Cerebrovascular function and cognition in childhood: a systematic review of transcranial doppler studies

Primary and secondary screens of the literature were performed. A primary screen, including PubMed, Embase, and PsycINFO searches, was conducted for peer-reviewed papers published between January 1970 and 31 May 2013. The following search terms were employed in the primary screen: (ultrasonography doppler transcranial OR TCD OR transcranial doppler*) AND (baby OR newborn* OR infant* OR preschool* OR teenager* OR child* OR adoles*). A total of 2778 articles were retrieved.

Titles and abstracts were read by at least two of the authors (MJB, JH or HADK). Articles were retained if they collected TCD data at rest or during a cognitive task, from any cerebral artery accessible via TCD (e.g. anterior, middle, posterior or basilar) and where the mean age of participants was less than or equal to 18 years of age. Bibliographies from the identified articles were also manually searched for relevant publications. Articles were excluded if they were not written in English, were a case-report, or did not look at the statistical association between a TCD measure and a measure of cognitive performance. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed [10].

If the article was included, the following data was extracted: participant characteristics (i.e. number, age range or mean if unavailable, male/female ratio, clinical diagnosis if applicable), TCD protocol and vessel, cognitive measure, and relevant findings (linking TCD and cognition results).

There were thirteen studies investigating children with Sickle cell disease, with most reporting an association between high blood flow velocities and impaired cognition, from infanthood to adolescence. Hogan et al. [19] reported that in infants (assessed at nine months) with Sickle cell disease, high blood flow velocities (both systolic velocity and mean velocity in the BA, MCA and ICA) were associated with moderate to high risk of neurodevelopmental delay. Armstrong et al. [30] reported that lower scores in the communications, daily living and socialisation domains of the Vineland Adaptive Behaviour Scales were associated with higher systolic velocities, even for velocities within the normal range in infants with Sickle cell disease. Similarly, Schatz et al. [25] reported that preschool-aged children with Sickle cell disease and delayed development had significantly higher MCA mean velocities, compared to healthy controls. Sanchez et al. [24] reported a negative association between MCA systolic velocities and syntactical ability in children with Sickle cell disease. Furthermore, Ruffieux et al. [23] reported an association between abnormally high velocities (maximum velocity >200 cm/sec or peak systolic >250 cm/sec) and poor memory performance in children with Sickle cell disease.

Kral et al. [22] found that children and adolescents with Sickle cell disease and abnormal systolic velocities (defined as maximum velocity >200 cm/sec in the MCA and ICA) had lower verbal IQ scores than those with conditional/mid-range systolic velocity (defined as maximum velocity 170-200 cm/sec). Children displaying these mid-range systolic velocities performed worse than children with normal systolic velocities (defined as maximum velocity <170 cm/sec) on measures of sustained attention/concentration and executive function. The conditional systolic velocities group were however the slowest to complete the Trail Making Test. Employing the same sample, Kral and Brown [20] reported that these laboratory measures of cognitive function did not translate to problems with everyday measures of cognition (i.e. academic attainment and grade retention). However, children with abnormally high TCD measures received significantly more special education services [20].

With respect to sample size, the largest study (n = 143) included in this review was performed by Bernaudin and colleagues [17] and involved children with Sickle cell disease and non-affected siblings. Children with Sickle cell disease and abnormal TCD (defined as maximum velocity >200 cm/sec in the BA, MCA, ACA, PCA and ICA) had lower cognitive performance than children with Sickle cell disease and normal TCD for Picture Arrangement and Performance IQ (on the WISC-II/WPPSI-R). However, no significant difference was observed after the exclusion of participants with a history of stroke (n = 5).

