Introduction

Sickle cell disease (SCD) is the most common genetic disorder worldwide affecting 0.1 of 1000 newborn in non-endemic areas and around 2% of children in some African countries [1]. In the most severe forms of SCD, the homozygous SS and the double heterozygous Sβ°, the brain is frequently affected and specific alterations involve both the microcirculation and the macrocirculation. Overt ischemic stroke occurs in 11% of untreated children as a result from stenosis or occlusion in the large arteries of the Circle of Willis [2–3]. Cerebral silent infarcts (CSI), affecting 40% of children by the age of 14, are caused by small vessel disease [4–5] although recent evidence suggests that also a combination of chronic hypoperfusion or hypoxic events, favored by an underlying arteriopathy of the large vessels can lead to CSI [6]. In the past 15 years improvements have been made in the management of stroke and CSI [7]. In fact, algorithms for screening, prevention and management of stroke and CSI based on neuroimaging techniques such as Transcranial Doppler (TCD) and Magnetic Resonance Imaging/Angiography (MRI/MRA) are routinely used in clinical practice [5, 7–8]. Less progress has been made in the management of cognitive dysfunction, a further major morbidity among children and adults with SCD [3,9]. Impairment of cognitive function is reasonable in children who experienced an overt stroke or present CSIs, even at young age, due to the anatomical damage of the brain [3,10–12]. But the pathophysiology of cognitive impairment in children with SCD in the absence of structural abnormalities detected in neuroimaging studies is less clear, although limited evidence suggests that some clinical parameters such as the degree of anemia [13] and oxygen saturation [14] could be involved. In fact, patients with normal TCD and normal MRI/MRA still display cognitive deficits mainly in attention, memory and executive functions, with profound adverse impact on health, education and quality of life [3,9,10–11,15–16]. Hence, the need for biomarkers that bridge the gap between early pathophysiological alterations occurring in the brain (micro-vasoocclusion, small vessel vasculopathy, endothelial dysfunction-intimal proliferation, vascular tone dysregulation), and clinically evident impaired cognition, which seems to be a later manifestation of cerebral damage. Some cerebral abnormalities might be undetectable with conventional imaging studies [17] and the development of more sophisticated imaging techniques might reveal that MRIs which are considered normal with the currently available equipments, present subtle brain parenchymal lesions. In addition, patients with SCD could present functional brain abnormalities not evaluable by conventional studies. Functional imaging studies, whether electroencephalogram (EEG) based or MRI based, have recently been applied in SCD and have shown potential insights in exploring clinical manifestations and cognitive abnormalities in this disease [18–19].

Functional magnetic resonance imaging (fMRI) is a non-invasive imaging technique which allows the measurement of blood oxygen level dependent (BOLD) signal in different brain areas during task performance or resting state. Imaging the brain during resting state, reveals spontaneous, large-amplitude, low-frequency (<0.1 Hz) fluctuations of BOLD signal that are temporally correlated across functionally related areas [20–24]. Hence these signal fluctuations reveal the presence of functional neural networks.

With a short acquisition time, resting state fMRI can be applied in special populations such as children and patients with cognitive impairment at preclinical stages [25–28]. Some resting state networks are already present in utero, while others mature in childhood [24,27]. Around 10 major resting state networks are consistently found in adults.

One of the most widely studied functional network is the default mode network (DMN) [29]. The DMN involves the precuneus and posterior cingulate, the bilateral inferior–lateral–parietal and ventromedial frontal cortex. It seems to represent an organized, baseline mode of brain function that is suspended during any specific task-oriented behaviour and it activates when individuals are engaged in internally focused tasks such as introspection, memories retrieval and conceiving the future [23–24, 29]. DMN structures are typically deactivated during various externally focused or demanding cognitive processes. DMN alterations have been found in patients affected by Alzheimer’s disease [30], schizophrenia [31] and brain injuries [32] even before symptoms appear.

Given the abnormalities found in the DMN in other diseases with impaired cognition at a preclinical level, and our previous finding of abnormal activation in the precuneus during cognitive evoked potentials, in children with SCD independently from large vessel vasculopathy and CSI [18], we aimed at evaluating DMN connectivity by means of resting state fMRI hypothesizing that neurocognitive scores and parameters linked to cognition in SCD, such as hemoglobin [11,13] and oxygen saturation [14], could be related to brain connectivity in the DMN.

