Young Age as a Risk Factor for Impaired Cerebral Autoregulation after Moderate to Severe Pediatric Traumatic Brain Injury

After institutional approval from the University of Washington (Seattle, Washington) Human Subjects Review Committee, we performed a retrospective analysis of prospectively collected data from an ongoing larger observational cohort study examining cerebral autoregulation after pediatric TBI at Harborview Medical Center (level 1 pediatric trauma center) over a 5-yr period (May 2002 through June 2007). Consent was obtained from parents or legal guardians.

Eligibility criteria for the prospective study included age 16 yr or younger, admission to the Harborview Medical Center pediatric intensive care unit with a diagnosis of moderate to severe TBI (admission Glasgow Coma Scale [GCS] score <13), TBI on computed tomography (i.e. , subdural hematoma) scan, and tracheal intubation. Children with extracranial injuries were included. We excluded children with isolated focal TBI such as epidural or subdural hematoma, hemodynamic instability (per treating intensivist if significant hypotension or hypertension was present immediately before testing), or patients with no available parent or guardian at the time of enrollment. We reviewed medical records for eligibility, relevant medical history, and physiologic data. Study participants underwent cerebral autoregulation testing if they were considered hemodynamically stable by the treating pediatric intensivist. Cerebral autoregulation was examined at the patient's bedside in the pediatric intensive care unit. The research nurse, who was trained in GCS scoring and not involved in determining middle cerebral artery flow velocity (V^mca), autoregulatory index (ARI), or 12-month Glasgow outcome score (GOS), determined GCS score at the time of cerebral autoregulation testing. Temperature was measured via  the tympanic route in all patients by the treating bedside nurse as per pediatric intensive care unit protocol.

Transcranial Doppler ultrasonography (Multidop X; DWL Corp., Sipplingen, Germany) was used to measure bilateral flow velocities in the middle cerebral artery using a 2-MHz ultrasound probe. Previously described age-appropriate depths were referenced and used to insonate the middle cerebral artery.11One registered vascular technologist, with more than 10 yr of experience with transcranial Doppler, insonated the middle cerebral arteries during cerebral autoregulation testing (performed by Y.U.). At the time of testing, both of these parties were blinded to TBI severity. Cerebral autoregulation was calculated off-line and entered into our database by S.S.F. or M.S.V.

Study participants underwent static autoregulation testing after TBI as previously described.12Briefly, during steady state (technically satisfactory conditions where change or lack of change in V^mcais attributed to increase in CPP; usually within 10 s), intravenous phenylephrine was titrated using a slow infusion (0.05–0.1 μg · kg⁻¹· min⁻¹) over a 3- to 5-min period. CPP was increased according to whichever following variable was greater: (1) 20% above baseline; or (2) a set value of 80 mmHg for the group younger than 9 yr and 90 mmHg for the group aged 9–16 yr, respectively. CPP and V^mcawere simultaneously and continuously measured and were recorded in the computer for subsequent off-line analysis.

Autoregulatory capacity was quantified with the ARI, which was calculated according to a previously published formula.12Mathematically, the ARI is the percent change in estimated cerebrovascular resistance per percent change in CPP:

ARI =% ΔEstimated Cerebrovascular Resistance/% ΔCPP.

The estimated cerebrovascular resistance is the ratio of CPP to V^mca. Therefore, an ARI of 0 represents absent autoregulation (pressure-dependent V^mca), whereas an ARI of 1.0 represents perfect autoregulation. To dichotomize the results for statistical analysis and in accordance with the previous definition of intact cerebral autoregulation, autoregulatory capacity was considered intact if the ARI was 0.4 or greater.12 

Glasgow outcome score was determined at 12 months after TBI using either telephone or written questionnaires or in-person evaluations by a research nurse blinded to clinical data determined outcome. The Jennett five-point GOS classification was used13: GOS 1 = dead; GOS 2 = vegetative state, GOS 3 = alive but functionally impaired, GOS 4 = minimal handicap, and GOS 5 = premorbid level of functioning. A GOS less than 4 reflected poor outcome, whereas good outcome was defined as a GOS of 4 or 5.13 

Patients who underwent cerebral autoregulation testing during the first 72 h after TBI were considered for this analysis. SPSS version 11.5 (SPSS Inc., Chicago, IL) was used for data entry and analysis. Descriptive statistics were used to describe patient characteristics, clinical data, differences in cerebral autoregulation, intracranial pressure (ICP), and factors associated with long-term outcome.

