Dynamic sepsis endotypes: instability or expected signal of biological progression? Author's reply

Ford and colleagues [1], in response to our recent article in Intensive Care Medicine [2], argue that the frequent transitions between immune profiles observed in our cohort primarily reflect the expected biological evolution of sepsis during an ICU admission.While we recognize that there may be important biological signals in our data, we consider this explanation insufficient.Our findings instead challenge the prevailing paradigm that sepsis patients exhibit well-defined, static or longitudinal, immunological profiles that can be reliably captured by assigning patients to discrete subgroups.At a mechanistic level, it is likely that a substantial proportion of the observed cross-endotype transitions reflects classification error rather than true biological change.To estimate the relative contribution of such noise, one can consider a hypothetical upper bound for biologically plausible variation.Even under a generous assumption that each patient undergoes up to three genuine profile changes during their ICU stay, the resulting expected transition rates would not surpass 15-25%.In contrast, we observed 8-hourly transition rates of 41%, 39%, and 22% across the three different starting profiles and these rates persisted throughout the ICU course.Moreover, the rapidity of transitions-occurring within 8-h intervals-exceeds what can reasonably be attributed to biological evolution alone.Together, the unexpectedly high transition rates, their temporal persistence, and the implied short profile "half-lives" strongly suggest a major role for classification error.