Post Hoc and Precarious: A Cautionary Note on Terlipressin in Early‐Stage <scp>ACLF</scp>

We read with interest the article by Rockey et al. [1] on terlipressin in ‘early-stage’ acute-on-chronic liver failure (ACLF) patients with hepatorenal syndrome (HRS). While the analysis is thought-provoking, we have several concerns regarding its design and interpretation. First, the post hoc exclusion of patients with ACLF grade 3 or serum creatinine ≥ 5 mg/dL is a fundamental design flaw [1]. This unplanned subgroup selection, performed after randomization, introduces substantial selection bias and violates the initial randomization scheme. By omitting the sickest patients (those at highest risk of non-response and adverse outcomes), the authors likely inflated the apparent efficacy of terlipressin in the remaining cohort. Such retrospective stratification can compromise the validity of the findings, as a more favourable ‘benefit-to-risk’ subset was effectively cherry-picked, limiting the generalizability of the results. Second, the interpretation of clinical benefit in this study is misplaced. The authors emphasise improvements in surrogate markers—notably a greater drop in creatinine and MELD score with terlipressin, yet no actual outcome advantage was demonstrated [1]. Liver transplant rates remained equal and 90-day survival was similar between groups. Highlighting surrogate endpoints (HRS reversal, creatinine reduction) in the absence of any survival or transplant benefit risks a ‘surrogate endpoint fallacy.’ As others have noted, HRS reversal with terlipressin does not confer a survival advantage unless liver function is restored via transplant [2]. Indeed, a recent meta-analysis confirmed no improvement in survival with terlipressin therapy at 1–3 months despite higher HRS reversal rates [3]. Focusing on biochemical surrogates without translating to patient-centered outcomes is misleading. Finally, the safety conclusions drawn are problematic. The study reports a similar incidence of respiratory failure in terlipressin vs. placebo (11% vs. 7%, p = 0.36) after excluding high-risk patients [1]. However, this underpowered subgroup analysis cannot establish true safety equivalence—the wide confidence intervals and small sample mean that a clinically important risk increase could be missed. In fact, the FDA has issued a boxed warning for terlipressin about serious or fatal respiratory failure, especially in patients with volume overload or ACLF-3 [4]. The label explicitly recommends avoiding terlipressin in ACLF grade 3 due to the high risk of respiratory failure [4]. These broader safety signals (e.g., in the overall CONFIRM trial population) are downplayed by the authors. Ignoring the FDA's safety warnings and relying on an underpowered subset analysis to claim ‘comparable’ respiratory risk is inappropriate and potentially unsafe. In summary, while Rockey et al. contribute valuable data on a select HRS subgroup, the post hoc patient exclusions, overemphasis on surrogates without outcome gains, and optimistic safety assurances warrant caution. An evidence-based, balanced interpretation requires acknowledging these limitations and the need for prospective validation in unselected HRS patients. The author has nothing to report. The author declares no conflicts of interest. This article is linked to https://doi.org/10.1111/liv.70399. The author has nothing to report.