Efficacy and Safety of Vasopressin and Terlipressin in Preterm Neonates: A Systematic Review

Introduction: The use of arginine vasopressin (AVP) and terlipressin to treat hypotension in preterm neonates is increasing. Our aim was to review the available evidence on the efficacy and safety of AVP and terlipressin for use in preterm neonates. Methods: MEDLINE, EMBASE, the Cochrane Central Register of Controlled Trials, Web of Science, and Google Scholar from inception to September 2021 were searched for studies of AVP and terlipressin in the treatment of hypotension of any cause in preterm neonates. Primary outcomes were improvement in end-organ perfusion and mortality. The risk of bias assessment and certainty of the evidence were performed using appropriate tools. Results: Fifteen studies describing the use of AVP (n = 12) or terlipressin (n = 3) among 148 preterm neonates were included. Certainly, the available evidence for the primary outcome of end-organ perfusion rated as very low. AVP or terlipressin were used to treat 144 and 4 neonates, respectively. Improvement in markers of end-organ perfusion was reported in 143 (99%) neonates treated with AVP and 3 (75%) treated with terlipressin. The mortality rate was 41% (n = 59) and 50% (n = 2) for neonates who received AVP and terlipressin, respectively. Hyponatremia was the most frequently reported adverse event (n = 37, 25%). Conclusion: AVP and terlipressin may improve measured blood pressure values and possibly end-organ perfusion among neonates with refractory hypotension. However, the efficacy–safety balance of these drugs should be assessed on an individual basis and as per the underlying cause. Studies on the optimal dosing, efficacy, and safety of AVP and terlipressin in preterm neonates with variable underlying conditions are critically needed.

Population: Preterm neonates born at less than 37 weeks' gestation with hypotension (defined as mean blood pressure less than gestational age or hypotension requiring fluid or vasoactive therapy) or persistent pulmonary hypertension. Intervention: Arginine vasopressin or terlipressin administered intravenously, initiated at any time, and for any duration as a primary or rescue treatment for hypotension or persistent pulmonary hypertension. Comparator: Standard treatment, placebo, or any other vasoactive agent. Outcomes: Primary Outcomes: (i) improvement in end-organ perfusion defined as an increase in systolic, diastolic or mean blood pressure, or an increase in urine output, a decrease in the need for inotropes, or a reduction in serum lactate as reported by the authors of the primary studies, (ii) mortality prior to discharge. Secondary Outcomes: (i) major neurosensory disability defined as moderate to severe motor or cognitive impairment or severe visual or hearing impairment as reported by the authors of the primary studies; and (ii) the occurrence of adverse events defined as peripheral tissue ischemia, gastrointestinal events (occurrence of perforation, necrotizing enterocolitis, or gastrointestinal bleed), hepatic events, renal events or hyponatremia as reported by the authors of the primary studies.

Data Sources
Ovid MEDLINE (1964-September 2021), EMBASE (1974-September 2021), Web of Science (1900( -September 2021, and the Cochrane Central Register of Controlled Trials (CEN-TRAL) were systematically searched. This search strategy, containing database-specific subject headings and text word terms for concepts, was first developed in MEDLINE (Ovid interface) and was translated as appropriate for the other databases ( Figure 1). We also searched the bibliographies of relevant studies and citations of the studies included for additional references. Using Google Scholar, we searched for relevant studies that were not commercially published, such as conference abstracts, dissertations, policy documents, and book chapters. No language or study design limitations were applied. We excluded animal studies and duplicate studies. A professional librarian peer-reviewed our strategy using the Peer Review for Electronic Search Strategies (PRESS) guideline [23] (Appendix A).
Secondary Outcomes: (i) major neurosensory disability defined as moderate to severe motor or cognitive impairment or severe visual or hearing impairment as reported by the authors of the primary studies; and (ii) the occurrence of adverse events defined as peripheral tissue ischemia, gastrointestinal events (occurrence of perforation, necrotizing enterocolitis, or gastrointestinal bleed), hepatic events, renal events or hyponatremia as reported by the authors of the primary studies.

