SERUM HUMANIN IN PEDIATRIC SEPTIC SHOCK–ASSOCIATED MULTIPLE-ORGAN DYSFUNCTION SYNDROME

ABSTRACT Background: Multiple-organ dysfunction syndrome disproportionately contributes to pediatric sepsis morbidity. Humanin (HN) is a small peptide encoded by mitochondrial DNA and thought to exert cytoprotective effects in endothelial cells and platelets. We sought to test the association between serum HN (sHN) concentrations and multiple-organ dysfunction syndrome in a prospectively enrolled cohort of pediatric septic shock. Methods: Human MT-RNR2 ELISA was used to determine sHN concentrations on days 1 and 3. The primary outcome was thrombocytopenia-associated multiorgan failure (TAMOF). Secondary outcomes included individual organ dysfunctions on day 7. Associations across pediatric sepsis biomarker (PERSEVERE)–based mortality risk strata and correlation with platelet and markers of endothelial activation were tested. Results: One hundred forty subjects were included in this cohort, of whom 39 had TAMOF. The concentration of sHN was higher on day 1 relative to day 3 and among those with TAMOF phenotype in comparison to those without. However, the association between sHN and TAMOF phenotype was not significant after adjusting for age and illness severity in multivariate models. In secondary analyses, sHN was associated with presence of day 7 sepsis-associated acute kidney injury (P = 0.049). Furthermore, sHN was higher among those with high PERSEVERE-mortality risk strata and correlated with platelet counts and several markers of endothelial activation. Conclusion: Future investigation is necessary to validate the association between sHN and sepsis-associated acute kidney injury among children with septic shock. Furthermore, mechanistic studies that elucidate the role of HN may lead to therapies that promote organ recovery through restoration of mitochondrial homeostasis among those critically ill.


INTRODUCTION
Multiple-organ dysfunction syndrome (MODS) is a leading cause of morbidity and mortality among critically ill children admitted to pediatric intensive care units (PICUs) around the world (1).Several clinical phenotypes of pediatric sepsis MODS have been described including the thrombocytopenia-associated multiorgan failure (TAMOF) spectrum (2,3).Current therapeutic strategies for this phenotype-extracorporeal support with continuous renal replacement therapy and therapeutic plasma exchange-holds potential for iatrogenic injury (4).Thus, there remains a critical need to identify novel disease targets, which may be amenable to therapeutic intervention, and shift care paradigms toward prevention of and recovery from organ failures.
The current understanding of the pathobiology of the TAMOF phenotype is that endothelial activation during sepsis results in the release of ultralarge von Willebrand factor-a major platelet adhesion molecule (5).A decrease in the cleaving activity of a disintegrin and metalloproteinase with a thrombospondin type 1 motif, member 13 (ADAMTS13) that serves to break down ultralarge von Willebrand factor is thought to drive the pathognomonic features of TAMOF including microvascular thromboses, consumption of platelets, and hypoxic tissue injury (6).Although current therapeutic approaches against TAMOF seek to restore ADAMTS13 activity, upstream factors that may contribute to endothelial and platelet dysfunction remain incompletely understood.
Mitochondrial failure in immune cells is well documented in immune cells during sepsis (7,8).Recent data suggest that a failure of mitochondrial homeostasis contributes to abnormal endothelial (9)(10)(11) and platelet responses (12)(13)(14) during health and disease.Thus, it is conceivable that interventions that seek to restore normal mitochondrial function may lead to prevention or recovery from organ failures among patients with TAMOF.
Serum humanin (sHN) is a small mitochondrial derived peptide encoded by the 16s ribosomal RNA MT-RNR2 that modulates intracellular and extracellular signaling (15).It serves to inhibit proapoptotic processes and promote mitochondrial homeostasis.When secreted, humanin (HN) has been shown to interact with several pathways and transcription factors of the metabolic and inflammatory responses, including the AMP-activated protein kinase, nuclear factor κB and the signal transducer and activator of transcription 3 pathways, with net anti-inflammatory and antioxidative effects (16).Recently, HN has been shown to have important roles in stabilizing endothelial (17)(18)(19) and platelet function (20), with potentially important clinical implications (21).
Although the effects of HN analogues have been tested by our group in murine models of sepsis (22) and hemorrhagic shock (23), the association between sHN and human sepsis outcomes remains untested.Given its known protective effects in endothelial cells and platelets, we sought to test the association between sHN and TAMOF phenotype and individual organ dysfunctions in pediatric septic shock.We further sought to test the correlation between sHN and platelet counts and biomarkers of systemic inflammation and endothelial activation in a cohort of critically ill children.