Impaired cerebral autoregulation (main outcome) was defined using both (1) average ARI of both cerebral hemispheres (mean ARI [m  ARI]) and (2) lowest ARI (l  ARI) of either cerebral hemisphere. Patients were categorized as having either impaired (ARI <0.4) or intact (ARI ≥0.4) autoregulation by each definition. Based on our previous work showing hemispheric differences in cerebral autoregulation, and because final outcome may be determined by severity of TBI and worst autoregulation, an l  ARI less than 0.4 defined impaired cerebral autoregulation.

The Student t  test, chi-square test, or Fisher exact test was used to analyze differences in baseline characteristics, physiologic conditions, and autoregulation data between (1) young (aged <4 yr) and older (aged ≥4 yr)x children, (2) impaired (l  ARI <0.4) versus  intact (l  ARI ≥0.4) cerebral autoregulation, and (3) good (GOS ≥4) versus  poor (GOS <4) outcome. The relation between impaired cerebral autoregulation and (1) sedation type, (2) ICP greater than 20 mmHg, and (3) V^mcagreater than 2 SDs was determined by the Fisher exact test. Patients were also divided into two groups, (1) poor outcome (12-month GOS <4) and (2) good outcome (12-month GOS ≥4), to examine the relation between age, cerebral autoregulation, and outcome. These data are presented as mean ± SD or n (%). P < 0.05 reflects significance. Linear relations between l  ARI and age as well as l  ARI and GCS at time of autoregulation testing were examined using the Spearman rank correlation and are reported using R  ²and P  values.

To determine potential factors independently associated with impaired cerebral autoregulation, variables with significance or variables with trends toward significance (0.05 ≤P < 0.10) for impaired l  ARI were entered into a binary logistic regression model for multivariate analysis. Adjusted odds ratios and 95% confidence intervals for the relation between variables of interest and impaired cerebral autoregulation (l  ARI) were determined.

During the 5-yr study period, 140 children were eligible for enrollment in the larger ongoing National Institutes of Health–funded study examining cerebral autoregulation and quality of life in children with TBI. We did not capture 42 (38 patients were extubated before consent) study participants, leaving 98 families available for approach. Sixteen guardians did not consent. Forty-five patients were then excluded from this analysis based on the following criteria: brain death (n = 1) or death (n = 6) before autoregulation testing, autoregulation testing greater than 72 h after injury date (n = 20), language barriers (n = 5), hemodynamic instability (as determined by the treating intensivist; n = 6), no available guardian (n = 6), and mild TBI (n = 1). Consequently, data from 37 children who underwent cerebral autoregulation testing within the first 72 h after moderate to severe TBI were examined. All 37 patients completed the protocol without dropout or withdrawal. Data from 12 children have previously been published in a study designed to answer a different question.4 

Children were 8.9 ± 5.1 (0.8–16) yr old (table 1). Ten children (27%) were young (aged <4 yr; mean, 2.0 yr; median, 2.2 yr; range, 0–3 yr), and 27 children (73%) were older (aged ≥4 yr; mean, 11.4 yr; median, 12.0 yr; range, 4–16 yr). The number of boys exceeded the number of girls in both age groups. Two children had inflicted trauma. All patients received a head computed tomography scan in the emergency department. All children had moderate to severe TBI (GCS 3–12) at the time of admission to the emergency department. Intracranial pressure data were available in 30 patients (81%) at the time of cerebral autoregulation testing. There were no demographic or morphometric differences between patients with and without ICP monitoring or between the patients included (n = 37) and excluded (n = 45) in this study (table 1).

