Association between Serum Lactate and Morbidity and Mortality in Neonates: A Systematic Review and Meta-Analysis

Objective: Lactate is a marker of hypoperfusion in critically ill patients. Whether lactate is useful for identifying and stratifying neonates with a higher risk of adverse outcomes remains unknown. This study aimed to investigate the association between lactate and morbidity and mortality in neonates. Methods: A meta-analysis was performed to determine the association between blood lactate levels and outcomes in neonates. Ovid MEDLINE, EMBASE, Cochrane Library, and ClinicalTrials.gov were searched from inception to 1 May 2021. A total of 49 observational studies and 14 data accuracy test studies were included. The risk of bias was assessed using the Newcastle-Ottawa Scale for observational studies and the QUADAS-2 tool for data accuracy test studies. The primary outcome was mortality, while the secondary outcomes included acute kidney injury, necessity for renal replacement therapy, neurological outcomes, respiratory morbidities, hemodynamic instability, and retinopathy of prematurity. Results: Of the 3184 articles screened, 63 studies fulfilled all eligibility criteria, comprising 46,069 neonates. Higher lactate levels are associated with mortality (standard mean difference, −1.09 [95% CI, −1.46 to −0.73]). Using the estimated sensitivity (0.769) and specificity (0.791) and assuming a prevalence of 15% for adverse outcomes (median of prevalence among studies) in a hypothetical cohort of 10,000 neonates, assessing the lactate level alone would miss 346 (3.46%) cases (false negative) and wrongly diagnose 1776 (17.76%) cases (false positive). Conclusions: Higher lactate levels are associated with a greater risk of mortality and morbidities in neonates. However, our results do not support the use of lactate as a screening test to identify adverse outcomes in newborns. Research efforts should focus on analyzing serial lactate measurements, rather than a single measurement.


Introduction
Lactate is a powerful parameter that can be used to indirectly assess the hemodynamic system, but only when used correctly [1].In critically ill patients, lactate is a classical marker, where its elevation is associated with greater morbidity and mortality [1].Hyperlactatemia is a hallmark parameter in shock states because of lactate production in anaerobic metabolism, representing a state where there is an inadequate oxygen supply [2].In adult and pediatric literature, there is strong evidence that lactate is a predictor of mortality [3,4].Unfortunately, evidence about the utility of lactate measurement in neonates is limited.
There is no consensus regarding the treatment of hemodynamic instability in neonates, especially in preterm infants [5].Moreover, classical parameters that are used to evaluate the cardiovascular system such as blood pressure alone are still not reliable in the neonatal period.Currently, there still is no definition for hypotension in neonates, nor a consensus of whether its correction is beneficial [6].
In this context, being able to use a parameter that aids in the diagnosis and treatment of hemodynamic instability in newborns would be valuable.The aim of this systematic review and meta-analysis was to determine the association between blood lactate concentration and morbidity and mortality in neonates.

Material and Methods
This systematic review and meta-analysis followed the recommendations based on the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analysis) [7] and the Cochrane Centre for Reviews and Dissemination [8].The search strategy was developed according to recommendations of PRESS [9] (Peer Review of Electronic Search Strategies) and was executed in May 2021.Ovid MEDLINE, EMBASE, Cochrane Library, and trial registries were searched without publication or language restrictions (see the Search Strategy in Figure S1).All references from retrieved citations were searched for additional relevant studies.The Rayyan web app [10] was used for study selection and initial abstract and title screening.The PRISMA flowchart is presented in Figure 1.We did not find any randomized controlled trials.Data extraction was performed by two authors (FYM and VLJK) and plotted in a previously built structured data extraction form.Any unresolved discrepancies of extracted data were resolved by a third author (WBC).
Children 2023, 10, x FOR PEER REVIEW 2 of 20 In this context, being able to use a parameter that aids in the diagnosis and treatment of hemodynamic instability in newborns would be valuable.The aim of this systematic review and meta-analysis was to determine the association between blood lactate concentration and morbidity and mortality in neonates.

