Time course of the placebo response and associated factors in clinical trials of biologic agents for ankylosing spondylitis

To characterize the time course and influencing factors of placebo response in randomized trials of biologic and targeted synthetic disease-modifying antirheumatic drugs (tsDMARDs) for ankylosing spondylitis (AS). We systematically searched major electronic databases (up to June 14, 2024) for randomized double-blind placebo-controlled trials of biologics or Janus kinase inhibitors in AS. Model-based meta-analysis with nonlinear mixed-effects modeling was conducted to pool placebo-arm ASAS20 (≥20% improvement by Assessment of SpondyloArthritis International Society criteria) and ASAS40 (≥40% improvement) response trajectories, using an exponential rise-to-maximum model with covariate screening and model validation. Thirty-eight trial arms from 36 trials (3046 placebo-treated participants) were included. The estimated maximum response (Pmax) was 31.1% for ASAS20 (90% of Pmax reached at 6.5 weeks) and 18.6% for ASAS40 (90% of Pmax reached at 17.0 weeks). ASAS20 Pmax increased by 2.7% per publication year. Smaller trials, no background therapy, lower male proportion, and lower baseline HLA-B27 positivity were associated with higher placebo response. The AS placebo response, particularly for ASAS20, shows a significant upward secular trend. This model provides a quantitative benchmark for trial design, anomalous placebo responses identification, and interpreting efficacy in trials lacking a concurrent placebo arm. Ankylosing spondylitis (AS) is a long-term inflammatory disease that mainly affects the joints of the spine. To find out whether a new medicine works, researchers compare it with a placebo, a dummy treatment with no active drug. The difficulty is that people given a placebo often improve anyway, and by very different amounts from trial to trial. When the placebo response is unusually large, an effective medicine can look as though it does not work. We collected data from 38 published AS trials of biologic medicines or Janus kinase inhibitors, covering 3046 people who received a placebo. Using a mathematical model, we described how the placebo response changes with time and what makes it larger or smaller. Two findings stood out. First, the placebo response grows slowly and then levels off: it takes about 6.5 weeks to reach its maximum when improvement reaches 20% and about 17 weeks when it reaches a stricter 40%, so a trial that stops early may underestimate the placebo group’s improvement. Second, the placebo response has grown by about 2.7% a year since 2002. It also tends to be larger in smaller trials, in trials where background painkillers were not allowed, in trials with fewer men and in trials where fewer patients carried the human leukocyte antigen B27 (HLA-B27) gene. Our model is a reference point: it helps researchers judge whether a new trial’s placebo response is unusually large or small, and interpret trials that had no placebo group. It may help design smaller, faster trials. Placebo responses vary substantially in ankylosing spondylitis (AS) clinical trials and interfere with true-drug-efficacy evaluation.This model-based meta-analysis (MBMA) included 38 placebo arms (3046 participants) to quantify ASAS20 (≥20% improvement by Assessment of SpondyloArthritis International Society criteria) and ASAS40 (≥40% improvement) placebo-response time-courses.The exponential rise-to-maximum model estimated Pmax at 31.1% for ASAS20 and 18.6% for ASAS40; 90% of Pmax was attained at 6.5 and 17.0 weeks, respectively.ASAS20 placebo response rose 2.7% annually; this secular trend needs consideration for cross-trial comparisons.Higher placebo responses correlated with small sample size, no background therapy, fewer male participants and lower baseline human leukocyte antigen B27 (HLA-B27) positivity.This model detects anomalous placebo outcomes and acts as a historical external control to support AS-trial design and efficacy interpretation. Placebo responses vary substantially in ankylosing spondylitis (AS) clinical trials and interfere with true-drug-efficacy evaluation. This model-based meta-analysis (MBMA) included 38 placebo arms (3046 participants) to quantify ASAS20 (≥20% improvement by Assessment of SpondyloArthritis International Society criteria) and ASAS40 (≥40% improvement) placebo-response time-courses. The exponential rise-to-maximum model estimated Pmax at 31.1% for ASAS20 and 18.6% for ASAS40; 90% of Pmax was attained at 6.5 and 17.0 weeks, respectively. ASAS20 placebo response rose 2.7% annually; this secular trend needs consideration for cross-trial comparisons. Higher placebo responses correlated with small sample size, no background therapy, fewer male participants and lower baseline human leukocyte antigen B27 (HLA-B27) positivity. This model detects anomalous placebo outcomes and acts as a historical external control to support AS-trial design and efficacy interpretation.