All patients underwent cerebral autoregulation testing within 72 h (15 ± 17 h; range, 0–72 h) after pediatric intensive care unit admission. Three patients (30%) in the young group and 6 patients (22%) in the older group underwent neurosurgery during the first 72 h after TBI (Fisher P = 0.6), but none of the children had autoregulation testing within 6 h of receiving general anesthesia. The distribution of mannitol (M) and/or 3% hypertonic saline (HS) either before or at the time of autoregulation testing was 9/10 (7M and 2M/HS; P = 0.6) in the young group versus  24/27 (15M and 9M/HS; P = 1.0) in the older group. There were no adverse outcomes due to autoregulation testing.

Overall, children had intact cerebral autoregulation (table 2). However, children younger than 4 yr had lower autoregulatory indices than older children (table 2). Children younger than 4 yr had a higher incidence of impaired cerebral autoregulation than older children (8 of 10 vs.  7 of 27; P = 0.006). There was no linear relation between age and l  ARI. However, higher GCS at the time of autoregulation testing was associated with higher l  ARI (fig. 1, actual data shown; r  ²= 0.49 based on rank, P = 0.002).

There was no difference in admission GCS, arterial carbon dioxide tension (Paco²), hematocrit, temperature, ICP, hyperemia, or mean arterial pressure between young and older children during the time of cerebral autoregulation testing (table 2). There was no difference in increase in CPP between young and older children (R  ²= 0.2). Sedation regimens were determined by the treating clinicians, and typical practice is to use benzodiazepine sedation followed by propofol for brief periods for refractory agitation and/or ICP. At the time of cerebral autoregulation testing, 29 children received sedation with a combination of either (1) an intravenous benzodiazepine (midazolam or lorazepam) and fentanyl (n = 19, 66%) or (2) an intravenous propofol (n = 10, 34%). There was no association between type of sedation and impaired cerebral autoregulation. Six (16%) of 37 patients had fever (temperature ≥38.5°C) during the time of autoregulation testing, and 4 patients (67%) with fever had impaired cerebral autoregulation (l  ARI 0.32 ± 0.38; range, 0–0.89). Absolute hyperemia, defined as V^mcagreater than 2 SDs for age and sex,11,14,15was present in 7 of 37 patients (19%), and there was no association between hyperemia and impaired (n = 2/15; 13%) cerebral autoregulation (P = 0.68; table 3). In each case where ICP monitoring was discontinued before cerebral autoregulation testing (7 of 37 patients), the ICP was less than 10 mmHg, so the mean arterial pressure was used as the CPP.

Univariate factors associated with impaired cerebral autoregulation (l  ARI <0.4) included age less than 4 yr (P = 0.01) and lower GCS score at time of autoregulation testing (P = 0.03; table 3). Factors independently associated with impaired cerebral autoregulation were age less than 4 yr (adjusted odds ratio, 12.2; 95% confidence interval, 1.5–98.5) and lower GCS at the time of autoregulation testing (adjusted odds ratio for higher GCS, 0.53; 95% confidence interval, 0.30–0.96; table 4). Age less than 4 yr predicted impaired cerebral autoregulation with a sensitivity of 53% and a specificity of 91%. GCS less than 9 predicted impaired cerebral autoregulation with a sensitivity of 100% and a specificity of 18%. Intracranial hypertension (ICP >20 mmHg) or hyperemia was not associated with impaired cerebral autoregulation.