Patients and Methods

In this cross-sectional study, consecutive children affected by SS-Sβ° SCD and regularly attending the Sickle Cell Clinic of the Azienda Ospedaliera-Università di Padova [33] were enrolled. As part of the comprehensive care program at our center patients undergo a complete check-up every year and are clinically evaluated at least every 4 months. Since 2009 the complete check-up includes: annual TCD starting at 2 years of age; MRI/MRA every two years starting when sedation is no longer necessary, generally at 5 years, and cognitive evaluation around age 5 or as soon as diagnosis is made if the child is taken in care during school years. MRIs are performed before 5 years of age in case of inability to perform TCD, indeterminate (poor window) or conditional/abnormal TCD, or clinical indication. MRI is also performed as part of pre-bone marrow transplantation evaluation. Functional MRI sequences were acquired during a standard MRI examination (see below). Since 2006 clinical and demographic information on all patients are prospectively and systematically collected in the Sickle Cell Access Database [33]. Inclusion criteria for participating in this study were: 1) brain MRI not affected by movement artifacts and 2) a detailed cognitive assessment and TCD performed within a year of MRI. Forty patients fulfilled the criteria. Previous stroke was not considered an exclusion criteria. Steady state Hemoglobin (Hb) values and day time Oxygen Saturation (SatO2) were recorded at each visit within the year of TCD, MRI and cognitive assessment. For patients on chronic transfusion the blood test was performed on the day of transfusion before the procedure.

Sixteen healthy subjects represented the control group, recruited among healthy age matched controls attending the Neurology Outpatient Clinic who performed MRI for episodic headaches.

Results

Forty patients (39 SS, 1 Sβ°), 21 M 19 F, mean age at brain MRI 8.08 ± 2.83 years (range 4.6–15) were enrolled. Thirty eight patients (95%) were Africans (24 Nigeria, 5 Ghana, 3 Congo, 3 Senegal, 1 Cameroon, 1 Burkina Faso, 1 Togo) and 2 from Central America. Main clinical and demographics characteristics are shown in Table 1. None of the patients had suffered a stroke and neurological examination was normal in all. Mean steady state Hb levels were ≥ 8 mg/dL in 30/40 (range 7.1–10.6 mg/dL) and steady state daytime SatO2 was≥97% in 31/40 (range 95–100%). At time of MRI, 27 patients (67,5%) were not receiving any disease modifying treatment; nine were on chronic transfusion (6 exchange transfusion, 3 top-up transfusion) for abnormal TCD (n.7) and for lack of response to hydroxyurea (n.2) while 4 were receiving hydroxyurea for recurrent acute chest syndromes (n.3) or previous chronic anemia <7 g/dl (n.1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Patients’ demographics, clinical characteristics and neurocognitive scores.

*Only in patients performing WISCIII.

https://doi.org/10.1371/journal.pone.0157090.t001

Sixteen healthy controls, 5 M 11 F, had mean age of 9.98±2.8 (range 5–15).

Default Mode Network Analysis with Resting State Functional MRI

Children with SCD presented significant increased connectivity in a small area of the DMN located in the posterior precuneus (cluster size 34 voxels, peak 28, 21, 37, Fig 1A) compared to controls. Although subgroup comparison (controls, SCD children with FsIQ≥75, SCD children with FsIQ<75) did not reveal significant differences in this area, the mean connectivity within the cluster showed an increasing connectivity from controls, to SCD children with normal cognition and SCD children with low FsIQ (Fig 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Functional MRI group analysis.

(A) Increased connectivity in the Default Mode Network (DMN), in the posterior precuneus, in patients versus controls; (B) Increased connectivity in the DMN, in the medial prefrontal regions, in patients with normal Verbal IQ compared to controls; (C) Increased connectivity in the DMN, in the right prefrontal region, in patients with normal MRI compared to controls; (D) Increased connectivity in the posterior calcarine region in patients with Oxygen Saturation≤97% compared to patients with higher Oxygen Saturation.

https://doi.org/10.1371/journal.pone.0157090.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Connectivity values in the Precuneus in controls, patients with preserved cognition, i.e. normal FsIQ (FsIQ≥ 75), patients with impaired cognition, i.e low FsIQ (FsIQ<75).

https://doi.org/10.1371/journal.pone.0157090.g002

VIQ and PIQ comparison showed no differences among SCD patients, although SCD patients with normal VIQ presented a higher connectivity compared to controls in medial prefrontal regions (3 clusters, n.1: size 211 voxels, peak 22, 53, 40; n.2: size 200 voxels, peak 32, 56, 40; n. 3 size 6 voxels, peak 18, 46, 42, Fig 1B).

Among SCD patients, there was no connectivity difference according to CSI lesion volume, although SCD patients with normal MRI showed a small cluster of increased connectivity in the right prefrontal region (cluster size 23 voxels, peak 25, 57, 39, Fig 1C) compared to controls.

No significant difference in connectivity could be found between SCD patients with normal TCD and those with a history of abnormal TCD nor according to the presence and degree of stenosis at MRA.

Among SCD patients, there was no DMN connectivity difference according to steady state hemoglobin levels, while a SatO2≤97% was associated with increased connectivity in a small area located in the posterior calcarine region (Fig 1D).