In the only study to report a positive association between blood flow velocity and cognition in a Sickle cell disease sample, Kral et al. [21] reported that after controlling for age and hematocrit, children with Sickle cell disease and high and mid-range systolic velocities had better verbal memory (Children’s Memory Scale Stories) compared with children with Sickle cell disease and normal TCD velocities. There were also reports of a lack of association between blood flow velocities and cognition in Sickle cell disease child samples [16, 18, 26, 31]. Onofri et al. [31] found no relationship between IQ (performance and verbal) and systolic velocities in a sample of children with Sickle cell disease. A study by Aygun et al. [16] observed no association between scores on an academic screening test and mean velocities (from multiple vessels). Strouse and colleagues [26], reported no correlation between ACA systolic velocities and IQ in children with Sickle cell disease. Similarly, Hijmans et al. [18] reported no association between mean velocity (MCA, ACA, and ICA) and cognitive measures (including IQ, response inhibition, sustained attention, planning and visuospatial working memory) in children with severe Sickle cell disease. However, children with right greater than left asymmetries in resting mean velocities had better sustained attention [18].

Alatas and colleagues [15] reported that in babies from high-risk pregnancies there was a positive relationship between in-utero MCA pulsatility index and five-minute Apgar scores after delivery, where low pulsatility index was associated with poor physical and neurological health. There was no association in babies from mothers with normal pregnancies.

Three studies were conducted in premature infants [11, 12, 14]. Scherjon et al. [14] reported that in preterm infants without intracranial haemorrhage, there was a relationship between MCA mean velocity ratio (defined as mean velocity in the first 12 hours after birth, divided by mean velocity in the period from 12 hours until 168 hours after birth) and Prechtl score at 40 weeks corrected gestational age; where if MCA mean velocity did not increase after birth, the child was more likely to have a suspected neurological abnormality. They reported no relationships between mean velocity ratio and neurological functioning at six and 12 months [14]. Arditi and colleagues [11] reported that in a group of infants born prematurely with TCD data collected at corrected 37 weeks gestational age: (1) higher right MCA systolic velocities were related to poorer neonatal perceptual performance (on habituation and orientation); (2) higher left MCA systolic velocities were associated with a better Mental Development Index score at 24 months; and (3), those with higher left MCA systolic velocities (as compared to right MCA systolic velocities) had better neonatal orientation and habituation scores as neonates and MDI scores at 24 months. In another study of preterm infants, Ojala and colleagues [12] compared those being ventilated for respiratory distress syndrome (V-RDS) with preterm infants with no signs and not ventilated (N-RDS) using the ACA resistance index. They reported that during the early postnatal transition (the first day of life measured at six, 12 and 24 hours), a lower resistance index was associated with a higher risk of adverse neurologic outcome in the group of N-RDS preterm infants at 12 months of age; this was not the case for the V-RDS infants.

Rennie et al. [13] found that developmentally delayed infants (defined as a score of <70 on the Bayley Scale of Infant Development at age 18 months) did not show the usual steady increase in ACA mean velocity over the first three days of life, as compared to typically developing infants. A rise then a fall in mean velocity occurred in four of nine of the developmentally delayed group (44%), whereas this pattern occurred in a smaller percentage (four of 31 participants, 13%) of the group who were subsequently normal.

Bruneau et al. [27] compared children with autistic behaviour, mentally retarded children without autistic symptoms and typically developing controls. In the healthy control group they found that auditory stimulation resulted in increased left MCA mean velocity and decreased left MCA resistance index. This pattern was also observed in the mentally retarded without autistic symptoms group, although the degree of lateralisation was reduced. However, in the autistic group, children displayed a symmetrical decrease of MCA mean velocities and an increase in the resistance index.

Hill and colleagues [7] reported that MCA systolic velocity was significantly higher, processing speed slower, and visual attention poorer, in a group of children with mild sleep disordered breathing as compared to controls. However, these cognitive and TCD measures did not significantly correlate.

Five studies assessed healthy children in order to investigate relationships between fTCD and cognition using visual stimulation [2, 3], passive and expressive language tasks [4, 29], and (non-) lexical word generation paradigm [28]. A recent study by Groen et al. [3] reported left-lateralized activation for language production and right-lateralized activation for visuospatial memory in the majority of children (58%). In this study, boys showed a trend for stronger right-hemisphere lateralization for visuospatial memory than girls, but there was no gender effect on language laterality. Children with left-lateralized language production achieved higher standard scores on vocabulary and non-word reading, regardless of the laterality of visuospatial memory. Having language and visuospatial functions in the same hemisphere was not associated with poor cognitive performance [3]. In a previous study, Groen et al. [2] reported a right hemispheric specialization for a visuospatial task in children, however the degree of lateralisation was unrelated to cognitive performance.