Glasgow outcome scores were available in 30 patients (84%). The overall 12-month GOS score for these patients was 3.9 ± 1.1 (range, 1–5; table 1). Eight children (27%) had a poor outcome (table 5). Children younger than 4 yr had a worse outcome than older children (GOS 3.0 ± 1.3 vs.  4.1 ± 0.9; P = 0.02). Poor outcome was associated with lower autoregulatory indices and impaired cerebral autoregulation (table 5). There was no difference in demographic characteristics between the 30 patients with cerebral autoregulation and GOS data and the 7 patients for whom we did not have outcome data.

The main findings of this study are that age less than 4 yr and low GCS score at the time of autoregulation testing were independently associated with impaired cerebral autoregulation. We also found that children younger than 4 yr and children with impaired cerebral autoregulation had worse 12-month GOS. These findings are the first to describe age as an independent risk factor for impaired cerebral autoregulation in children early after TBI.

Despite the relatively small number of patients in this study, the current analysis show that children younger than 4 yr had lower autoregulatory indices, a higher prevalence of impaired cerebral autoregulation, and worse long-term outcome (table 2). We defined young children as children younger than 4 yr to examine the relation between age and impaired cerebral autoregulation because children aged 0–4 yr have the highest rate for TBI-related deaths, hospitalizations, and emergency department visits compared with any other age group in the United States.7Furthermore, we did not observe a linear relation between age and impaired cerebral autoregulation. We speculate that the higher incidence of impaired cerebral autoregulation in young children at least partially explains the relation between young age and poor outcome, thereby making young age an important confounder in the causal pathway between cerebral autoregulation and outcome (fig. 2). Although this study does not provide a mechanistic explanation for the relation between either young age and impaired cerebral autoregulation, or impaired cerebral autoregulation and poor outcome, these observations provide new neuroepidemiologic data of some cerebrovascular changes after pediatric TBI and may provide the rationale for further examining age-related differences in cerebral autoregulatory mechanisms and response after pediatric TBI.

Fluid percussion injury in animals is thought to mimic pediatric TBI,16and studies of fluid percussion injury in piglets indicate age-related differences in both the incidence of impaired autoregulation and mechanisms regulating autoregulatory capacity. When compared with older juvenile piglets with fluid percussion injury, newborn piglets with fluid percussion injury have a higher incidence of impaired autoregulation and possible unique biologic autoregulatory mechanisms.17,18Cell signaling pathways, endogenous peptides, and membrane potentials have been examined as mechanisms of vascular control of cerebral autoregulatory capacity.17–19Increased endothelin-1 levels18; blunted K⁺channel activation17; and decreased levels of nitric oxide, cyclic guanosine monophosphate, cyclic adenosine monophosphate, and prostanoids19,20have been reported in newborn versus  juvenile piglets after fluid percussion injury when subjected to hemorrhagic hypotension as the autoregulatory stimulus. Potential mechanisms that might contribute to age-related differences in autoregulatory outcome status include age-related up-regulation of endothelin 1, which impairs K⁺channel function, an important mechanism for vasodilation.17,18Age differences in autoregulatory mechanisms likely exist similarly in humans and thus merit further study.

Cerebral autoregulatory status may be an important marker of or, more importantly, a contributor to injury severity in children with TBI. Although previous studies by Sharples et al.  9and Vavilala et al.  6described a high incidence of impaired cerebral autoregulation in children with moderate and severe TBI, admission GCS was used to describe the cohort, and the incidence of impairment was not examined in relation to GCS at the time of testing. Although age was also an independent predictor of impaired cerebral autoregulation at the time of testing, unlike age, GCS and cerebral autoregulation may improve or deteriorate early after TBI.21This fact and the fact that patients with severe TBI have more impairment than patients with mild TBI6suggest that point-of-care evaluation of GCS may serve as a screening tool for impaired cerebral autoregulation when formal autoregulation testing is not feasible. However, a larger study is needed to verify these preliminary findings.

