Early multimodal vasopressor strategy in septic shock: results of the TRICYCLE randomized controlled trial

Abstract

Background

Given multifactorial mechanisms driving vasodilation in septic shock, there is a strong physiological rationale for early initiation of multiple vasopressors (angiotensin II, vasopressin and norepinephrine) with complementary mechanisms of action. Such an approach could more effectively counteract the complex pathophysiological processes characteristic of septic shock. Here we tested the physiological and clinical effects of an early multimodal vasopressor strategy in patients with septic shock.

Methods

We randomly assigned patients with septic shock receiving ≥ 0.15 µg/kg/min of norepinephrine base to either intervention (early multimodal vasopressor administration: angiotensin II, vasopressin, and norepinephrine) group or control (classic stepwise vasopressor administration: norepinephrine dose increase followed by vasopressin and then by angiotensin II or other vasopressors) group. The primary outcome was change in plasma renin concentrations. This phase II trial was not powered to detect differences in mortality.

Results

A total of 79 patients were included (intervention group: n = 38, control group: n = 41). We observed a faster decline in plasma renin concentrations (2.61% per hour vs. 1.67% per hour; p = 0.015) and lactate concentrations (1.53% per hour vs. 1.01% per hour; p = 0.006) during the first 72 h in the intervention group compared to the control group. At 72 h, Sequential Organ Failure Assessment (SOFA) score was reduced by 3 points in the intervention group vs. 0 in the control group (p = 0.018). In the additional non-prespecified analyses, norepinephrine equivalent (NEE) dose decreased at a similar rate in the intervention and control groups over 72 h (− 5.41% per hour vs. −5.19% per hour; p = 0.47), whereas over the first 24 h, NEE dose decreased significantly faster in the intervention group (− 6.87% per hour vs. −3.03% per hour; p < 0.0004). The incidence of tachyarrhythmias was significantly lower in the intervention group (15.6% vs. 68.6%; p < 0.01). Intensive care unit mortality (26.3% vs. 41.5%; RR 0.635; 95% CI 0.33 to 1.21; p = 0.235) and 28-day mortality (36.8% vs. 46.3%; RR 0.795; 95% CI 0.47 to 1.35; p = 0.494) were similar.

Conclusion

Early multimodal vasopressor therapy with Ang II, vasopressin and norepinephrine in patients with septic shock demonstrated promising physiological effects. These findings provide a rationale for a larger trial to assess whether early multimodal vasopressor therapy improves patient-centered outcomes.

Trial registration

ClinicalTrials.gov ID: NCT06155812. URL: https://clinicaltrials.gov/study/NCT06155812?term=NCT06155812&rank=1. Trial registration date: 4 December 2023.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s13054-026-06323-z.

Background

Vasopressors are considered lifesaving in patients with septic shock when hypotension is unresponsive to fluid resuscitation. Norepinephrine is considered the vasopressor of first choice and is superior to dopamine [1, 2]. Although clinical trials comparing vasopressin, angiotensin II (Ang II) or other non-adrenergic vasopressors versus norepinephrine have failed to demonstrate a survival benefit [2, 3], retrospective data suggest these agents may improve patient outcomes [4].

Vasodilation, vascular hyporeactivity and hypotension can persist with preserved cardiac output despite exogenous catecholamine infusion, which can be attributed to interpatient differences in adrenergic receptor function [5, 6] and dysregulation of multiple neurohormonal pathways [7–10]. Given that mortality increases substantially with escalating doses of norepinephrine [11, 12], initiation of additional vasopressors is commonly considered to shorten hypoperfusion and minimize toxicity associated with excessive catecholamine exposure. A balanced approach with the introduction of multiple mechanistically distinct vasopressors could theoretically better counteract the complex pathophysiology of septic shock [13–15]. The precise timing and selection of additional vasopressors remains unclear [1, 16]. Although observational studies, meta-analyses, and post-hoc analyses suggest potential benefits of earlier introduction of non-adrenergic vasopressors, prospective evidence remains limited [1, 17, 18].

Over the past decade, several studies showed that elevated renin concentrations on admission and increasing concentrations during the intensive-care-unit (ICU) stay are associated with mortality and outperform lactate in outcome prediction [19–24]. Two recent studies showed that renin retains its prognostic value in patients who are treated with Ang II [25, 26].

These findings have prompted the initiation of the “Early multimodal vasopressor strategy in septic shock (TRICYCLE)” trial to determine whether early initiation of multimodal vasopressor therapy (defined as simultaneous use of Ang II, vasopressin and norepinephrine) accelerates the decline in plasma renin concentrations and improves physiological and clinical outcomes compared with classic stepwise use of vasopressors in adult patients with septic shock.

Methods

Study design and setting

This was an investigator-initiated and -designed (ZK, MG, GL and AM) randomized controlled trial in three separate ICUs in two countries (University Medical Centre Maribor and University Hospital Centre Zagreb). The trial was single blinded and enrolled adult septic shock patients to evaluate the physiological and clinical effects of an early multimodal vasopressor strategy compared to a classic stepwise vasopressor approach. Sepsis and septic shock were defined according to the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3) [27]. The protocol was approved by ethics committees in all participating institutions (KME RS 0120–212/2023-2711-13 and KME Rebro 02/013 AG). All participant screening and enrollment procedures were performed by the investigators (Fig. 1). Written informed consent was obtained from all patients. If the patient was unable to provide informed consent due to disease severity, written consent was obtained from a legal guardian or the closest relative before study enrolment. Detailed rationale for the trial and trial protocol was published separately [28]. Collection and analysis of the trial data and preparation of the manuscript were all performed by the investigators. No protocol deviations were documented during the conduct of the trial. Differences between the protocol-specified and reported outcomes and analyses are summarized in Supplementary Appendix Table S2.

