The time-course response after a single injection of botulinum toxin (BoNT) for post-stroke spasticity is debated. We addressed this issue by conducting a systematic review and a pharmacodynamic model-based meta-analysis.

We searched Medline, PeDro and Google Scholar databases up to March 2020, selecting randomized controlled trials of post-stroke and traumatic brain injury patients with arm or leg muscle hypertonia, comparing BoNT to placebo, or different BoNT preparations. The main outcome was change in Modified Ashworth Scale (MAS) score. A non-linear mixed effect model was used to estimate maximal toxin and placebo effects (Emax and EPlacebo), the effect disappearance half-life (T1/2off) of BoNT and the doses achieving 50 and 80% of Emax (D50 and D80). The equivalence ratios between different BoNT preparations were calculated from D50 values. Adverse events were recorded.

Altogether, 2,236 unique records were screened by 2 independent reviewers: 35 eligible trials including 3011 patients (95% post-stroke) were identified. For all BoNT preparations, the BoNT Emax of -1.11 (95% credible interval -1.31; -0.29) was reached at 5 weeks; the maximal placebo effect was -0.30 (-0.37; -0.22). Both D50 and D80 differed significantly by muscle volume. At D50, the equivalence ratio was significantly higher for abobotulinumtoxinA (3.35) than onabotulinumtoxinA and lower for letibotulinumtoxinA (0.41). T1/2off was longer for abobotulinumtoxinA than for onabotulinumtoxinA and the other preparations (13.1 weeks [95% credible interval 7.7; 19.3] vs 8.6 weeks [7.1; 10.1]). Adverse events were minor, with a weak, but significant, dose–response relation for muscle weakness.

This first pharmacodynamic model-based meta-analysis of individuals with stroke revealed that for all BoNT-A preparations, BoNT-A injections to treat spasticity have maximal effect at 5 weeks. The T1/2off was longer for abobotulinumtoxinA than other preparations. Differences between certain BoNT unit scales were also confirmed.