Data Sources
Ovid MEDLINE (1964-September 2021), EMBASE (1974-September 2021), Web of Science (1900( -September 2021, and the Cochrane Central Register of Controlled Trials (CENTRAL) were systematically searched. This search strategy, containing database-specific subject headings and text word terms for concepts, was first developed in MEDLINE (Ovid interface) and was translated as appropriate for the other databases ( Figure 1). We also searched the bibliographies of relevant studies and citations of the studies included for additional references. Using Google Scholar, we searched for relevant studies that were not commercially published, such as conference abstracts, dissertations, policy documents, and book chapters. No language or study design limitations were applied. We excluded animal studies and duplicate studies. A professional librarian peer-reviewed our strategy using the Peer Review for Electronic Search Strategies (PRESS) guideline [23] (Appendix A).

Study Selection
All original research studies and conference abstracts describing AVP or terlipressin as primary or rescue treatment for hypotension of any cause in preterm neonates born at less than or equal to 37 weeks gestational age were eligible. These included randomized controlled trials (RCTs), quasi-RCTs, prospective and retrospective cohort studies, descriptive studies, case series, and case reports. We included studies with mixed populations (term and preterm neonates) if separate data for preterm neonates were available. Studies were eligible for inclusion irrespective of the dose, administration frequency, and duration of AVP or terlipressin treatment. Standard practice or other therapeutic interventions were the comparators in studies with a control group. If there was no comparator group, the reported efficacy and safety of AVP/terlipressin were extracted.
Our primary outcomes were: (1) improvement in end-organ perfusion, defined as an increase in systolic, diastolic, or mean blood pressure, or an increase in urine output, a decrease in the need for inotropes, or a reduction in serum lactate, as reported by the authors in the primary studies and (2) mortality before discharge. Our secondary outcomes

Study Selection
All original research studies and conference abstracts describing AVP or terlipressin as primary or rescue treatment for hypotension of any cause in preterm neonates born at less than or equal to 37 weeks gestational age were eligible. These included randomized controlled trials (RCTs), quasi-RCTs, prospective and retrospective cohort studies, descriptive studies, case series, and case reports. We included studies with mixed populations (term and preterm neonates) if separate data for preterm neonates were available. Studies were eligible for inclusion irrespective of the dose, administration frequency, and duration of AVP or terlipressin treatment. Standard practice or other therapeutic interventions were the comparators in studies with a control group. If there was no comparator group, the reported efficacy and safety of AVP/terlipressin were extracted.
Our primary outcomes were: (1) improvement in end-organ perfusion, defined as an increase in systolic, diastolic, or mean blood pressure, or an increase in urine output, a decrease in the need for inotropes, or a reduction in serum lactate, as reported by the authors in the primary studies and (2) mortality before discharge. Our secondary outcomes were: (1) major neurosensory disability, defined as a moderate to severe motor or cognitive impairment or a severe visual or hearing impairment, as reported by the authors of the primary studies and (2) occurrence of adverse events, defined as peripheral tissue ischemia, gastrointestinal events, hepatic events, renal events, or hyponatremia, as reported by the authors in the primary studies (Appendix B).
We used Covidence as the primary screening and data extraction tool. Two independent reviewers (AA, KS) screened the titles and abstracts of retrieved studies to assess their eligibility. The eligible studies were then reviewed in duplicate at the full-text level by the same reviewers. We resolved disagreements through discussion with a third reviewer (SSZ).

Data Extraction and Synthesis
Two reviewers (AA, KS) independently conducted data extraction from the full-text studies meeting the inclusion criteria using a standardized data extraction form developed in Covidence (Appendix C). We resolved any disagreements throughout the data extraction process through discussion with a third reviewer (SSZ). We conducted a narrative synthesis of the study results structured around the Population, Intervention, Comparator, Outcome (PICO) framework. We described the details of the population (gestational weeks at birth, birth weight, postnatal age, indication for treatment), intervention, comparator, and outcome.