Fig. 1.

CONSORT flow diagram of patient screening, enrollment, randomization, and analysis

Study population

We included patients who fulfilled the following inclusion criteria: adult (≥ 18 years), with sepsis and persisting hypotension requiring vasopressors to maintain mean arterial pressure (MAP) ≥ 65 mmHg and serum lactate level > 2 mmol/L despite adequate volume resuscitation (20–30 ml/kg in 3 h), with vasopressor requirement ≥ 0.15 µg/kg/min of norepinephrine base equivalent, with central venous access, arterial line and urinary catheter inserted and expected to remain present for at least the initial 72 h of treatment, and with cardiac index (CI) > 2.3 L/min/m² (measured by bedside echocardiography, pulse contour analysis cardiac output (PiCCO) or Swan-Ganz catheter) [28]. Exclusion criteria were as follows: death expected within 24 h, suspected or confirmed pregnancy, surgery expected for the source of infection or inter-hospital transfer expected during the first 72 h of hospitalization (because both situations could have compromised protocol adherence), liver failure with a Model for End-Stage Liver Disease (MELD) score ≥ 30, acute mesenteric ischemia or history of mesenteric ischemia, Raynaud’s phenomenon, systemic sclerosis or vasospastic disease, active bleeding and an anticipated need (within the first 48 h of study initiation) for transfusion of > 4 units of packed red blood cells, known allergy to mannitol (an ingredient in Ang II 2,5 mg/ml vials), or veno-arterial ECMO.

Study intervention

Patients were randomized to the intervention (Tricycle) group or the control group in a 1:1 ratio. Patients in the intervention group received multimodal vasopressor therapy with Ang II, vasopressin and norepinephrine, while patients in the control group received standard stepwise vasopressor therapy starting with norepinephrine, adding vasopressin at higher doses of norepinephrine and adding additional vasopressors at high doses of norepinephrine and vasopressin. The intervention group vasopressor titration regimen is shown in Fig. 2. Detailed treatment protocols for both groups are provided in the Supplementary Appendix and the published trial protocol [28]. Vasopressor doses were converted to NEE using the conversion factors proposed by Kotani et al. [29] (Table S1 Supplementary Appendix). The protocol allowed the use of additional inotropes and vasopressors in both the control and intervention groups (dobutamine, levosimendan, epinephrine, methylene blue, or dopamine). Use of Ang II was allowed in the control group in patients with refractory hypotension (persistent hypotension despite exceeding a norepinephrine base dose of 0.5 µg/kg/min and administration of vasopressin at a dose of 0.03 IU/min). Vasopressor therapy for both the intervention and control groups was protocolized for the first 72 h. After 72 h, vasopressor therapy was prescribed as per the treating physician. All patients received thromboprophylaxis. Following randomization, additional intravenous fluids could be administered in both groups at the discretion of the treating team, based on clinical and ultrasound assessment of fluid status and responsiveness. All other aspects of treatment were as per standard guidelines and recommendations.

Fig. 2.

Vasopressor titration regimen in the Tricycle group. Reproduced from [28]. Abbreviations: NE norepinephrine base, AVP arginine vasopressin, Ang II angiotensin II

Assessments and end points

Extended laboratory examination was performed at baseline and at 24, 48 and 72 h after enrollment. Plasma renin concentrations were determined at baseline and at 12, 24, 36, 48 and 72 h after enrollment. Plasma lactate concentrations were determined at baseline and at 12-h intervals until 72 h after enrollment. SOFA score was determined daily. Plasma renin concentrations were measured as direct renin using a commercially available chemiluminescent immunoassay on the LIAISON XL automated analyzer (LIAISON Direct Renin assay, DiaSorin S.p.A., Saluggia, Italy) according to the manufacturer’s instructions. Results were expressed as mU/L. The assay has a reported reference range of 4–50 mU/L and an upper quantification limit of 500 mU/L.

The primary endpoint was the change in plasma renin concentrations within 72 h after inclusion. Prespecified secondary endpoints included change in lactate concentrations within 72 h after inclusion and change in SOFA score between day 1 and day 3 (ΔSOFA) [30]. Prespecified exploratory endpoints were RRT requirement during ICU stay, survival to ICU discharge (reported as ICU mortality) and 28-day mortality. Longitudinal NEE trajectories over 0–72 h and 0–24 h and change in eGFR over 0–72 h were additional non-prespecified analyses. A comparison of protocol-specified and reported outcomes/analyses and rationale for modifications are presented in Supplementary Appendix Table S2.