Discussion

To our knowledge, our study is the first to investigate the DMN in children with SCD by means of resting state fMRI revealing significant functional changes early in the disease course. In particular, our study revealed increased connectivity within the precuneus in patients compared to controls, more evident among those with decreased cognitive performances. Since the DMN is involved in speculative activities such as memory and planning, its connectivity results decreased in patients with severe cognitive impairment [29–31], but can be increased at a preclincal stage (low or mild impairment). In fact, there is a growing evidence of a compensatory role of the connectivity changes in the early stages of cognitive failure [41–42]. Therefore, the increased connectivity in the precuneus of children with SCD likely represents a similar phenomenon, i.e. the recruitment of the DMN for counterbalancing the progressively emerging cognitive deficits. Interestingly, the precuneus has been shown to present also morphological changes in children with SCD: the precuneus and the posterior cingulate gyrus showed progressive thinning [43–44], and this finding was interpreted as a consequence of regional vulnerability to decreased oxygen delivery in relation to an impaired blood flow. According to our functional data, cortical thickness changes in SCD might result from an increased functional demand combined to an impaired blood flow indicating that the connectivity modulation in areas of the DMN might help to preserve cognitive functions but also induce a progressive regional cortical exhaustion. The absence of significant differences between SCD patients subgroups (normal FsIQ and low FsIQ) vs controls likely reflects the decreased statistical power of small samples and prompts for larger population studies on SCD children and for longitudinal studies evaluating cognition and DMN connectivity changes.

Regarding the functional involvement of the DMN in SCD, cortical sources of abnormal EEG activity in the precuneus [18] in SCD have already been demonstrated, indicating that the brain areas of the DMN (precuneus, posterior cingulate, the bilateral inferior–lateral–parietal and ventromedial frontal cortex) are critically involved in this condition. A similar mechanism of compensation through increased connectivity might occur to maintain the VIQ thus explaining the changes detected in the medial prefrontal regions of the DMN in patients with low VIQ. Nevertheless, language abilities could be better evaluated with specific networks of the dominant hemisphere that require a dedicated analysis of areas specifically involved in language functions.

The reasons of connectivity changes within the DMN remain elusive and confounding factors such as chronic anemia or decreased SatO2 need to be investigated. Interestingly, there was no significant alteration in DMN connectivity in patients with more severe anemia, showing that BOLD signal is not influenced by a low steady state Hb level ≥7 g/dL. Chronic anemia (Hb< 7 g/dL or Ht<20%) is a strong independent risk factor for both CSI [6] and cognitive impairment in the presence of normal MRI [3,13,45]. None of the children in our cohort had a mean steady state Hb<7 g/dL which might justify why there was no difference in brain connectivity according to the degree of anemia in our cohort. Nevertheless, our previous analysis using functional EEG based techniques and analysis of cortical sources of cognitive evoked potentials also failed to demonstrate a correlation with the degree of anemia or a history of anemia in the first 5 years of life, suggesting that anemia alone might not be sufficient to determine impaired cognition in SCD [18]. Some of the patients were on chronic transfusion and this also could have contributed in reducing the importance of Hb value in our analysis. Further studies in larger cohorts would be useful to compare the effect of disease modifying treatments such as HU or chronic transfusions on brain connectivity changes.

Low daytime SatO2, on the contrary, was significantly related to the connectivity of the DMN in our cohort, but revealed connectivity changes in a very limited areas outside the regions that were significantly involved by cognitive performances. Children with lower daytime mean steady state SatO2 presented increased connectivity, with possibly the same mechanism of an initial attempt to compensate the lower delivery of O2 to the tissues. In fact, SatO2 is a clinical factor related to cognition [14,46] and was recently related to white matter damage other than CSI [47]. Probably, the lower O2 delivery to the cerebral tissue, as documented by low SatO2, is only one of the means through which cortical disruption can occur.

The increased connectivity seems to be unrelated to large vessel vasculopathy alone because there was no difference in connectivity according to history of abnormal TCD or presence of stenosis on MRA. Moreover the increased connectivity was also unrelated to CSI. These data might indicate a preclinical alteration in brain connectivity, undetectable by standard neuroimaging. This data suggests that abnormal brain function might begin particularly early in children with SCD, considering the relatively young age of our population (8 years). It might implicate also that cognition in SCD is influenced more by factors limiting O2 delivery to the tissues than by CSI or large vessel vasculopathies. In this sense, correlation of nighttime SatO2 with connectivity and interventions aimed at increasing SatO2 could be significant [48].

Finally, the percentage of CSI and stenosis in our population was higher compared to other cohorts both in the United States and Europe. The majority of our patients were from West Africa and it could be that ethnic differences could play a role in the variability of cerebral vasculopathy in children with SCD as it happens in stroke for adults [49].

Conclusions

In conclusion, resting state fMRI analysis reveals functional abnormalities in brain connectivity that seem independent from the known mechanisms of cerebral vasculopathy in SCD, such as large vessel disease or CSI and are thus not revealed by routinely used imaging techniques. fMRI, together with other functional EEG imaging studies, might represent a useful complimentary analysis for investigating the disease natural course in studies linking physiopathology, anatomical and functional changes in the DMN. In addition, as neural network connectivity might be positively influenced by behavioral, educational or environmental factors [50], children with SCD could benefit from specific treatments and their response could be closely monitored by resting state fMRI up to cognition improvement.

Acknowledgments

The research was performed with Funding from Fondazione Città della Speranza Grants 07/04, 11/04, 14/02.