A small case series suggests that children with inflicted TBI may be at greater risk for impaired cerebral autoregulation and poor outcome than children with TBI from other causes, but the mechanisms explaining this association are not well defined and the number of children with inflicted TBI was exceedingly small.22Cerebrospinal fluid and serum markers of neuronal damage such as S100B and neuronal specific enolase are found differentially and in higher concentrations in inflicted TBI compared with noninflicted TBI, suggesting either a more severe form or a different type of TBI23and potentially a different mechanism involved in impaired cerebral autoregulation after inflicted TBI.

Certain physiologic factors may also affect cerebral autoregulatory capacity after TBI (fig. 2). Paco²is the most potent cerebral vasodilator, increasing CBF linearly by 2–4% per mmHg Paco²within a 25- to 75-mmHg range.24However, carbon dioxide reactivity can be temporarily impaired after TBI,25and carbon dioxide reactivity changes of less than 2% may be associated with poor outcome after TBI.26In our study, there was no association between Paco²and cerebral autoregulation, but the number of children with either hypocapnia or hypercapnia was very small (table 3). Hyperthermia may also be an independent predictor of injury severity27and may be associated with poor outcome after pediatric TBI.28Although we found no difference in temperature between children with and without impaired cerebral autoregulation, few children had fever at the time of testing (table 3).

Severe pediatric TBI can lead to increased CBF29and hyperemia in the presence of impaired cerebral autoregulation.6,8In 1996, Biagas et al.  30showed that posttraumatic hyperemia may be an age-related trend in immature (3.5- to 4.5-week-old) rats and suggested variations in vasoreactivity or metabolism as a possible mechanism. In our study, young children did not have a higher incidence of cerebral hyperemia than children aged 4 yr and older (table 2). Although we compared our V^mcadata to age- and sex-specific reference values,11,14our definition of cerebral hyperemia was not ideal because we did not have a measure of cerebral metabolic rate. While adult and pediatric studies document an association between hyperemia and loss of cerebral autoregulation,6,31we found no such association. Low cerebrovascular resistance results in vasodilatation and higher CBF, and limits cerebral vasoconstriction during hypertension. Overall, the number of patients with cerebral hyperemia in our cohort was small, and a true association between hyperemia and loss of autoregulation cannot be determined from this data set.

The primary limitation of this study is the small sample size, reducing our ability to better examine the relation between age and cerebral autoregulation. Transcranial Doppler ultrasonography measures V^mca, not CBF. However, changes in CBF velocity are proportional to changes in CBF,32and transcranial Doppler is a noninvasive bedside method for estimation of CBF in children. Furthermore, transcranial Doppler is commonly used to measure cerebral blood flow velocity and autoregulatory capacity.6,33Many of our patients received sedation, which may (in particular propofol) decrease CBF and improve autoregulation.34However, we found no relation between autoregulation and type of sedation. In the current study, there were only two patients with inflicted TBI, and we cannot comment on the relative contribution of age versus  inflicted TBI on impaired cerebral autoregulation. Given the small sample size and the observational (and not randomized) nature of this study, we have not altered our treatment strategy. Finally, mechanistic explanations of the observed age-related differences are lacking. Despite these limitations, these data suggest a need to further examine the age-related mechanisms involved in cerebral autoregulation in children with TBI. Perhaps the most important consequence of this study is that we now have an increased awareness that young children may have an increased risk of impaired cerebral autoregulation; thereby prompting autoregulation testing in these younger children, early after TBI.

In summary, our data show that children younger than 4 yr have an increased risk of impaired cerebral autoregulation, independent of GCS at the time of autoregulation testing. Developmental age-related factors may play a role in the mechanisms maintaining or worsening this homeostatic process in children with TBI. Understanding alterations in these mechanisms may be important to improving the cerebrovascular milieu and, ultimately, the outcome of young children, who are at most risk of poor outcome after TBI.

The authors thank the children and families at Harborview Medical Center who participated in this study and Domonique Calhoun, M.F.A. (Secretary Senior, Department of Anesthesiology, Harborview Medical Center, Seattle, Washington), for her assistance in the preparation of the manuscript.