Adverse events recorded during the study period included thrombotic events, tachyarrhythmias, thrombocytopenia, hyponatremia and cardiac, mesenteric and digital ischemia. Adverse events were identified from clinical assessment, laboratory findings and diagnostic investigations. During data collection, tachyarrhythmias were recorded when atrial fibrillation or atrial flutter, paroxysmal supraventricular tachycardia, or ventricular tachyarrhythmias required pharmacological treatment or electrical cardioversion. Events were identified during continuous cardiac monitoring and documented by the treating clinical team.

Data analysis

The planned sample size was 80 patients (40 per group), based on a priori Wilcoxon-Mann-Whitney comparison of renin concentrations to provide at least 80% statistical power as described in the published study protocol [28]. This approach was based on available published renin data, which did not provide sufficient information to estimate the parameters required for a sample-size calculation based on a longitudinal mixed-effects model. The sample size calculation therefore did not formally assess power for the mixed-effects model used in the present study.

Data were analyzed using IBM SPSS Statistics 25.0 (IBM Inc., Armonk, New York, USA) and R 4.2.3 statistical environment (R Core Team 2020, Vienna, Austria). We used standard statistical approaches to test for significant differences between the intervention and control groups. Data were summarized using descriptive statistics for continuous and categorical variables [28]. The Shapiro-Wilk test was used to assess the normality of data distribution, Chi-square test and Fisher’s exact test were used to analyze categorical variables and Student’s t-test and Mann–Whitney U-test were used to analyze continuous variables between groups. 95% confidence intervals for binary outcomes were calculated using the Wilson method. For the primary analysis of ΔSOFA, patients who died before day 3 were assigned a day-3 SOFA score of 24. A sensitivity analysis was performed excluding patients who died during first 72 h to assess the influence of mortality on between-group comparisons. A post hoc exploratory analysis of tachyarrhythmias was performed using logistic regression adjusted for baseline atrial fibrillation and beta-blocker use. Repeated measurements of renin, lactate, and NEE dose were analyzed using linear mixed-effects models (LMM) in R version 4.2.3 using the packages readr 2.1.4, dplyr 1.1.4, tidyr 1.3.1, ggplot2 3.5.0, lme4 1.1–34, emmeans 1.8.7, mice 3.19.0 and lme4cens 0.1.14. Outcomes were log-transformed due to right-skewed distributions. For NEE, a small constant was added before the log transformation. Time was treated as a continuous variable in hours and centered at its mean to improve model stability. The models included fixed effects for time, intervention group, the group-by-time interaction, and relevant covariates. Renin and lactate models were adjusted for baseline eGFR, age, and sex, based on their potential influence on study outcomes. NEE models were also adjusted for hydrocortisone and ACE inhibitor or ARB use, based on their potential influence on vasopressor requirements. Random intercepts for subjects were included to account for the correlation within subjects. As a sensitivity analysis, the models were repeated with subject-specific random slopes for time in addition to random intercepts. The group-by-time interaction was the main parameter to test whether the rate of change in renin, lactate and NEE differed over time between the control and intervention groups. Estimated marginal trends were used to obtain group-specific slopes. Pairwise comparisons between groups were performed using estimated marginal trends and p values were adjusted for multiple comparisons using the Holm method. Results were subsequently back-transformed from the log scale and reported as percentage change per hour, percentage change over 24 h, and corresponding 95% confidence intervals. NEE was analyzed over the full 72-hour period and over the first 24-hour period. Additional analyses evaluated associations between individual renin trajectories and ICU mortality, 28-day mortality, RRT requirement during ICU stay and ΔSOFA. Binary outcomes were assessed using logistic regression with ORs expressed per 1%-point/h slower renin decline and ΔSOFA using Spearman correlation. Analyses were also performed separately by treatment group.

Because renin measurements at the assay upper quantification limit of 500 mU/L represent right-censored and not exact values, an additional sensitivity analysis was performed using a right-censored mixed effects model, with measurements at the upper quantification limit treated as right-censored rather than exact values. The fixed-effects and random intercept structure was otherwise consistent with the primary renin analysis.

Fluid balance was omitted from the primary analyses because of missing data, which would have reduced number of complete case data to only 65 subjects. As a sensitivity analysis, missing fluid-balance values were imputed using multivariate imputation by chained equations (MICE) with predictive mean matching to generate 20 imputed datasets. The mixed-effects models were then repeated with fluid balance included as an additional covariate, and parameter estimates were pooled according to Rubin’s rules.

The models were fitted using maximum likelihood estimation with the bobyqa optimizer and higher iteration limits to ensure convergence. Statistical significance was assessed at an alpha level of 0.05.

Results

Patients and baseline characteristics

The trial was conducted from December 2023 to December 2025, and a total of 79 patients were included in the study. Patients had a mean age of 66 years and 58% were males (Table 1). Baseline clinical characteristics and demographics were generally comparable, although baseline renin and lactate concentrations were higher in the intervention group.

Table 1.