Risk of Bias Assessment
Two independent reviewers (AA, KS) conducted a qualitative assessment of included studies. The Tool for Evaluating the Methodological Quality of Case Reports and Case Series [24], the adapted Retrospectoscope for Reducing Bias in Chart Review Studies [25] and the Cochrane Collaboration Risk of Bias 2.0 tool [26] were used, as appropriate. The scales that we proposed in our original protocol [22] were not applicable to the included studies, thus, we modified the scales for our risk of bias assessment based on applicability.

Assessment of the Certainty of Evidence
Two reviewers (AA, KS) rated the certainty of the evidence using the Cochrane Grading of Recommendations Assessment, Development, and Evaluation approach (GRADE) [27]. We resolved any disagreements through discussion with a third reviewer (SSZ).

Data Analysis
The included articles did not present comparative effect estimates. Therefore, we did not conduct a meta-analysis.

AVP
Twelve studies described the use of AVP for the treatment of hypotension in 144 neonates (219 episodes) [18,[28][29][30][31][32][33][34][35]38,39,41], with median (minimum, maximum) gestational age (GA) and postnatal age (PNA) of 26 (23,36) days, respectively. AVP was a rescue treatment in 10 studies (134 neonates, 199 episodes, 93%). One RCT examined the effect of AVP versus dopamine in the initial treatment of early transient hypotension (n = 10, 7%) [18]. If adequate mean BP was not reached after the highest study drug dose, one dose of intravenous hydrocortisone (1 mg/kg) was administered. The treatment groups were comparable in baseline characteristics, except for neonates in the vasopressin group that had a lower mean PaCO2 (p < 0.05) and bicarbonate (17.5 vs. 19.5, p-value not significant) and higher base deficit (10.3 vs. 8.9, p-value not significant) during study drug administration and received fewer doses of surfactant (p < 0.05; Table 2). Furthermore, seven neonates in the vasopressin and three in the dopamine treatment group received a normal saline bolus (10 mL/kg) before the study drug initiation. The study reported successful treatment, defined as reaching mean blood pressure 2 mmHg above gestational age (in weeks), for nine subjects in each treatment group, with three and one neonates in the vasopressin and dopamine groups also requiring hydrocortisone to achieve target blood pressure. Urine output was similar in the vasopressin and dopamine treatment groups (3.5 ± 1.4 vs. 4.4 ± 1.4), and there was no report of other markers of end organ perfusion. Mortality was reported as a secondary outcome and analyzed along with BPD as a composite outcome. Four and two neonates in the vasopressin and dopamine treatment group died, but composite outcome for BPD or death were similar in both groups (p = 0.26). In one case report, AVP was the initial pharmacotherapy for hypotension secondary to hypertrophic obstructive cardiomyopathy [39].

Terlipressin
Three case reports described the use of terlipressin for the treatment of refractory hypotension as a rescue therapy in four preterm neonates (four episodes) with a median (minimum, maximum) GA and PNA of 30 (25, 34) weeks and 7.5 (4, 11) days, respectively. Terlipressin was administered as a bolus with cumulative dosing of 0.12 and 0.2 mg/kg/day in two studies (three neonates, three episodes) and as a bolus and infusion of 5 mcg/kg followed by 1-10 mcg/kg/h in one study (one neonate, one episodes) [37].

Terlipressin
Three of the included studies (100%) with a total of three (75%) neonates (3 episodes) reported improvement in end-organ perfusion. In two (50%) of the cases, the increase in blood pressure was reported along with decreased inotropic use. None of the available studies reported an improvement in urine output or a decrease in serum lactate levels after terlipressin administration.

Mortality, Major Neurosensory Disability, and Adverse events Vasopressin
Of all neonates treated with AVP, 59 (41%) died. Examining the rate of mortality as per the underlying conditions, neonates with gastrointestinal (5 of 8, 68%) and cardiac disease 2 of 3, 66%) as the causes of refractory hypotension had the highest mortality rate.

Terlipressin
Two of four neonates (50%) treated with terlipressin for refractory septic and vasodilatory shock of unknown origin died. Normal neurodevelopmental assessments were reported for the remaining two survivors treated for refractory septic shock and PPHN [36,37]. There were no reports of adverse events among neonates treated with terlipressin.

