The Vasopressin Loading for Refractory septic shock (VALOR) study: a prospective observational study

Background Vasopressin is a second-line vasoactive agent for refractory septic shock. Vasopressin loading is not generally performed because of the lack of evidence for its effects and safety. However, based on our previous findings, we hypothesized it can predict the responsibility to vasopressin infusion with safety, and prospectively examined it in the present study. Methods Vasopressin loading was performed via the intravenous administration of a bolus of 1 U, followed by its continuous infusion at 1U/h in patients with septic shock treated with ≥ 0.2 μg/kg/min noradrenaline. An arterial pressure wave analysis was conducted, and endocrinological tests were performed immediately prior to vasopressin loading. We classified patients into responders/non-responders based on mean arterial pressure (MAP) changes after vasopressin loading. Based on our previous findings, the lower tertile of MAP changes was selected as the cut-off. The change in the catecholamine index (CAI) after 6 h was assigned as the primary outcome. Digital ischemia, mesenteric ischemia, and myocardial ischemia during the admission period were prospectively and systematically recorded as adverse events. Results Ninety-two patients were registered during the study period and examined. Sixty-two patients with a MAP change > 22 mmHg were assigned as responders and the others as non-responders. Blood adrenocorticotropic hormone levels were significantly higher in non-responders. Stroke volume variations were higher in responders before loading, while stroke volume and dP/dtmax were higher in responders after loading. Median CAI changes were − 10 in responders and 0 in non-responders, which was significantly lower in the former (p < 0.0001). AUROC of MAP change with vasopressin loading to predict CAI change < 0 after continuous infusion was 0.843 with sensitivity of 0.92 and specificity of 0.77. Ischemia events were observed in 5 cases (5.4%). Conclusions Vasopressin loading may be safely introduced for septic shock. Vasopressin loading may be used to predict responses to its continuous infusion and select appropriate strategies to increase blood pressure. Supplementary Information The online version contains supplementary material available at 10.1186/s13054-023-04583-7.


Introduction
Vasopressin is one of the strongest vasopressor agents used to treat septic shock and acts via a different receptor from catecholamines 1.While randomized control trials previously demonstrated its efficacy [2-4, vasopressin causes adverse effects, such as ischemia events [5-7.Therefore, it is recommended as a second-line agent after noradrenaline to increase blood pressure 8.
Vasopressin is generally administered in a continuous infusion of up to 0.03 U/min (1.8 U/h) 8.However, continuous vasopressin administration takes some time to increase blood pressure from the decision to introduce vasopressin because the minimum concentration to increase blood pressure needs to be achieved 9, 10, although an urgent pressure increase is essential to treat septic shock refractory to noradrenaline.Furthermore, some patients respond poorly to vasopressin and, thus, other approaches need to be introduced immediately 11.
We previously examined vasopressin loading by intravenously administering a bolus of 1 U in 21 consecutive cases and retrospectively analyzed its effects 12.We demonstrated the potential of vasopressin loading to rapidly increase blood pressure and predict subsequent responses to its continuous infusion by identifying responders/non-responders to a bolus infusion without many adverse events.Based on these findings, we conducted a prospective observational study named the VAsopressin Loading for Refractory septic shock (VALOR) study, in which vasopressin loading was performed for patients with septic shock in whom vasopressin was required.We examined responses to bolus loading, assessed its safety, and clarified whether it may be used to predict responses to a subsequent continuous infusion using endocrinological and hemodynamic evaluations.

Ethics approval and informed consent
This clinical study was conducted after receiving approval from the Ethics Board of our hospital (2020-130).It was registered at the University Hospital Medical Information Network (UMIN) with registration number UMIN000044041 13 (https:// cente r6.umin.ac.jp/ cgiopen-bin/ ctr_e/ ctr_ view.cgi? recpt no= R0000 50288).We received informed consent from eligible participants or their proxies.

Patient selection
Patients admitted to the ICU at the Hitachi General Hospital Emergency and Critical Care Center between April 2021 and March 2023 were included.This medical and surgical ICU is reserved for patients from the emergency department, for patients exhibiting in-hospital acute deterioration, and severe patients following surgery.Inclusion criteria were as follows: (1) age ≥18 years, (2) septic shock (Sepsis-3 criteria), (3) requiring the administration of vasopressin following the continuous infusion of noradrenaline ≥0.2 μg/kg/min, (4) not given continuous steroid therapy until the administration of vasopressin, and (5) an arterial line inserted.Study exclusion criteria were designated as end of life/terminal care.