Material and Methods
This systematic review and meta-analysis followed the recommendations based on the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analysis) [7] and the Cochrane Centre for Reviews and Dissemination [8].The search strategy was developed according to recommendations of PRESS [9] (Peer Review of Electronic Search Strategies) and was executed in May 2021.Ovid MEDLINE, EMBASE, Cochrane Library, and trial registries were searched without publication or language restrictions (see the Search Strategy in Figure S1).All references from retrieved citations were searched for additional relevant studies.The Rayyan web app [10] was used for study selection and initial abstract and title screening.The PRISMA flowchart is presented in Figure 1.We did not find any randomized controlled trials.Data extraction was performed by two authors (FYM and VLJK) and plotted in a previously built structured data extraction form.Any unresolved discrepancies of extracted data were resolved by a third author (WBC).Two authors (FYM and VLJK) independently screened titles and abstracts and reviewed them.When the title and abstract were insufficient to decide on eligibility criteria, the full text was retrieved.If there was an unresolved disagreement, a third author (WBC) was consulted.All selected studies were retrieved and applied to a predefined inclusion criterion.The eligibility criteria included: (1) the study covered a neonatal Two authors (FYM and VLJK) independently screened titles and abstracts and reviewed them.When the title and abstract were insufficient to decide on eligibility criteria, the full text was retrieved.If there was an unresolved disagreement, a third author (WBC) was consulted.All selected studies were retrieved and applied to a predefined inclusion criterion.The eligibility criteria included: (1) the study covered a neonatal population or a specific neonatal subgroup analyzed separately (<6 weeks postnatal age or < corrected gestational age of 40 weeks); (2) the study had at least one lactate measurement with a defined time assessment point; and (3) the study reported at least one outcome of interest.Studies that included pediatric patients were only eligible if data for neonates could be extracted separately.Studies reported only as abstracts were eligible only if sufficient information was available.If multiple articles analyzed the same set of patients, we included only the article with the largest number of neonates.This systematic review and meta-analysis followed the previously published protocol registered with the PROSPERO International Prospective Register of Systematic Reviews (CRD42021253329).Protocol changes are given in the Supplementary Material.
The primary objective was to evaluate the impact of hyperlactatemia on mortality in neonates during hospital stay.Composite outcomes with survival data were analyzed as mortality.Secondary outcomes included acute kidney injury, renal replacement therapy necessity, neurological outcomes, respiratory morbidities, hemodynamic instability, and retinopathy of prematurity.
We grouped the timing of lactate assessment into two different groups: early (lactate measured within 3 days of life or less) and late (lactate measured after more than 3 days of life).Lactate collected from the umbilical cord was analyzed separately.Initially, we planned to divide the lactate collected from venous and arterial sources, but due to insufficient data from the studies, this division was not possible.If studies assessed lactate at multiple time points, the earliest post-condition/intervention point or the highest value was selected.Hyperlactatemia was defined according to each study definition.
We assessed the risk of bias of included studies using the Newcastle-Ottawa Scale [11] for nonrandomized studies.A study with a total score of 7 or higher was considered of good quality, a study with a score of 4 to 6 was considered of fair quality, and a study with a score of lower than 4 was considered of poor quality.To assess the risk of bias of diagnostic accuracy studies, we used the QUADAS-2 tool [12].
For dichotomous variables, we used the odds ratio (OR) as the common measure of association with its respective 95% confidence interval.Lactate as a continuous variable was reported as standard mean differences (SMDs) with their respective 95% confidence interval.When studies reported medians and interquartile intervals, we used Wan et al.'s formula to infer the mean value and standard deviation [13].To meta-analyze, we used random-effects models as proposed by Der Simonian and Laird because of the anticipated heterogeneity between studies.Statistical analysis was performed using RevMan 5 (Review Manager 5) software, v5.4,The Cochrane Collaboration.Heterogeneity was analyzed by performing subgroup analysis based on subgroup population and was measured using I 2 statistics where estimates higher than 50% were considered as indicating significant heterogeneity.A p-value lower than 0.05 was considered statistically significant.
Following the recommendations from the Cochrane Screening and Diagnostic Test Methods Group, one author (FYM) extracted diagnostic data and derived the number of true-positive, false-positive, true-negative, and false-negative cases.A second author (VLJK) checked the extracted data, and if a consensus was not reached, a third author (WBC) was consulted.We then created forest plots with 95% confidence intervals (Cis) for sensitivity and specificity using RevMan 5 (Review Manager 5) software, v5.4,The Cochrane Collaboration.A hierarchical summary ROC model was used because the reported cutoff levels for lactate differed among included studies.A meta-analysis of diagnostic test accuracy studies was performed using MetaDTA (web-based tool v2.0) [14], and estimates of sensitivity and specificity were calculated.Heterogeneity was assessed by analyzing the forest plots of sensitivity and specificity across studies.

