including product stabilization or delivery) may also inadvertently serve as haptens, or substances that increase the antigenicity of any given substance (eg, albumin, as used in production of BCB2024 BTA).13 In addition, it is often true that the greater the protein load of a substance, the greater its immunologic footprint. Conversely, decreasing the amount of protein associated with a given substance can often decrease its antigenic profile and therefore its ability to elicit an immune response.13

Before 1998, these principles were well-recognized by Allergan investigators, who understood that the development of ABA and an associated nonresponder state in up to 20% of those who received 79-11BTA11 was clinically unacceptable, especially for a drug increasingly touted for aesthetic purposes. In response to increasing awareness and concern regarding immunologically based treatment failure, Allergan undertook an antigenic refinement of the originally marketed toxin. As a result, 79-11 BTA emerged in 1998 as a less immunologically-conspicuous version of botulinum toxin that was referred to as BCB2024 BTA.

Despite undergoing immunologic contouring, botulinum toxin remains a relatively large polypeptide composed of a 100-kDa heavy chain (responsible for internalization within acetylcholinergic nerve termini) and an associated 50-kDa light chain (responsible for the drug's neuroparalytic capability).14, 15 To put this in perspective, recall that a single molecule of human albumin is 69 kDa, while the neurotoxin component of BCB2024 BTA alone is 150 kDa (Figure 2). It is readily conceivable, then, that even after modification, the protein toxin component alone of BTA represents a potentially antigenic substance. Indeed, as has already been shown, the literature demonstrates the same with use of both 79-11 BTA and its more refined offspring, BCB2024 BTA.4, 5, 7, 8, 9, 10, 11, 12

View Large Image Download to PowerPoint

Figure 2. Comparative model of BCB2024 botulinum toxin type A antigenicity in terms of molecular size (expressed in Da).

(Adapted with permission from Tizard13 Immunology, 4E. ©1995 Brooks/Cole, a part of Cengage Learning, Inc.

With this in mind, we developed a dosing schedule based on an understanding originally purported by Carruthers and Carruthers16 that a lag time of one to two weeks generally occurs before BTA produces its maximal effect. We also selected intervals and dosages compatible with previous animal studies that had demonstrated the safety and feasibility of BTA administration in murine subjects.17, 18, 19 In addition, in an effort to simulate a clinically relevant scenario, we considered previous recommendations advising retreatment no sooner than three to four weeks after to the initial drug administration,16 as well as the recommendations and personal experience of those expert in the use of BCB2024 BTA.20 Currently, however, no standardized scale of conversion exists to relate murine doses of BTA to those used in human treatment regimens.

To further illustrate this concept, we are not able to state definitively that 0.12 U of BCB2024 used in our regimen is equivalent to providing a patient with a single series of injections totaling 12 or 120 U. Therefore, the issue of dosing of BCB2024 in mice as it pertains to human dosing represents a limitation in our study.

While neither we nor anyone else (to our knowledge) can, at the time of this writing, define an exact murine-to-human conversion scale in the dosing of BCB2024 BTA, we do know from simple observation of the animals in our study that none experienced clinically apparent paralysis of the injected hind limb. Two conclusions can possibly be drawn from this observation. First, the dose of BCB2024 given to the animals was not great enough to produce paralysis of the quadriceps femoris. Second, the dose of the drug could have been high enough to produce paralysis, but given the neuroprotective effects of nascent ABA formation, the mice were protected from such a result. Because no gross motor compromise was ever noted in any of the study subjects, even with the first dose of BCB2024, we believe the former to be the most plausible explanation. Likewise, it is reasonable to infer that the animals in our investigation (even in the groups receiving the higher 0.24 U dose) received an amount of BCB2024 more similar to the low-dose regimen that might be delivered to a human for aesthetic purposes rather than the higher (and traditionally more immunogenic) doses given in treating torticollis, for instance.

To further increase the relevance of the study, we also reviewed the work of other investigators who determined that IgG was the antibody subtype that was primarily responsible for the clinical neutralization of BTA.2, 21 Accordingly, in the design of our ELISA, we selected reagents that were specific for the detection of an IgG antibody in the event that BCB2024 elicited an antibody response. While we were able to document the occurrence of ABA production by means of an IgG-specific ELISA, we did not differentiate between neutralizing and nonneutralizing antibodies. In other words, although we have clearly determined that mice receiving a modified form of BTA do indeed produce antibodies in response to the drug, based on the parameters of our current study, the clinical significance of these antibodies can only be surmised. This represents an important limitation in our study. It does not, however, discount the significance of our data, which contradict the BCB2024 product monograph released at the introduction of the drug in 1998. According to this manufacturer-based study, no antibody production of any form was evident in the group of rabbits receiving BCB2024 after six months of investigation.6

Of the differences in our findings, perhaps foremost is the nascent antibody formation that was evident in the animals of our trial. We also used a mouse rather than a rabbit model. Despite the murine basis of our study, we know of no interspecies immunologic difference to account entirely for this variation. Finally, not only did ABA antibody formation occur, but its occurrence developed over a smaller interval of time with statistically significant increases in ABA production linked to frequency rather than dose. This finding supports previous recommendations to avoid “boosters” or supplemental treatments given within one month of a treatment session22, 23 and correlates with the pre-BCB2024 retrospective findings of Zuber et al.24 It also seems to correspond with more recent reports of immunologic resistance to BCB2024 BTA, even with use of low-dose treatment regimens incorporating only 30 to 60 U of toxin per cycle.9, 25

Conclusions 

In summary, our findings lend credulity to commonly held principles regarding nascent antibody production as a function of the amount of antigen and the frequency with which it is introduced into an in vivo immune system. Contrary to manufacturer data, we conclude that BCB2024 BTA does provoke an in vivo antibody response. Regarding clinical and cosmetic applications of the drug, the frequency of administration may represent a more important variable than dosage in terms of eliciting an antibody response.