Using a sample of right-handed children, Stroobant et al. [32] reported that both receptive and expressive language tasks elicited left hemispheric lateralization which was more pronounced in the expressive task. However, lateralization did not relate to language performance or hand dominance. Similarly, Lohmann et al. [4] reported that a language task induced a left-lateralised velocity increase in children between two and nine years, however this lateralisation was unrelated to language performance. Haag et al. [28] reported that in right-handed children and adolescents, the picture description paradigm failed to indicate left hemispheric dominance in a substantial proportion of the sample regardless of age. However, although Haag et al. [28] measured the effect of a cognitive process on blood flow velocity, they did not correlate changes with a behavioural cognitive measure.

Resting cerebral blood flow velocities such as systolic, diastolic and mean were commonly assessed in relation to cognitive performance [7, 11‐14, 16‐26, 30]. The majority of studies reported an association between blood flow velocity and cognition, with the extremes of systolic and mean blood flow velocities (i.e., very low or very high) being associated with poorer cognitive or neurological functioning. However, six studies reported no associations, including four Sickle cell disease studies [16, 18, 26, 31], one study on preterm infants [12] and one on children with sleep disordered breathing [7]. The interpretation of these studies reporting no association is limited by the lack of information about potential confounders (e.g., medication doses and age), which are required to adequately examine the outcome of cognitive performance in children by means of TCD.

Much of the research reviewed here related to children with Sickle cell disease. Higher systolic and mean velocities were associated with lower IQ or poorer cognitive performance in seven samples of children with Sickle cell disease [17, 19, 22‐25, 30], although this effect did not hold in one sample when accounting for a history of stroke [17]. The reported cognitive deficits did not translate into academic achievement impairments, but those with higher systolic velocities did receive more special education services [20]. Conflicting other studies, Kral et al. [21] reported that after controlling for age and medication, children with Sickle cell disease and higher systolic velocities displayed better verbal memory performance than those with lower velocities. Four Sickle cell disease studies reported no association between cognitive performance and resting blood flow velocities [16, 18, 26, 31]. There have been mixed findings from other brain imaging techniques in Sickle cell samples, for example poor cognitive function has been associated with decreased regional cerebral blood flow as measured using PET [33] and increased cerebral blood flow using fMRI [26].

Clinical studies assessing premature and low birth weight infants showed consistent associations between low blood flow velocities and poor cognitive outcomes. In a study tracking low birth weight infants, Rennie et al. [13] showed how over the first three days of life systolic velocity increased, and that infants with developmental delay at 18 months of age were less likely to show this characteristic increase. In a similar study of premature infants, Scherjon et al. [14] reported that a larger increase in mean velocity in the first days of life related to better neurological functioning at 40 weeks corrected gestational age. Likewise, another study found higher systolic velocities measured at birth were related to lower neonatal habituation performance and neurological functioning at two years [11].

There was some suggestion that asymmetries (i.e., differences between the hemispheres) in resting blood flow velocities were associated with cognitive performance. Greater left than right systolic velocity was related to better neonatal habitation performance and neurological functioning at two years [11]. Further, children with Sickle cell disease had better sustained attention when right mean velocities were greater than the left [18].

The study by Bruneau et al. [27], also mentioned above, reported that blood flow velocities (mean, systolic and diastolic) increased during auditory stimulation in healthy and mentally retarded children, however velocities decreased in a group with autistic behaviours. They suggested that instead of downstream vasodilation in response to cognitive demands, those with autistic behaviours may vasoconstrict [27].