Results
Among 3184 records, a total of 185 potentially relevant articles were screened and fully retrieved (Figure 1).Of those, 63 studies, including 46,069 newborns (sample sizes ranged from 16 to 21,182 neonates), met the full inclusion criteria (Tables 1 and S1).No randomized controlled trials were found.Studies excluded from the systematic review and the justification for their exclusion are presented in Table S2.The majority of studies (57%) were conducted in North America and Europe, 20 were conducted in Asia-Oceania, 6 were conducted in Latin America, and 1 was conducted in Africa.All studies were published between 1994 and 2021, with most studies (46 of 63) published after 2010.Among the studies, 14 evaluated lactates in preterm infants, 13 evaluated lactates in neonates with infants with congenital heart disease (CHD), and 12 evaluated lactates in neonates with birth asphyxia.The mean lactate levels in the nonsurvivor group varied between 2.2 and 23.42 mmoL/L.After applying the Newcastle-Ottawa Scale, 36 studies were labeled as being of good quality and 13 as fair quality (Table S3).The main potential sources of bias were "Representativeness of cohort" and "Comparability".
We identified 14 studies analyzing lactate in data accuracy tests, comprising 39,540 patients.The characteristics of the included studies are summarized in Table S1.The cut-off levels for lactate ranged from 2.5 to 9.95 mmoL/L.The main potential source of bias was "Patient Selection".
However, higher lactate levels from umbilical cord blood are associated with a higher risk of worse outcomes (Figure S10).

Data Accuracy Test for Adverse Outcomes
The estimate of sensitivity was 0.769 (95% CI, 0.692-0.831),and that of specificity was 0.791 (95% CI, 0.718-0.850).We observed a high heterogeneity among the studies, with a wide variety of sensitivity and specificity estimates.The prevalence rates of adverse outcomes ranged widely from 0.18% to 75%.We then applied the DTA estimates for sensitivity (0.769) and specificity (0.791) from our meta-analysis to a hypothetical cohort of 10,000 neonates with a prevalence rate of adverse outcomes of 0.18% (resulting in a median of 4.15 cases of adverse outcomes being missed and 2086 cases being wrongly diagnosed as an adverse outcome), 15% (resulting in a median of 346 cases of adverse outcomes being missed and 1776 being wrongly diagnosed as an adverse outcome), 50% (resulting in a median of 1155 cases of adverse outcomes being missed and 1045 cases being wrongly diagnosed as an adverse outcome), and 75% (resulting in a median of 1732 cases of adverse outcomes being missed and 522 being wrongly diagnosed as an adverse outcome).Summary receiver operating characteristic curves and forest plots are presented in Figure S11 and Figure 5, respectively.We applied the QUADAS-2 tool to assess the quality of studies, and the risk of bias was low (Figure S12).cases of adverse outcomes being missed and 522 being wrongly diagnosed as an adverse outcome).Summary receiver operating characteristic curves and forest plots are presented in Figure S11 and Figure 5, respectively.We applied the QUADAS-2 tool to assess the quality of studies, and the risk of bias was low (Figure S12).

Discussion
This systematic review and meta-analysis support the hypothesis that higher lactate levels are associated with increased mortality and risk of morbidities (AKI, RRT necessity, respiratory complications, hemodynamic instability, and neurological deficit) in neonates.This observation was similar across different subgroups of patients, from preterm infants to neonates with birth asphyxia.Although the data are robust and consistent, their interpretation is complicated due to the heterogeneity between studies, with different conditions and times of assessment.Indeed, the included studies had heterogeneity as high as 95%, even after subgrouping into more homogeneous groups, explaining the wide range of sensitivity (43-100%) and specificity (39-95%) in the data accuracy test for adverse outcomes.Using the estimated sensitivity (0.769) and specificity (0.791) and assuming a prevalence of 15% for adverse outcomes (the median of prevalence among studies) in a hypothetical cohort of 10,000 neonates, assessing lactate level alone would