Study design and patient selection
The study protocol was approved by institutional review boards of participating institutions (24)(25)(26).Briefly, patients younger than 18 years were recruited from 14 PICUs across the United States between 2003 and 2019.Inclusion criteria were patients meeting pediatric-specific consensus criteria for septic shock (27).Patients were excluded if they did not receive vasoactive support within 24 h of ICU enrollment.There were no study-related interventions except for blood draws within 24 h of study enrollment (day 1) after meeting the enrollment criteria and 72 h thereafter (day 3).Clinical and laboratory data were available between days 1 and 7.All patients were followed for a total of 28 days from enrollment.Baseline illness severity among patients was determined by Pediatric Risk of Mortality III (PRISM-III) score (28).The primary outcome of interest occurrence of thrombocytopenia-associated multiple-organ failure (TAMOF).TAMOF was defined as new-onset thrombocytopenia (lowest platelet count <100,000/μL between days 1 and 7 after meeting the septic shock criteria) and two or more organ failures without evidence of preexisting conditions resulting in thrombocytopenia.Secondary outcomes were presence of individual organ dysfunctions (cardiovascular, respiratory, kidney, hepatic, hematologic, and neurologic) on day 7 based on modifications to Proulx criteria (29), as previously detailed (24) Of note, sepsis-associated acute kidney injury (SA-AKI) was defined based on serum creatinine (SCr) criteria as per Kidney Disease Improving Global Outcomes stage 2 AKI or higher, which corresponds to a ≥-2-fold increase in SCr relative to baseline (30).Other outcomes tested included complicated course-a composite measure of death within a 28-day study period or >2 or more organ dysfunctions on day 7 of septic shock.We have used this measure extensively to capture short-term morbidity and mortality related to septic shock (26,31).In addition, we measured PICU length of stay and PICU-free days, with the latter derived by subtracting PICU length of stay from a theoretical maximum of 28-day study period.

sHN measurement
Human MT-RNR2 ELISA kit (AVIVA Systems Biology, San Diego, CA; lot number KF1067; range, 28-1,800 pg/mL) was used to measure sHN concentrations in serum from days 1 to 3 collected from patients.

PERSEVERE-based risk stratification
PERSEVERE mortality probability and risk strata were previously determined according to published methods (32).Briefly, IL-8, heat shock protein 70 kDA, C-C chemokine ligand 3, C-C chemokine ligand 4, granzyme B, IL-1α, and matrix metallopeptidase 8 were previously measured in day 1 septic shock serum.Classification and Regression Tree analyses were used to derive a mortality probability risk score (0.000-0.999) using R software (version 4.2.2) (GNU Project, Boston, MA).Patients were subsequently classified as low risk (mortality probability score range, ≤0.019), intermediate risk (mortality probability score range, >0.019 to ≤0.300), or high risk (mortality probability score range, >0.300).

Statistical analyses
Minitab Software (version 21.1.0;Minitab, Pittsburgh, PA) was used for data analyses.GraphPad Prism (version 9; GraphPad Software, Boston, CA) was used

Demographics, clinical characteristics, and outcomes of the cohort
A total of 140 patients were included in the study, of whom 39 patients (27.8%) of the cohort had TAMOF.Table 1 shows comparison of demographic, clinical characteristics, and outcomes between patients with and without TAMOF in the study.Patients with TAMOF clinical phenotype were younger and had greater illness severity at study enrollment based on PRISM-III score relative to those without this phenotype.Unsurprisingly, the median for day 1 and the lowest value of the platelet counts recorded between days 1 and 7 from study enrollment were lower among patients with TAMOF.Patients with TAMOF were more likely to remain on invasive organ support on day 7 of illness.Although there were no differences in 7-or 28-day mortality, patients with TAMOF were more likely to have a complicated course, greater PICU length of stay, and fewer PICU-free days.
Association between sHN concentrations and multiple-organ dysfunctions sHN concentrations were higher among samples collected on day 1 compared with day 3 (Fig. 1A).Platelet counts on day 1 of meeting the septic shock criteria were negatively correlated with sHN concentrations (slope, −0.12 [95% confidence interval {CI}, −0.03 to −0.24]; r = 0.04, P = 0.010) as shown in Figure 1B.Patients with TAMOF phenotype had higher sHN concentrations relative to those without at both sampling time points as shown in  Results of multivariable logistic regression analyses testing the association between log10-transformed sHN concentrations and primary and secondary outcomes of interest are shown in Table 2. Upon adjusting for age and PRISM-III score, sHN concentrations were not associated with increased odds of TAMOF (adjusted odds ratio, 1.28 [95% CI, 0.64-2.56];P = 0.480).However, an increase in sHN concentrations was associated with 2-fold increased odds of day 7 SA-AKI (adjusted odds ratio, 2.16 [95% CI, 1.01-4.66];P = 0.04 9) in the cohort.There were no significant associations with any other organ dysfunctions.