Baseline demographic, clinical, hemodynamic and treatment characteristics

Baseline characteristics	Tricycle group N = 38	Control group N = 41
Age, Mean ± SD	64.7 ± 11.3	67.2 ± 11.3
Gender, n (%male)	23 (60.5%)	23 (56.1%)
BMI, median (IQR), kg/m2	25.8 (22.1, 34)	27.8 (23.9, 32.2)
Underlying diseases, n (%)
Arterial hypertension	21 (55.3%)	23 (56.1%)
Diabetes mellitus	8 (21.1%)	11 (26.8%)
Atrial fibrillation‡	7 (21.9%)	10 (28.6%)
Chronic kidney disease stages 4–5	3 (7.9%)	3 (7.3%)
Coronary artery disease‡	1 (3.1%)	6 (17.1%)
Cirrhosis‡	6 (18.8%)	3 (8.6%)
Cerebrovascular disease‡	1 (3.1%)	2 (5.7%)
History of DVT or PE‡	5 (15.6%)	1 (2.9%)
Site of infection, n (%)
Respiratory tract	23 (60.5%)	24 (58.5%)
Urinary tract	1 (2.6%)	7 (17.1%)
Gastrointestinal tract	8 (21.1%)	6 (14.6%)
Skin and soft tissue	4 (10.5%)	3 (7.3%)
Other	2 (5.3%)	1 (2.4%)
Invasively mechanically ventilated, n (%)	8 (21.1%)	14 (34.1%)
Renal replacement therapy, n (%)	1 (2.6%)	1 (2.4%)
Creatinine, median (IQR), mg/dL	1.64 (1.1, 2.4)	1.60 (1.1, 2.9)
eGFR, median (IQR), mL/min/1.73m2	36 (26.5, 57.5)	41.5 (23.3, 65.5)
Renin, median (IQR), mU/L	203.6 (41.1, 500)	66.7 (20, 155)
Lactate, median (IQR), mmol/L	4.4 (3.2, 7.6)	3.5 (2.4, 5.6)
SOFA score¶, median (IQR)	10 (8, 12.3)	9 (7, 11)
NEE dose, median (IQR), µg/kg/min	0.39 (0.27, 0.56)	0.35 (0.2, 0.48)
Cardiac index§, median (IQR), L/min/m²	3.0 (2.6, 3)	2.85 (2.5, 3.3)
Hemoglobin, median (IQR), g/L	116.5 (95.8, 127.5)	116 (103.5, 131)
Volume of fluids/RBW†, median (IQR), mL/kg	24.3 (20, 33)	26 (19.3, 37.1)
Use of ACEi or ARBs, n (%)	17 (44.7%)	17 (41.5%)
Use of beta-blockers, n (%)	12 (31.6%)	17 (41.5%)
Time to randomization*, median (IQR), h	5.2 (3.9, 12)	7.1 (3.9, 9.8)

‡ Data were missing for 6 patients in the Tricycle group and 6 patients in the Control group

§ Data were missing for 5 patients in the Tricycle group and 3 patients in the Control group

† Data were missing for 3 patients in the Tricycle group and 1 patient in the Control group

Scores on Sequential Organ Failure Assessment (SOFA) range from 0 to 24, with higher scores indicating greater disease severity

* Defined as the earliest documented hospital presentation with sepsis or septic shock. Patients could present with established septic shock or progress from sepsis before meeting the eligibility criteria. Data were missing for 6 patients in the Tricycle group and 6 patients in the Control group

Abbreviations: SD standard deviation, IQR interquartile range, BMI body mass index, DVT deep vein thrombosis, PE pulmonary embolism, eGFR estimated glomerular filtration rate, NEE norepinephrine equivalence, RBW real body weight, ACEi angiotensin-converting enzyme inhibitors, ARBs angiotensin II receptor blockers

Primary outcome

Renin significantly declined over time in both groups with the observed decline being 2.61% per hour in the intervention group (95% CI 2.11% to 3.09%) and 1.67% per hour in the control group (95% CI 1.18% to 2.14%), p = 0.015, (Table 2; Fig. 3), corresponding to a computed reduction of 46.9% in the intervention group and 33.2% in the control group over 24 h.

Table 2.

Prespecified primary, secondary and exploratory outcomes and additional non-prespecified analyses

Outcome	Tricycle group N = 38	Control group N = 41	P
Prespecified primary outcome
Renin*, % change per hour (95% CI)	−2.61 (− 3.09, − 2.11)	−1.67 (− 2.14, − 1.18)	0.015
Prespecified secondary outcomes
Lactate*, % change per hour (95% CI)	−1.53 (− 1.76, − 1.29)	−1.01 (− 1.24, − 0.79)	0.006
ΔSOFA, change from baseline, median (IQR)	−3 (− 5.3, − 0.75)	0 (− 3.5, 2)	0.018
Prespecified exploratory outcomes
RRT¶ during ICU stay, n (%; 95% CI)	8 (21.6%; 11, 37)	11 (27.5%; 16, 43)	0.605
ICU mortality, n (%; 95% CI)	10 (26.3%; 15, 42)	17 (41.5%; 28, 56)	0.235
28-day mortality, n (%; 95% CI)	14 (36.8%; 23, 53)	19 (46.3%; 32, 61)	0.494
Additional non-prespecified analyses
NEE dose* (0–72 h), % change per hour (95% CI)	−5.41 (− 5.81, − 4.99)	−5.19 (− 5.59, − 4.80)	0.47
NEE dose* (0–24 h), % change per hour (95% CI)	−6.87 (− 7.93, − 5.80)	−3.03 (− 4.08, − 1.97)	< 0.0004
eGFR (0–72 h), change from baseline, mean (95% CI), mL/min/1.73m2	19.2 (9.9, 28.4)	8.2 (− 3.5, 19.8)	0.140

* Data availability at scheduled time points is presented in Supplementary Appendix Tables S8-10

¶ N = 77; 1 patient in the Tricycle group and 1 in the Control group had preexisting CKD on RRT and were not included in analysis

Abbreviations: CI confidence interval, NEE norepinephrine equivalence, eGFR estimated glomerular filtration rate, RRT renal replacement therapy, SOFA sequential organ failure assessment, IQR interquartile range, ICU intensive care unit

Fig. 3.