Vasopressin administration protocol
Vasopressin was introduced to patients with septic shock who were administered noradrenaline ≥ 0.2 μg/kg/min.In the study protocol, vasopressin was intravenously administered as a bolus of 1 U for loading, followed by its continuous intravenous administration at 1 U/h.Prior to vasopressin loading, the following procedures were performed: (1) blood sampling for endocrinological tests and (2) ProAQT (Getinge, Japan) was connected to the arterial line for an arterial pressure wave analysis.Clinical practices were not limited other than vasopressin loading and its subsequent continuous infusion.

Outcomes and measurements
The change in the catecholamine index CAI 6 h after vasopressin loading was assigned as the primary outcome 13.CAI was calculated with dopamine + dobutamine + (noradrenaline + adrenaline) × 100 μg/kg/min.The catecholamine dose was controlled independently from vasopressin loading by an Emergency and Critical Care physician to maintain the optimal blood pressure.The target of blood pressure was not specified by protocol and determined by each physician.The change in CAI ΔCAI 2 and 4 h after the introduction of vasopressin, urine output every 2 h after the introduction of vasopressin, 24 h fluid IN/OUT balance after vasopressin loading, lactate and pH changes for 2 h, the total vasopressin administration time, steroid use after the administration of vasopressin, in-hospital mortality, and the lengths of intensive care unit (ICU) and hospital stays were analyzed as secondary outcomes.
As reported in our previous study 12, we classified patients into responders and non-responders to vasopressin loading based on changes in mean arterial pressure (MAP) just after vasopressin loading.In this prospective study, the maximum MAP change was prospectively and carefully observed by the physicians within 3-5 min after loading, and was recorded as MAP change (ΔMAP).Since we previously selected a cut-off for a loading response from MAP change retrospectively obtained from medical records, we expected a different ΔMAP distribution and cut-off for this prospective study.Therefore, to follow our previous study in which the cut-off of 18 mmHg was the lower tertile, we herein selected the lower tertile of ΔMAP as the cut-off in this study.
We recorded hemodynamics, including an arterial pressure wave analysis, just before vasopressin loading and when maximum ΔMAP was achieved.Blood pressure from the arterial line, heart rate, cardiac output CO, stroke volume, stroke volume variation (SVV), central venous pressure, systemic vascular resistance SVR, and dPmx, the maximum pressure increase over time calculated in ProAQT, were evaluated.dPmx, generally expressed as dP/dt max , is expected to represent left ventricular contractility 14, 15, however, the limitation has been demonstrated that preload and actual aortic pressure also affected dP/dt max 16.Digital ischemia, mesenteric ischemia, and myocardial ischemia during the admission period were prospectively and systematically recorded as adverse events.
Age, sex, height, body weight, acute physiology and chronic health evaluation (APACHE II) scores and sequential organ failure assessment (SOFA) scores on admission, comorbidities, and atrial fibrillation at the administration of vasopressin were analyzed as basic characteristics.The use of adjunctive therapies, such as mechanical ventilation, blood purification, and extracorporeal membrane oxygenation, was extracted.The timing of the administration of vasopressin from the initiation of noradrenaline, the noradrenaline dose and CAI when vasopressin was introduced, and maximum lactate levels within 24 h of the administration of vasopressin were also evaluated.In endocrinological tests, v, adrenocorticotropic hormone (ACTH), and cortisol levels were evaluated just before vasopressin loading.

Sample size estimation
The primary outcome was already examined in 21 consecutive cases with significance in our previous retrospective study 12.Therefore, the sample size for the safety analysis was estimated in the present study.Previous studies reported ischemia events in 5-10% of cases without loading [2][3][4]12].When the adverse events ratio was assumed to be 5%, 73 patients were required to detect more than a + 5% increase with a 5% error.Therefore, we estimated that a maximum of 100 cases needed to be registered.

Statistical analysis
The normality of the distribution of each parameter was assessed using the Shapiro-Wilk test.The significance of differences was evaluated using the Student's t-test and chi-squared test for parametric data.The Mann-Whitney U test was performed for non-parametric data.The area under the receiver operating characteristic (AUROC) for ΔCAI 6 h < 0 was calculated with factors that significantly differed between responders and nonresponders to vasopressin loading.For the sensitivity analysis, we performed the same analysis with the cut-off of 18 mmHg ΔMAP for responder/non-responder identification similarly with the previous study.All statistical analyses were conducted using JMP 14 software (SAS Institute Inc., Cary, NC, USA).Results were expressed as a mean ± standard deviation for normally distributed continuous variables with parametric test, or median (interquartile range) for non-normally distributed continuous variables with non-parametric test.P-values < 0.05 were considered to be significant.