sHN concentrations were higher among patients with high-relative to low-mortality risk based on PERSEVERE biomarkers
The association between sHN concentrations and PERSEVERE biomarker mortality-risk strata is shown in Figure 2. Patients deemed to be at high risk of mortality were noted to have significantly higher sHN concentrations on day 1 compared with those categorized as low-mortality risk (P = 0.021).There were no significant differences when comparing patients with intermediate-versus low-mortality risk and those with high-versus intermediate-mortality risk groups.

DISCUSSION
We report on the association between sHN concentrations and pediatric septic shock-associated MODS.Among critically ill children, serum concentrations were higher early in the course of illness and among patients with TAMOF clinical phenotype.Moreover, sHN demonstrated an inverse correlation with platelet counts.Although we did not identify an independent association between sHN and TAMOF, in secondary analyses, we identified that higher sHN concentrations on day 1 were associated with an increased risk of persistent SA-AKI.Finally, we provide evidence that sHN is higher among patients with a high-mortality risk, based on PERSEVERE biomarkers, indicative of degree of systemic inflammation and weakly correlated with several markers of endothelial activation.
To the best of our knowledge, our study is the first to test the association of sHN in human sepsis.Our results are supported by preclinical studies in murine models of shock.Our group has identified that in murine hemorrhagic shock, circulating levels of HN increase in an AMPα1-independent mechanism and administration of synthetic HN exerted beneficial effects through STAT-3 activation (23).Urban et al. (22) tested whether Colivelin, a synthetic derivative of HN, could ameliorate endothelial dysfunction and improve outcomes among male C57Bl/6 mice subject to  cecal ligation and puncture model of sepsis.Treatment with Colivelin was reported to improve survival in comparison with septic animals receiving antibiotics alone.Moreover, Colivelin attenuated the inflammatory response, endothelial, and glycocalyx injury, and ameliorated glycocalyx density and mitochondrial structure among septic animals relative to shams.Similar findings were reported among male albino rats subject to an model of mesenteric artery ischemia reperfusion injury, wherein animals treated with HN analogue were reported to demonstrate reduced inflammation, apoptosis, and iNOS activity relative to shams (33).To the best of our knowledge, no studies have thus far evaluated whether HN directly influences platelet structure and function in shock and as such requires future mechanistic investigation.
Although we observed that sHN was higher among patients with TAMOF phenotype relative to those without, we could not reject our null hypotheses after adjusting for the confounding influence of age and baseline illness severity upon multivariate testing.
It is likely that, given the relatively few patients with TAMOF phenotype included in our study, we were underpowered to detect differences between groups.In secondary analyses, we identified an association with presence of sHN and persistent SA-AKI in the cohort.Given the fact that SA-AKI is a common and consequential organ dysfunction in pediatric septic shock (34) and specifically among those exhibiting the TAMOF disease spectrum (35), future studies are necessary to validate our observational data and specifically test whether sHN can predict patients at risk of persistent SA-AKI.
Loss of mitochondrial homeostasis is a well-recognized mechanism that contributed to organ dysfunctions in critical illness (36,37).Impaired systemic perfusion coupled with microcirculatory changes result in insufficient tissue oxygen delivery for oxidative phosphorylation of ADP to ATP.At critically low levels, ATP generation is compromised with triggering of cell death pathways and consequent organ damage.More recent research implicates the role of mitochondrial dysfunction within microvascular endothelial cells and platelets (9,10,12), both of which are key determinants of microcirculatory blood flow.Our data demonstrate higher sHN concentrations among patients with high-mortality risk strata based on PERSEVERE biomarkers, reflective of degree of systemic inflammation, and weak but significant correlation between sHN and markers of endothelial activation and platelet counts.Thus, further mechanistic studies are necessary to determine whether recombinant HN or its analogues can restore mitochondrial homeostasis in the microcirculation and thus used as a therapeutic intervention to promote organ recovery among the most critically ill patients.
Our study has several limitations, as follows: 1. Identification of organ dysfunctions was based on modifications to historic consensus criteria with inclusion of Kidney Disease Improving Global Outcomes staging (30) and SCr data alone to determine patients with SA-AKI.The data set did not have the necessary granularity to use the updated PODIUM criteria for pediatric organ dysfunctions (38).Of note, our data set lacked oxygen saturation data, degree of noninvasive ventilator support, ventilator settings including PEEP to measure oxygenation index, and neurological criteria including delirium scores.2. Use of only a clinical definition to identify patients with TAMOF phenotype is a limitation, as no data on ADAMTS13 activity were available among patients.3. Small sample size with relatively low number of patients with TAMOF may have resulted in type II error and an inability to reject our null hypothesis.
4. Given the small sample size, we did not perform multiple comparison testing for secondary outcomes.As such, these analyses were meant to generate pilot data with the intent of guiding future validation and mechanistic studies. 5. Potential residual confounders were not accounted for and may explain between sHN and SA-AKI.6. sHN may be a marker of increased mitochondrial damage-associated molecular patterns, and thus, associations shown may reflect epiphenomena of sepsis without causal role.

