The investigators performed a post hoc analysis of the VANISH randomized trial to assess whether baseline renal function, measured by proenkephalin A (PENK), modified hydrocortisone effects in patients with septic shock. VANISH was a double-blind randomized study comparing vasopressin versus norepinephrine and hydrocortisone versus placebo across multiple UK ICUs. The analysis cohort comprised 173 patients whose plasma PENK concentrations were available at baseline. The study team dichotomized patients using a PENK cutoff of 89 pmol/L to define high and low groups; that threshold was applied as the primary stratifier in all subgroup analyses reported.
The rationale for focusing on PENK was clinical and mechanistic. PENK acts as a circulating marker that correlates with renal function, and the kidney plays central roles in sepsis pathophysiology and steroid metabolism. Investigators hypothesized that baseline renal impairment, reflected by higher PENK, might alter hydrocortisone pharmacodynamics or net clinical benefit.
Data indicated that a baseline PENK concentration ≥ 89 pmol/L associated with increased 28-day mortality. The authors reported an odds ratio of 3.24 (95% CI, 1.62–6.74) for mortality in the high-PENK group versus the low-PENK group. That finding positions PENK as a prognostic biomarker in this septic shock sample.
The analysis also tested whether PENK modified the effect of the trial's vasopressor randomization (vasopressin vs norepinephrine). The study team observed no significant interaction between PENK group and vasopressor assignment on clinical outcomes. In other words, baseline PENK did not appear to change the relationship between vasopressor strategy and measured endpoints in this dataset.
Investigators focused a prespecified subgroup on patients who reached the trial's maximum vasopressor threshold (n = 111). Within that subgroup they tested for an interaction between PENK group and hydrocortisone versus placebo for kidney failure–free days. The analysis produced a statistically significant interaction (p = 0.01). The reported kidney failure–free days by group were as follows: high PENK, placebo median 26 (IQR 13–28) versus hydrocortisone median 11 (IQR 0–26); low PENK, placebo median 28 (IQR 15–28) versus hydrocortisone median 28 (IQR 28–28). The study team interpreted these differences to indicate that patients with elevated baseline PENK who received hydrocortisone accrued the fewest kidney failure–free days.
These observations imply heterogeneity of treatment effect for hydrocortisone when stratified by a renal biomarker, at least among patients who progressed to the trial’s maximum vasopressor threshold. The source did not provide additional mechanistic data such as hydrocortisone pharmacokinetics by PENK level, nor did it report granular timelines for kidney injury onset relative to steroid dosing beyond the kidney failure–free day endpoint.
Although the interaction for kidney failure–free days reached statistical significance in the maximum vasopressor subgroup, the interaction for 28-day mortality did not meet conventional significance (p = 0.14), according to the authors. Nevertheless, raw mortality proportions varied across PENK and steroid groups: the high-PENK hydrocortisone group had the highest reported mortality (50%), compared with high-PENK placebo (32%) and low-PENK groups (hydrocortisone 14%, placebo 24%). Those absolute differences raise clinical concern but—again—did not achieve statistical interaction significance for the mortality endpoint in this dataset.
Readers should treat the mortality pattern as hypothesis-generating. The analysis was exploratory and post hoc, so the observed mortality distribution may reflect chance, confounding, or sample-size limitations rather than a causal harm signal attributable to hydrocortisone among high-PENK patients.
The authors described this work as exploratory and recommended validation in independent prospective cohorts. Several important methodological constraints limit interpretation. The analysis was post hoc and therefore subject to selection and multiplicity biases inherent to secondary subgroup testing. The source did not report detailed hydrocortisone dosing regimens, timing of baseline PENK sampling relative to randomization beyond the designation of baseline, or pharmacokinetic measures that could clarify whether reduced renal clearance influenced steroid exposure.
The findings hinge on a single PENK threshold (89 pmol/L) applied within this cohort; the external validity of that cutoff and its operating characteristics across broader septic shock populations were not provided. The analysis also concentrated its statistically significant interaction on a subset (those reaching maximum vasopressor threshold, n = 111), which reduces sample size and increases the possibility of random variation. Safety outcomes beyond mortality and kidney failure–free days were not extensively reported in the abstract, and the source did not present stratified adverse-event rates tied to steroid assignment and PENK strata.
Prospective validation is necessary. A preplanned randomized trial or prospectively stratified analysis using baseline proenkephalin A to guide hydrocortisone use would clarify whether the interaction observed here replicates and whether it represents a clinically actionable signal. Additional studies should capture steroid dosing details, timing of biomarker sampling, pharmacokinetic data, and comprehensive renal and nonrenal safety endpoints to resolve remaining uncertainties documented by the authors.