All other fTCD studies were in healthy samples [2‐4, 28, 32]. In all studies a lateralised response during a cognitive operation was observed, with exception of one task (picture description, which was one of two tasks employed, the other being word generation) used by Haag et al. [28]. Stroobant et al. [32] noted that the expressive language task generated a greater lateralisation than the receptive language task. The lateralisation index (i.e., the difference in the degree of blood flow velocity increase between the hemispheres relative to a cognitive operation; see Figure 2) was related to cognitive performance in only one study [3], but not in four studies [2, 4, 28, 32].

Measures of vessel resistance such as pulsatility and resistance indexes have shown to relate to cognitive impairment in aged and demented samples [8]. These measures were only assessed in three childhood studies [12, 15, 27]. Low pulsatility index recorded in-utero was related to lower physical and neurological functioning five-minutes post-birth in infants from high risk pregnancies, but not in controls [15]. Similarly, low resistance index taken in the first day of life was related to neurological impairments at 12 months in a group of preterm infants who required ventilation after birth. However this effect was not found in preterm infants not requiring ventilation. It therefore appears that in preterm infants with compromised physical reserve, low cerebrovascular resistance puts them at risk for neurological impairment (at least temporarily).

Bruneau et al. [27] was the only study to assess changes in cerebrovascular resistance in relation to cognitive stimulation. They reported that the resistance index decreased on the left side during auditory stimulation in typically developing children, to a lesser extent in mentally retarded children, but increased in children with autistic behaviours [27].

One limitation of the TCD method is the assumption that artery diameter remains constant, therefore any change in velocity represents a change in flow. It has been reported however, that the diameter of the MCA does not significantly change during moderate alterations in blood pressure [34, 35] and therefore any change in velocity reflects change in blood volume through the artery. Notably, and similar to other psychophysiological measurements, the evidence base for the reliability of TCD requires more work. In a recent paper, McDonnell et al. [36] reported that the test-retest reliability (three testing sessions over a few weeks) of the TCD-measured cerebrovascular response (CVR; change in blood flow velocity relative to inhalation of gas with 95% O₂ and 5% CO₂) was strong, particularly for measurements taken in sitting as opposed to supine. Sanchez et al. [24] reported elevated systolic velocities apparent in a group of five children with Sickle cell disease remained elevated over a six month period. These studies provide evidence for the reliability of TCD measures however further work is required, particularly in relation to fTCD measures.

Many studies did not detail the manufacturer of the TCD set-up employed. Specifications vary greatly between different manufacturers, for example sampling rates range from 1 Hz to 100 Hz. It is unknown if TCD measurements produced are equivalent between commercially available set-ups. Future papers must note the TCD manufacturer and associated recording parameters. Further, simple experimental design elements were often missing in identified studies, such as the time period used to collect resting TCD measures (e.g. 30 seconds, 1 minute or 2 minutes). A concerted effort to include these experimental design factors is required.

A further limitation of this review, and of all reviews of retrospectively published data, is that findings may be biased, with significant associations more likely to be published than non-significant associations [37]. Most studies investigated associations between cognition and TCD measures in Sickle cell disease samples, which limits extrapolation to other clinical as well as healthy populations. Future research therefore needs to focus on non-Sickle Cell disease samples. Further, age and sex effects have been under-investigated [9]. Of the articles reviewed here, one reported stronger right-hemisphere lateralization for visuospatial memory in boys as compared with girls [3]. In terms of assessing age-related change, TCD is particularly well placed to investigate changes in infancy, an age-range where other techniques (such as fMRI) are not feasible. Longitudinal analyses would provide opportunities for tracking the development of lateralisation of function. Recent analyses have shown that the relationship between the lateralisation index from fTCD and handedness are complex [38], deserving further assessment. Finally, neurovascular coupling has been investigated in adults simultaneously recording TCD and electroencephography [39, 40]. This experimental design could be translated into child samples to investigate neurovascular coupling in healthy and clinical samples, and relationships with cognitive functioning.

Although all studies employed standardised cognitive and behavioural measures, they varied greatly, which made direct comparisons difficult. We were therefore unable to investigate how TCD measures differentially associate with cognitive domains. This would be a good avenue of future research.