CONCLUSION
Although sHN concentrations were higher among patients with TAMOF clinical phenotype relative to those without, this association did not hold after adjusting for age and illness severity.In secondary analyses, sHN was associated with presence of persistent SA-AKI among critically ill children with septic shock necessitating independent validation studies.sHN was increased across mortality risk strata reflective of systemic inflammation and correlated with platelet counts and markers of endothelial activation.Further mechanistic studies are necessary to elucidate the role of HN in the pathobiology of organ dysfunctions, as they may lead to the development of therapies that promote organ recovery through restoration of mitochondrial homeostasis.

FIG. 1 .
FIG. 1. A, Box and whisker plots comparing serum humanin (sHN) concentrations in day 1 and 3 serum samples.B, Correlation between day 1 sHN concentrations and day 1 platelet counts.C and D, Comparison of day 1 and 3 sHN concentrations among patients with and without thrombocytopenia-associated multiorgan failure clinical phenotype (TAMOF) in the cohort.

Figure 1 ,
Figure 1, C and D. Changes in sHN concentrations between day 1 and 3 time points were not associated with any outcome of interest (data not shown).Results of multivariable logistic regression analyses testing the association between log10-transformed sHN concentrations and primary and secondary outcomes of interest are shown in Table2.Upon adjusting for age and PRISM-III score, sHN concentrations were not associated with increased odds of TAMOF (adjusted odds ratio, 1.28 [95% CI, 0.64-2.56];P = 0.480).However, an increase in sHN concentrations was associated with 2-fold increased odds of day 7 SA-AKI (adjusted odds ratio, 2.16 [95% CI, 1.01-4.66];P = 0.04 9) in the cohort.There were no significant associations with any other organ dysfunctions.

FIG. 2 .
FIG. 2. Box and whisker plots comparing serum humanin (sHN) concentrations across pediatric sepsis biomarker based (PERSEVERE) low, intermediate, and high mortality risk strata.P values are shown for nonparametric Kruskal-Wallis test with adjustment for multiple comparison testing.There were no statistically significant differences in sHN concentrations between low and intermediate, or intermediate and high strata.However, the difference in sHN concentration when comparing low-versus high-risk strata was significant ( P = 0.02), with the highest concentrations noted among the latter group.

TABLE 1 .
Demographic, clinical characteristics, and outcomes among patients with and without TAMOF phenotype in pediatric septic shock generate figures.Demographic data were summarized with percentages or median with interquartile ranges.Differences between groups were determined by χ 2 test for categorical variables, Kruskal-Wallis test for continuous variables, and Wilcoxon test for differences in sHN concentrations between paired groups (days 1 and 3).We also tested whether changes in sHN between the two time points were associated with outcomes of interest.Multivariable logistic regression analyses adjusted for age and PRISM-III score between log10-transformed day 1 sHN concentrations and outcomes of interest were determined.One-way ANOVA with the Dunnett test for multiple comparison testing was used to compare differences across PERSEVERE mortality-risk strata.Spearman correlation was used to test the association between sHN, platelet counts, and markers of endothelial activation.A two-tailed P value <0.05 was used to test significance.

TABLE 2 .
Multivariable logistic regression analyses to test the association between log-transformed sHN concentrations and primary and secondary outcomes of interest All models were adjusted for age and PRISM-III score.Adjusted OR reported for sHN concentration in pg/mL transformed to log10 scale.CI indicates confidence interval; CVS, cardiovascular; OR, odds ratio; PRISM-III, Pediatric Risk of Mortality III; sHN, serum humanin; TAMOF, thrombocytopenia-associated multiorgan failure.