Risk of Bias Assessment
We assessed the risk of bias of the included retrospective studies by answering the questions, adapted from the Retrospectoscope for Reducing Bias in Chart Review Studies [25]. All four studies did not provide information regarding six out of 10 required questions. Thus, the lack of sufficient data limited our ability to judge the risk of bias ( Table 2).
The included case series, case reports, and case studies were evaluated across four domains: risk of selection bias, ascertainment bias, causality bias, and reporting bias, using the Tool to Evaluate the Methodological Quality of Case Reports and Case Series [24]. Selection and causality were the domains through which bias might have been introduced in most studies (Table 3). We assessed the risk of bias for the included RCT using the Cochrane Risk of Bias 2.0 tool [26]. We judged the RCT to have a high risk of bias. The risks arising from the randomization process, missing outcome data and measurement of the outcome was low, while we judged the risk of bias from outcome selection and deviation from the intended intervention to be high ( Figure 3). The included case series, case reports, and case studies were evaluated across four domains: risk of selection bias, ascertainment bias, causality bias, and reporting bias, using the Tool to Evaluate the Methodological Quality of Case Reports and Case Series [24]. Selection and causality were the domains through which bias might have been introduced in most studies (Table 3). We assessed the risk of bias for the included RCT using the Cochrane Risk of Bias 2.0 tool [26]. We judged the RCT to have a high risk of bias. The risks arising from the randomization process, missing outcome data and measurement of the outcome was low, while we judged the risk of bias from outcome selection and deviation from the intended intervention to be high ( Figure 3).

Assessment of the Certainty of the Evidence
As we did not conduct a meta-analysis, we followed the recommendations of Murad et al. for rating the certainty of evidence in the absence of a single estimate of effect [41]. The certainty of the evidence for the primary outcome end-organ perfusion was very low ( Table 4). The risk of bias assessment for the RCT determined that the study had a high risk of bias. Furthermore, there was no information available about several risks of bias domains for the retrospective studies. Therefore, we judged the included studies to have serious methodological limitations. The participants, interventions/exposures, and comparators were directly comparable to our clinical question. We did not suspect indirectness. For imprecision, the total number of participants in the included studies did not meet the threshold of 400 (n = 148) [42]. Furthermore, none of the included studies reported a 95% confidence interval. Thus, we judged concerns about imprecision as serious. For inconsistency, the results showed an unclear direction of effect of AVP/terlipressin on end-organ perfusion (improved blood pressure in 92% of studies; decreased need for inotropes in 83% of studies; improved urine output in 50% of studies; decreased serum lactate in 29% of studies). We judged concerns about inconsistency as serious. We did not strongly suspect publication bias as both negative and positive studies were included in our review and our search strategy was comprehensive.

Certainty in the Evidence
End-organ perfusion assessed using the following outcomes: systolic, diastolic or mean blood pressure, need for inotropic support, urine output, and serum lactate.
The direction of effect was unclear; the majority of studies showed improvements in blood pressure and need for inotropes but not urine output and serum lactate.