Plasma renin concentrations. Temporal trends in renin levels. Points represent raw data. The colored lines show the LMM predicted values, and the shaded area indicates the 95% confidence intervals

Secondary outcomes

Lactate significantly declined over time in both groups with the observed decline being 1.53% per hour in the intervention group (95% CI 1.29% to 1.76%) and 1.01% per hour in the control group (95% CI 0.79% to 1.24%), p = 0.006, (Table 2; Fig. 4), corresponding to a computed reduction of 30.9% in the intervention group and 21.7% in the control group over 24 h.

Fig. 4.

Plasma lactate concentrations. Temporal trends in lactate levels. Points represent raw data. The colored lines show the LMM predicted values, and the shaded area indicates the 95% confidence intervals

Comparison of the ΔSOFA score (SOFA score at day 3 minus the baseline SOFA score) showed a significantly greater decrease in the intervention group. The observed median change was − 3 in the intervention group (IQR − 5.3 to − 0.75) and 0 in the control group (IQR − 3.5 to 2), p = 0.018, (Table 2). The between-group difference persisted in a sensitivity analysis excluding patients who died during the first 72 h (Table S6 Supplementary Appendix). Available SOFA component scores are detailed in Supplementary Appendix Table S11.

Exploratory outcomes

During the ICU stay, 8 out of 37 patients (21.6%) in the intervention group and 11 out of 40 patients (27.5%) in the control group received RRT, (RR 0.79; 95% CI 0.36 to 1.74), p = 0.605, (Table 2), while 2 patients, one in each group, had preexisting CKD on RRT and were excluded from this analysis.

ICU mortality was 26.3% in the intervention group and 41.5% in the control group (RR 0.635; 95% CI 0.33 to 1.21), p = 0.235, (Table 2). 28-day mortality among groups was 36.8% vs. 46.3% (RR 0.795; 95% CI 0.47 to 1.35), p = 0.494, (Table 2; Fig. 5).

Fig. 5.

Kaplan-Meier survival plot. Kaplan-Meier survival curves for the Tricycle group and the Control group. The difference in survival between the groups was not statistically significant (Mantel-Cox test, p = 0.562)

Additional analyses

72-hour measurement period of NEE doses has shown a significant decline in both groups. The decline was 5.41% per hour in the intervention group (95% CI 4.99% to 5.81%) and 5.19% per hour in the control group (95% CI 4.80% to 5.59%), corresponding to a reduction of 73.7% in the intervention group and 72.2% in the control group. The observed difference was not statistically significant, p = 0.47, (Table 2; Fig. 6).

Fig. 6.

NEE dose requirement. Temporal trends in required NEE dose. Points represent raw data. The colored lines show the LMM predicted values, and the shaded area indicates the 95% confidence intervals. Abbreviations: NEE norepinephrine equivalence

In the additional analysis restricted to the first 24 h, the observed decline was 6.87% per hour in the intervention group (95% CI 5.80% to 7.93%) and 3.03% per hour in the control group (95% CI 1.97% to 4.08%), corresponding to a reduction of 81.9% in the intervention group and 52.3% in the control group over the first 24 h. The decline was statistically significantly greater in the intervention group, p < 0.0004, (Table 2; Fig. 6).

MAP and heart-rate trajectories during the first 72 h after randomization are presented in Supplementary Appendix Figures S2 − 3 and showed substantial overlap between treatment groups.

.

Analysis of kidney function showed a mean increase in eGFR of 19.2 mL/min/1.73m² in the intervention group (95% CI 9.9 to 28.4) and 8.2 mL/min/1.73m² in the control group (95% CI − 3.5 to 19.8), p = 0.14, (Table 2).

In an additional analysis evaluating the clinical relevance of renin kinetics, a 1%-point per hour slower decline in plasma renin concentrations was associated with higher odds of ICU mortality (OR 1.41; 95% CI 1.13 to 1.75), p = 0.002, 28-day mortality (OR 1.46; 95% CI 1.18 to 1.82), p < 0.001 and higher odds of RRT requirement during ICU stay (OR 1.61; 95% CI 1.23 to 2.1), p < 0.001. A slower decline in plasma renin concentrations also correlated with less improvement in SOFA score (Spearman ρ = 0.37), p = 0.001. There was no evidence that these associations differed between treatment groups (Table S7 Supplementary Appendix).

Adjunctive interventions

Daily fluid balance and hydrocortisone use

There were no statistically significant differences in daily fluid balance and daily hydrocortisone use during the first 72 h (Table S12 Supplementary Appendix). The observed median fluid balance was − 2.5 mL/kgRBW in the intervention group (IQR − 38 to 33.2) and 12.5 mL/kgRBW in the control group (IQR − 3.9 to 44), p = 0.173. The observed daily hydrocortisone use was 200 mg/day in the intervention group (IQR 193 to 233) and 200 mg/day in the control group (IQR 175 to 200), p = 0.073.