Results
A study outline is shown in Fig. 1.In the study period, 365 patients with septic shock were admitted into the ICU, and 215 were treated with ≥ 0.2 noradrenaline.Among them, 90 patients did not require the infusion of vasopressin and 33 for whom vasopressin loading was not performed were not registered to this study.Therefore, 92 patients were registered and administered a bolus of 1 U vasopressin followed by its continuous infusion at 1 U/h.The lower tertile of ΔMAP after vasopressin loading was 22 mmHg; therefore, 62 patients with ΔMAP > 22 mmHg were assigned as responders and 30 with ΔMAP ≤ 22 mmHg as non-responders.
Outcomes are shown in Table 2 and Fig. 2. As the primary outcome, ΔCAI 6 h was − 10 in responders versus 0 in non-responders, which was significantly lower in the former (p < 0.0001).The AUROC of ΔMAP by vasopressin loading to predict ΔCAI 6 h < 0 was 0.843 (Additional file 1: Fig. S1).ΔMAP > 22 mmHg, the definition of responders in the present study, had sensitivity of 0.92 and specificity of 0.77 for ΔCAI 6 h < 0. Furthermore, ΔCAI 2 h and 4 h were significantly lower in responders (Fig. 2).This dataset did not have sufficient power to detect a statistically significant difference in in-hospital mortality between responders and non-responders (37.7% vs. 57.1%,p = 0.087).The length of hospital stay and duration of mechanical ventilation/blood purification were also not significantly different.Urine output 2, 4, and 6 h after the infusion of vasopressin was significantly higher in responders, and fluid volume IN and net IN-OUT balance were also significantly lower in responders.Blood lactate changes after 2 h were significantly lower with a minus value as the median in responders.The final duration of the administration of vasopressin was not significantly different between the groups, whereas the rate of steroid use after the introduction of vasopressin was higher in non-responders.As adverse effects, ischemia events were observed in 5 cases (5.4%).Digital ischemia occurred in 2 (7.1%) non-responders and 0 (0%) responders, which was significantly higher in the former (p = 0.030).No significant differences were noted in mesenteric ischemia or cardiac ischemia.

Discussion
Responses to vasopressin loading may predict CAI changes after its continuous infusion with adequate sensitivity.Adverse events were not common following the administration of a bolus of 1 U vasopressin.Although responses correlated with some hemodynamics and endocrinological changes, they alone did not predict responses to the continuous infusion of vasopressin.
Rapid increases in blood pressure may be achieved with vasopressin loading because the maximum MAP change was observed within a few minutes in the present study.Since a continuous infusion requires more time for its administration to patients due to its preparation and priming/titration, vasopressive agents, including catecholamines, may need to be loaded in a bolus for emergency and critical care cases 17.Vasopressin increases blood pressure more slowly than catecholamines.Vasoconstriction and blood pressure increases induced by vasopressin are only achieved when plasma vasopressin concentrations are higher than 50 pg/ml 10.In contrast, noradrenaline and other catecholamines induce vasoconstriction linearly from the lowest concentration 18.These differences in the concentration-vasoconstriction relationship are due to the V1 and α1 receptors 19.Moreover, the half-life of vasopressin is 10-35 minutes 9, while that of catecholamines is a few minutes 20.Therefore, vasopressin loading appears to be appropriate for achieving a blood pressure target and steady state.
The bolus administration of vasopressin has not yet been examined in detail in clinical settings.Terlipressin, an analogue of vasopressin with a longer half-life 21, was administered with bolus loading in some clinical trials 22. Adverse events were only observed in patients in whom massive loading was performed 23.Regarding vasopressin, one small study reported 7 cases for which a bolus of 50 mU/kg vasopressin was administered, and 4 died due to mesenteric ischemia 24.In the present study, the frequency of adverse events was adequately low and similar to those reported in previous clinical trials on vasopressin without loading 3. The present results indicate that loading with a bolus dose of 1 U is safe.
It is important to predict responses to vasopressin because responses to its continuous infusion differ among patients.Although marked improvements occur in some patients 1, others do not respond to vasopressin 11, and, thus, other blood pressure management strategies need to be introduced.Strategy development based on responses to vasopressin loading is required, including a continuous infusion of vasopressin for responders and another intervention, such as the infusion of steroids or epinephrine, for non-responders without waiting to attempt the continuous infusion of vasopressin.Actually, it would be rather harmful that lactate was increased and cardiac output was decreased significantly in nonresponders in this study.Randomized control trials to examine the effects of vasopressin loading with these strategy changes on the prognosis of refractory septic shock are warranted in the future.
In the present study, endocrinological testing immediately prior to vasopressin loading was performed to assess relationships with responses to vasopressin.Previous studies demonstrated that the concentration of vasopressin was elevated in early septic shock, but decreased within 48 h as shock continued 25, 26, which is referred to as vasopressin deficiency.While vasopressin deficiency did not correlate with responses to vasopressin, the timing of the administration of vasopressin from the shock onset was longer in responders in the present study.These findings indicated that responders might include the less sick patients, or that there might exist the vasopressin deficiency in which it could not be explained by the absolute vasopressin concentration at one point, such as hypovolemia.Meanwhile, ACTH concentrations were significantly higher in non-responders.Vasopressin is one of the feedback regulators of ACTH release via the pituitary vasopressin 3 receptor 27.Although vasopressin concentration was not different between the groups, vasopressin excretion might have been continued longer in non-responders.In these patients, relative adrenal insufficiency was suspected and steroid supplementation may be an alternative treatment.One of the reasons why some randomized control trials showed that the effectiveness of vasopressin was maximized with steroid use 11 may be that steroids acted as a rescue therapy for vasopressin non-responders.
SVV, the dynamic index for fluid responsibility, was higher in responders.Since vasopressin induces venous vessel contraction 9, it may have increased venous return by contracting stressed volume.dP/dt max is a marker for cardiac contractility, when the appropriate preload and actual aortic pressure were maintained 15, 16.Vasopressin theoretically does not affect cardiac contractility, however, it would achieve a favorable hemodynamic adjustment with vaso-constriction, which resulted in dP/ dt max increase.If the cardiac contractility was originally better, vasopressin would be more effective, corresponding to our results.
These significant endocrinological and hemodynamics differences would be reasonable for vasopressin responses, but did not adequately predict the responses.Responses to vasopressin may be affected by many factors other than hemodynamics and endocrinological changes; therefore, vasopressin loading may be important for predicting these responses.
Furthermore, vasopressin loading increased urine output in a short period in responders.Because vasopressin increased vascular resistance more in efferent glomerular arterioles than in afferent glomerular arterioles 28, it was demonstrated that vasopressin could have increased urine output in early phase comparing with noradrenaline in septic shock 1.As the urine output was significantly different between responders and non-responders, vasopressin loading might have the predictability of renal impacts in septic shock.
There are several limitations that need to be addressed.This was an observational study.Therefore, a randomized control trial is needed to confirm the effects of vasopressin loading.Furthermore, the management of CAI was performed by an Emergency and Critical Care physician, who was not always available to make rapid adjustments to CAI in ICU patients.In addition, since Japanese ICU patients are often smaller and older than those in Western countries, a dose of 0.03 U/min (1.8 U/h) may be too high; therefore, we adopted the described protocol.