Discussion
The results of this systematic review show that AVP may improve blood pressure and urine output among neonates with refractory hypotension of various aetiologies. The observed increase in urine output among neonates receiving AVP is supported by the hypothesis that in a state of septic shock where endogenous vasopressin is low, vasopressin increases the glomerular filtration rate of the kidneys [43]. Interestingly, in the four neonates receiving terlipressin, a synthetic analogue of AVP with targeted selectivity for V1 receptors, there was no report of increase in urine output. This finding, along with the long half-life of terlipressin raises concern for its use in preterm neonates with highly dynamic needs.
In the only available study (RCT) describing the use of vasopressin in preterm neonates with early transient refractory hypotension [18], although vasopressin increased blood pressure, no other outcomes related to the end-organ perfusion were reported. Furthermore, the higher number of neonates in the vasopressin group needing hydrocortisone treatment (three vs. one) raises question on efficacy of vasopressin. Investigators reported significantly lower CO 2 levels and less need for surfactant in neonates treated with vasopressin, as an indicator of beneficial effects of vasopressin on pulmonary hemodynamics. However, these neonates also had a lower bicarbonate (17.5 vs. 19.5) and higher base excess (10.3 vs. 8.9), which could indicate that the lower CO 2 was in compensation for the more severe metabolic acidosis. If this lower level of bicarbonate was an adverse outcome associated with vasopressin treatment is another important question in need of investigation. The mortality rate during the treatment period was similar in the two groups (one vs. one) but three more neonates in vasopressin vs. one in the dopamine treatment group died prior to discharge. The cause of death for these neonates is not reported. Considering this lack of data, in preterm neonates with hypotension in the immediate postnatal transitional period, where myocardial maladaptation, uncompensated decrease in preload and delayed adrenal recovery are the main known associated pathophysiology [44][45][46], the use of vasopressin cannot be recommended.
We examined mortality as per the underlying cause of refractory hypotension and found it to be consistently high (30-70%) among all different groups. As this population has a high background rate of morbidity and mortality already, distinguishing the possible adverse effects of vasopressin in these poorly controlled studies is impossible. Information on the neurodevelopmental outcome of these neonates was scarce. Data were available on only ten survivors (12%) in the AVP group and two (100%) in the terlipressin group. Although severe neurodevelopmental disability occurred in only two neonates receiving AVP (20%), the lack of methodological precision in the included studies precludes any interpretation of the findings.
Outcome assessment for the occurrences of adverse events was presented in 12 of the 14 studies on AVP and all studies on terlipressin, which included 144 and four neonates, respectively. Hyponatremia occurred in 25% of cases, with most episodes reported as non-severe (n = 31, 79%). Gastric perforation and hepatic necrosis were the only other serious adverse events that occurred (n = 2, 1%). Potent pressor action of vasopressin on hepatosplenic circulation and other microcirculatory blood flow is the main area of concern for its use in preterm neonates. Juvenile animal data have been conflicting with studies in septic models, showing a marked redistribution of portal, pancreatic, and even renal blood flow despite an increase in urine output [47]. This decrease in microcirculatory blood flow to the hepatic, pancreatic, upper gastrointestinal tract, and renal systems has been reported in up to 30% of studied species, suggesting the fast and sustained increase in blood pressure following administration of AVP or its analogues might not be associated with an increase in systemic blood flow. On the contrary, two studies on the effect of AVP in juvenile pigs receiving cardiopulmonary bypass or having mesenteric ischemia showed that this drug can preserve capillary density and tissue blood flow and limit endothelin expression as a marker of intestinal microcirculatory disturbance [48,49]. These conflicting results, although limited to animal models, raises caution in use of vasopressin in preterm neonates and emphasizes the urgent need for population pharmacokinetic-pharmacodynamic data. The use of these drugs should also be limited to randomized controlled trials, specifically in patients with underlying conditions, such as early transient hypotension, where the scarce available data do not support a favorable efficacy-safety balance.

Limitations
We attempted to collect all available literature on the use of AVP and terlipressin in the treatment of hypotension among preterm neonates. However, the existing evidence mainly consisted of studies at the lowest level of medical evidence. Thus, we must interpret the results cautiously, as they are subject to vulnerabilities, including the inability to establish a cause-effect relationship between exposures and outcomes. Only one of the included studies used an RCT design, but this study also remained limited in the small sample size and limited reports of important outcomes.

Conclusions
Our review suggests that AVP and terlipressin may improve measured blood pressure values and hemodynamic indices among neonates with refractory hypotension. However, clinicians need to assess the efficacy-safety balance when using these drugs in each preterm neonate with refractory hypotension, individually. The limited data obtained through the current systematic review is cautiously supportive of the use of AVP in preterm neonates with refractory septic shock. However, close monitoring of indices of end organ perfusion, including renal, hepatic, pancreatic and gastrointestinal is highly warranted. There is no biologic plausibility, dosing, efficacy, or safety data to support the use of AVP in neonates with early transient hypotension. As AVP appears to be a viable option in critically ill preterm neonates with refractory hypotension, studies on its optimal dosing, efficacy, and safety profile in preterm neonates with variable underlying conditions is critically needed.