Use of vasopressin and additional vasopressors and inotropes

Vasopressin was administered to 23 of 41 patients (56.1%) in the control group and was initiated at a median NEE dose of 0.45 µg/kg/min (IQR 0.35 to 0.50) (Table S12 Supplementary Appendix) with no differences in the use of inotropes during the first 72 h. Dobutamine was administered to 6 patients in the intervention group (15.8%) and 5 patients in the control group (12.2%), p = 0.750. Levosimendan was administered to 1 patient in the intervention group (2.6%) and 0 patients in the control group, p = 0.481. In 4 (9.8%) patients in the control group, Ang II was used for refractory hypotension. No patients received milrinone, dopamine, epinephrine, or methylene blue.

Adverse events

We observed a lower rate of tachyarrhythmias in the intervention group compared to the control group (15.6% vs. 68.6%), p < 0.01, with the association persisting after post hoc adjustment for baseline atrial fibrillation and beta-blocker use (adjusted OR 0.07, 95% CI 0.02 to 0.26), p < 0.001. There were no differences in the rates of thrombosis, thrombocytopenia, hyponatremia and ischemic events (Table 3).

Table 3.

Adverse events

Adverse events	Tricycle group N = 38	Control group N = 41	P
Adverse event of any grade, n (%)	31 (81.6%)	36 (87.8%)	0.537
Total adverse events, n	47	61
Thrombosis*, n (%)	3 (7.9%)	1 (2.4%)	0.347
DVT or PE	2 (5.3%)	1 (2.4%)
Ischemic stroke	0	0
DIC	1 (2.6%)	0
Other	0	0
Tachyarrhythmias‡, n (%)	5 (15.6%)	24 (68.6%)	< 0.01
Thrombocytopenia, n (%)	29 (76.3%)	22 (53.7%)	0.059
Hyponatremia, n (%)	5 (13.2%)	4 (9.8%)	0.731
Cardiac ischemia, n (%)	2 (5.3%)	8 (19.5%)	0.089
Mesenteric ischemia, n (%)	2 (5.3%)	0	0.228
Digital ischemia, n (%)	1 (2.6%)	2 (4.9%)	0.999

*5 patients in the Tricycle group and 1 patient in the Control group had confirmed PE or DVT prior to randomization. Only new events are reported

‡ Tachyarrhythmias requiring pharmacological treatment or electrical cardioversion were recorded. Data were available for 32 patients in the Tricycle group and 35 patients in the Control group

Abbreviations: DVT deep vein thrombosis, PE pulmonary embolism, DIC disseminated intravascular coagulation

Sensitivity analyses of longitudinal outcomes

A sensitivity analysis accounting for right-censoring of renin measurements at the assay upper quantification limit of 500 mU/L produced results consistent with the primary analysis. Sensitivity analyses additionally adjusting all longitudinal mixed-effects models for fluid balance after multiple imputation produced materially unchanged estimates and conclusions compared with the primary analyses. In sensitivity analyses including subject specific random slopes for time, the estimated group-specific renin slopes remained similar in magnitude and direction, however the between-group difference was no longer statistically significant (Tables S3 − 5 Supplementary Appendix).

Discussion

In the TRICYCLE randomized controlled trial, early multimodal vasopressor therapy with Ang II, vasopressin and norepinephrine resulted in a more rapid decline in plasma renin and lactate concentrations and was associated with greater reduction in early vasopressor requirements, improved organ function and fewer tachyarrhythmias.

In septic shock, elevated plasma renin concentrations reflect alterations in the renin–angiotensin–aldosterone system which may occur at several levels. These include impaired generation of Ang II, possibly attributable to defects in angiotensin-converting enzyme activity, enhanced degradation of Ang II by peptidases and reduced availability of Ang II type 1 (AT1) receptors due to internalization or reduced expression. These alterations can occur either independently or in combination and are important contributors to vasodilation and vascular hyporeactivity characteristic of septic shock [8, 9]. In addition, hypotension and increased β1-adrenergic receptor stimulation secondary to sympathetic nervous system activation further stimulate renin release [19, 20]. Infusion of exogenous Ang II can at least partially counteract these alterations by restoring the AT1 receptor-mediated negative feedback on renin release. This mechanism is supported by a secondary analysis of the ATHOS-3 trial in which infusion of Ang II resulted in marked reduction in plasma renin concentrations within 3 h of treatment initiation [31]. Although the renin sampling period was longer in our study and baseline plasma renin concentrations were higher in the intervention group despite randomization, likely reflecting a chance imbalance due to the limited sample size, we similarly observed a faster decline in plasma renin concentrations over time in the intervention group. The magnitude and direction of the estimated renin trajectories remained similar in a sensitivity analysis additionally including for patient-specific random slopes for time, however uncertainty around between-group difference increased, and the difference was no longer statistically significant. This increased uncertainty may reflect substantial variability in renin trajectories between patients, potentially related to complex disturbances in RAAS signaling during septic shock. Previous studies have documented that elevated renin concentrations on admission and increasing concentrations during ICU stay are associated with mortality and outperform lactate in outcome prediction [19–24]. Consistently, in our study, a slower decline in renin concentrations was associated with higher odds of ICU mortality, 28-day mortality, RRT requirements during ICU stay and correlated with less improvement in SOFA score. Associations with mortality and RRT requirement were observed within both treatment groups, with no evidence that treatment allocation modified these associations (Table S7 Supplementary Appendix). These findings support the prognostic relevance of renin kinetics but do not establish that treatment induced reductions in renin concentrations improve clinical outcomes. Due to restoration of the AT1 receptor-mediated negative feedback, renin levels would be expected to decline with infusion of exogenous Ang II. A post hoc analysis of the ATHOS-3 trial suggested that patients with elevated renin had greater survival benefit from Ang II therapy [31], while a subsequent observational study from our group reported that persistent elevation in plasma renin concentration despite Ang II administration was associated with increased mortality [26]. Taken together, these findings suggest that renin kinetics may provide clinically relevant information during septic shock, including in patients treated with Ang II. However, whether modulation of RAAS and reduction in plasma renin concentrations during septic shock improves patient outcomes remains uncertain and requires further investigation.