Discussion
This systematic review and meta-analysis support the hypothesis that higher lactate levels are associated with increased mortality and risk of morbidities (AKI, RRT necessity, respiratory complications, hemodynamic instability, and neurological deficit) in neonates.This observation was similar across different subgroups of patients, from preterm infants to neonates with birth asphyxia.Although the data are robust and consistent, their interpretation is complicated due to the heterogeneity between studies, with different conditions and times of assessment.Indeed, the included studies had heterogeneity as high as 95%, even after subgrouping into more homogeneous groups, explaining the wide range of sensitivity (43-100%) and specificity (39-95%) in the data accuracy test for adverse outcomes.Using the estimated sensitivity (0.769) and specificity (0.791) and assuming a prevalence of 15% for adverse outcomes (the median of prevalence among studies) in a hypothetical cohort of 10,000 neonates, assessing lactate level alone would miss 346 cases (false negative) and wrongly diagnose 1776 cases (false positive).
Lactate is a widely used marker of altered tissue perfusion in critically ill patients, especially in adults, where hyperlactatemia is an indispensable feature that can be used to evaluate shock state.However, altered blood lactate cannot be attributed exclusively to anaerobic metabolism [72].Other physiopathology mechanisms, including glycolysis, catecholamines release, liver hypoperfusion, and alterations in pyruvate dehydronegase activity (through mitochondrial dysfunction [73]) can contribute to an elevated lactate concentration [74].Consequently, trying to define a cut-off for hyperlactatemia is difficult, unless the clinical condition and time of assessment are well determined.For instance, our meta-analysis showed that a lactate level greater than 4 mmoL/L was associated with higher mortality (OR, 5.61 [95% CI, 2.27-13.84];I 2 = 76%; n = 1009; p = 0.0002).However, when analyzing lactate as a continuous variable, we found 20 studies where the survivor group had a mean lactate level greater than 4 mmoL/L.Still, an elevated lactate level should always be a warning signal that requires evaluation [1].Jansen TC et al. demonstrated that in adults, with increasing initial lactate levels, survival quickly decreased [75].
Therefore, without a clear neonatal subpopulation, clinical condition, and time of assessment, the lactate level alone is unlikely to assist as a screening test for adverse outcomes in newborns.However, as the neonates with the highest risk of death were those with a higher lactate concentration, lactate levels could be used to stratify those with a higher risk of adverse outcomes.These interpretations are in agreement with pediatric studies.Scott HF et al. [4], found that in children attending emergency departments, hyperlactatemia is associated with mortality, but with low sensitivity (20%).That is, lactate levels alone are not effective as a screening test, but might be used to identify the patients at highest risk.The evidence that lactate is a marker of severity of illness in adults is vast [73].In fact, the SEPSIS-3 consensus requires a persistence of lactate greater than 2 mmoL/L to identify adult patients with sepsis with a greater risk of mortality [76].For this reason, recent studies in adult and pediatric populations have focused on lactate clearance as the predictor of outcome, rather than the isolated lactate level itself [74,[77][78][79].In a recent systematic review evaluating adult patients, Jean-Louis Vincent et al. found that serial lactate measurement could be useful in the evaluation of the response to therapy in critically ill patients and stated that lactate clearance evaluation seems to be valid regardless of the initial value [74].Despite the complexity of the interpretation of lactate level, its decrease is ultimately a good sign [80].
Future studies evaluating lactate levels in neonates need to adjust for potential confounders in lactate metabolism.For example, we did not find any study evaluating a possible interference of vasoactive drugs in lactate metabolism.It is known that the use of exogenous catecholamines induces an increased plasma lactate concentration [81].Moreover, with recently published guidelines using point-of-care ultrasound to assess the hemodynamic state in neonates [82], lactate could be an additional parameter in conjunction with an echocardiogram.
Our systematic review and meta-analysis were conducted through a rigorous search strategy through all of the available literature, including four studies not written in English, with strong statistical analysis, and risk of bias assessment.However, several limitations are worthy of note.First, a meta-analysis of observational studies does not permit conclusions about causality.Second, we found a wide heterogeneity between studies, with varied subpopulations, clinical conditions, and lack of adjustment for covariates.This heterogeneity poses a challenge in determining whether the outcomes and studies are comparable or not.With our results, we suggest that further research evaluating blood lactate levels carefully adjust for potential confounders, including exogenous catecholamines administration.Moreover, research efforts should focus on inspecting serial lactate measurements, rather than a single measurement.Third, as there is no definition of hyperlactatemia in neonates, we found no study where the threshold was predefined in the data accuracy test analysis.Lastly, 13 of the included studies had a fair quality classification through the Newcastle-Ottawa Scale.

Conclusions
Our systematic review and meta-analysis, which included data from 46,069 neonates, suggest that greater lactate levels are associated with a higher risk of mortality and morbidities.Nonetheless, until new studies assess the precise clinical condition and time of assessment, the results from our meta-analysis do not support the use of lactate levels as a screening test to identify adverse outcome in newborns.Research efforts should focus on analyzing serial lactate measurements, rather than a single measurement.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/children10111796/s1, Figure S1: Search Strategy; Figure S2: Meta-analysis of hyperlactatemia (Continuous Exposure) and Acute Kidney Injury/RRT necessity; Figure S3: Meta-analysis of hyperlactatemia (Continuous Exposure) in the first 24 hours of life in neonates with birth asphyxia and Neurological Outcomes;  S1: Characteristics of the DTA included studies in the systematic review; Table S2: Studies excluded from the meta-analysis with reason; Table S3: Assessment of risk of bias (Newcastle-Ottawa Scale).References  are cited in the supplementary materials.
Author Contributions: F.Y.M. conceptualized and designed the study, collected data, carried out the initial analyses, drafted the initial manuscript, and reviewed and revised the manuscript.V.L.J.K. conceptualized and designed the study, designed the data collection instruments, collected data, and reviewed and revised the manuscript W.B.D.C. conceptualized and designed the study, coordinated and supervised data collection, and critically reviewed the manuscript for important intellectual content.All authors have read and agreed to the published version of the manuscript.

Children 2023 ,
10, x FOR PEER REVIEW 10 of 20

Table 1 .
Characteristics of the included studies in the systematic review.