Conclusions
Vasopressin loading may be introduced for septic shock to safely achieve a rapid increase in blood pressure.Vasopressin loading may be used to predict responses to its continuous infusion and select appropriate strategies to increase blood pressure.
Additional file 2. Tables S1, S2 and S3.The sensitivity analysis of the baseline, outcomes and hemodynamics with the cut-off of 18 mmHg ΔMAP for responder/non-responder identification.

Fig. 2
Fig. 2 Changes in the catecholamine index in responders/non-responders to vasopressin loading.Responders to vasopressin loading were defined by a mean arterial pressure change > 22 mmHg after loading.Changes in the catecholamine index (CAI) (dopamine + dobutamine + (noradrenaline + adrenaline) × 100 μg/kg/min) from before loading to 2, 4, and 6 h after the initiation of vasopressin are shown.A Post 2 h-pre.B Post 4 h-pre.C Post-pre.Changes in CAI significantly differed between responders and non-responders at 2, 4, and 6 h

Table 1
Differences in baseline characteristics between responders and non-responders to vasopressin loadingData were expressed as a mean ± standard deviation for normally distributed continuous variables or median (interquartile range) for non-normally distributed continuous variables.Categorical variables were expressed as n (%) MAP mean arterial pressure; SOFA sequential organ failure assessment; APACHE acute physiology and chronic health evaluation; ACTH adrenocorticotropic hormone

Table 2
Differences in outcomes between responders and non-responders to vasopressin loading Data were expressed as a mean ± standard deviation for normally distributed continuous variables or median (interquartile range) for non-normally distributed continuous variables.Categorical variables were expressed as n (%) MAP mean arterial pressure; ICU intensive care unit

Table 3
Differences in hemodynamics between responders and non-responders to vasopressin loadingData were expressed as a mean ± standard deviation for normally distributed continuous variables or median (interquartile range) for non-normally distributed continuous variables MAP mean arterial pressure; SBP systolic blood pressure, DBP diastolic blood pressure; HR heart rate; CO cardiac output; SV stroke volume; SVV stroke volume variation, SVR systemic vascular resistance; CVP central venous pressure