In addition to a more rapid decline in renin concentrations, patients in the intervention group demonstrated a more rapid decline in lactate concentrations, lower NEE requirements during the first 24 h, greater improvement in SOFA score and fewer tachyarrhythmias.

Given the broadly similar achieved MAP in both study groups, the faster decline in lactate concentrations cannot be explained solely by restoration of macrocirculatory perfusion pressures. Reduced catecholamine exposure may have contributed, although differences in regional perfusion despite similar MAP targets cannot be excluded. Baseline lactate concentrations were also higher in the intervention group despite randomization and should therefore be considered when interpreting this finding. The contribution of reduced catecholamine exposure remains uncertain because norepinephrine, which was the predominant adrenergic vasopressor used in this study, exerts substantially less β₂-adrenergic effects than epinephrine and β₂-adrenergic receptor activation is considered the principal pathway mediating catecholamine-induced aerobic lactate production [32, 33].

Patients in the intervention group demonstrated a more rapid reduction in NEE during the initial 24 h, however this difference was no longer evident at 72 h. This early catecholamine-sparing effect is consistent with the reduction of background vasopressors following initiation of Ang II demonstrated in the ATHOS-3 trial [34] and the reduced norepinephrine requirements with vasopressin use reported in the VASST [35] and VANISH [36] trials. Recent meta-analyses and observational studies have also reported a catecholamine sparing effect with earlier vasopressin initiation [4, 17]. The lack of difference at 72 h suggests that early triple vasopressor therapy accelerated rather than fundamentally altered the trajectory of vasopressor requirement. Current Surviving Sepsis Campaign Guidelines recommend norepinephrine as the first-line vasopressor in septic shock, with vasopressin suggested as a second-line agent and epinephrine as a third-line agent [1]. These recommendations are based on low certainty evidence, and the guidelines acknowledge the need for further studies to better define the optimal selection, timing and combination of vasopressors. In this context, our findings support further investigation regarding concurrent administration of vasopressors acting through different neurohormonal pathways.

Patients in the intervention group also demonstrated greater improvement in SOFA score, primarily driven by the cardiovascular component, whereas the remaining SOFA components were broadly similar (Table S11 Supplementary Appendix).

Analysis of adverse events demonstrated a significantly lower incidence of tachyarrhythmias in the intervention group which is likely attributable to a lower stimulation of cardiac β1-adrenergic receptors with reduced catecholamine exposure. This finding is consistent with retrospective observational data comparing non-adrenergic vasopressors with adrenergic vasopressors in patients with vasodilatory shock [4]. Given that sepsis-driven tachyarrhythmias contribute to hemodynamic compromise and are associated with adverse outcomes [37], earlier introduction of Ang II and vasopressin may provide a mechanistic advantage without additional β1-adrenergic receptor stimulation and thereby reducing catecholamine-associated tachyarrhythmias.

Although thrombotic events occurred numerically more frequently in the intervention group, the absolute number of events was small, and the difference was not statistically significant. A history of DVT/PE was also more frequent at baseline in the intervention group (5 vs. 1 patient). Use of Ang II has been associated with concerns due to potential prothrombotic effects; however, retrospective analyses have not demonstrated an increased risk of thrombotic events [38]. Nevertheless, standard venous thromboembolism prophylaxis should be applied whenever not contraindicated.

No differences were observed between groups in ischemic events, hematologic abnormalities or electrolyte disturbances.

No significant differences were observed in the need for RRT, kidney function or mortality.

Although early triple vasopressor therapy with Ang II, vasopressin and norepinephrine was associated with a more rapid decline in plasma renin and lactate concentrations, accelerated vasopressor weaning and reduced tachyarrhythmias, it remains uncertain whether these physiological effects improve patient-centered outcomes. Confirmation in larger adequately powered trials is required to determine whether these physiological benefits translate into meaningful clinical benefit.

Strengths and limitations

This study has several strengths. It was a multicenter randomized controlled trial with a control arm reflecting current guideline-based management of septic shock. To the best of our knowledge, this is the first randomized trial evaluating the early introduction of three vasopressors with distinct mechanism of action. Serial plasma renin measurements provide novel insights into renin kinetics during septic shock.

This study also has limitations. First, baseline renin and lactate concentrations were higher in the intervention group despite randomization, likely reflecting chance imbalances. Although both biomarkers were analyzed using linear mixed-effects models incorporating repeated measurements, with fixed effects for time, treatment group, and treatment-by-time interaction, these baseline differences should nevertheless be considered when interpreting the estimated treatment effects. Second, this trial was not powered for non-primary outcomes and analyses of these outcomes should be considered exploratory and hypothesis-generating. Third, the primary endpoint was plasma renin concentration, which is a surrogate biomarker of RAAS activity, rather than a patient-centered clinical outcome. Although the intervention accelerated the reduction in plasma renin concentrations, it remains uncertain if this effect translates to improved clinical outcomes. Fourth, the study was single blinded and treating clinicians were aware of treatment allocation, therefore performance bias regarding vasopressor titration, fluid resuscitation, initiation of inotropes or renal replacement therapy and the use of other adjunctive therapies cannot be excluded. Tachyarrhythmias were identified and recorded by the treating clinical team and required therapeutic intervention to qualify as a recorded event, therefore ascertainment may have been influenced by the single blind design. In addition, tachyarrhythmia data were not available for 12 patients and therefore missing-data bias cannot be excluded. This design was chosen because of practical challenges associated with administration of three vasoactive agents under blinded conditions. Fifth, in accordance with the prespecified protocol, Ang II administration was permitted as a rescue therapy in persistent hypotension in patients requiring a norepinephrine base dose of > 0.5 µg/kg/min and vasopressin dose of 0.03 IU/min for ethical reasons to avoid withholding potential beneficial treatment. Subsequently Ang II was administered in 9.8% of patients in the control group, which may have reduced the contrast between treatment groups. Sixth, the highly selected study population of patients requiring NEE dose ≥ 0.15 µg/kg/min limits the generalizability of these findings to patients with less severe septic shock. Seventh, Ang II and vasopressin were introduced simultaneously, therefore individual contribution of each vasopressor cannot be reliably determined. It cannot be excluded that some of the observed effects could have been achieved with earlier initiation of Ang II or vasopressin. Eighth, calculation of NEE doses and cardiovascular SOFA component relied on published NEE conversion factors which represent pharmacologic estimates and may not fully account for interindividual variability in vasopressor responsiveness. Ninth, longitudinal measurements were truncated by death in a small proportion of patients, and the mixed-effects models did not explicitly account for potential informative missingness related to mortality. Tenth, although the estimated renin trajectories remained similar in magnitude and direction when subject-specific random slopes for time were included, uncertainty around between-group difference increased, and the difference was no longer statistically significant. Eleventh, several prespecified outcomes and subgroup analyses could not be evaluated because of limitations in data availability, including pre-inclusion kidney function measurements, incomplete follow-up and information regarding timing of medication exposure.

Finally, the longitudinal NEE analyses and eGFR analysis were not prespecified in the published protocol and should therefore be interpreted as hypothesis-generating.

Conclusion

By targeting complementary vasoactive pathways early in the course of septic shock, a multimodal strategy (Ang II, vasopressin and norepinephrine) favorably influenced renin and lactate kinetics, early vasopressor requirements, and other physiological outcomes. Whether these effects translate into meaningful benefits for patients remains uncertain and should be evaluated in a larger, adequately powered randomized trial.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

SOFA

Sequential organ failure assessment

NEE

Norepinephrine equivalence

RCT

Randomized clinical trial

ICU

Intensive care unit

Ang II

Angiotensin II

MAP

Mean arterial pressure

PiCCO

Pulse contour cardiac output

ECMO

Extracorporeal membrane oxygenation

CI

Cardiac index

MELD

Model for end-stage liver disease

CKD

Chronic kidney disease

RRT

Renal replacement therapy

AKI

Acute kidney injury

ACEi

Angiotensin-converting enzyme inhibitors

ARBs

Angiotensin II receptor blockers

LMM

Linear mixed-effects models

RBW

Real body weight

Author contributions

ŽK, AM, MG and GL designed the study. MG performed the statistical analysis. ŽK, AM, GL and MG interpreted the data. ŽK, AM, AVB and AV performed data collection. ŽK and MG created the figures and tables. ŽK, AM, IC, JF, MH, KJ, BK, MM, NP, FS, SS, AVB, AV, NE, DP and ASK performed patient acquisition. All the authors contributed to the drafting and writing of the manuscript, had full access to all the data, and approved the final version for publication.

Funding

The study was funded by an internal research grant from UMC Maribor (Grant No.: OZRD-IRP-2023/01). UMC Maribor has received a donation of Giapreza 2,5 mg/ml (donation contract number 00117/2023). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Before the trial was initiated, this trial received ethical approval from ethics committees in both participating institutions (KME RS 0120–212/2023-2711-13 and KME Rebro 02/013 AG). Written informed consent was obtained from all patients or their surrogates.

Consent for publication

Not applicable.

Competing interests

We declare the following disclosures: Author GL has received speaker fees from Medis, Paion and Viatris, and consultancy fees from Paion and Viatris. Author AM has received speaker fees from Medis. None of these were related to the present work.