Thrombospondin-1 and L-selectin are associated with silent cerebral infarct in children with sickle cell anaemia

Silent cerebral infarction (SCI) is a prevalent problem in sickle cell anaemia (SCA, defined herein as HbSS or HbSB0 thalassaemia); by age six years, 27% of children with SCA have had a SCI (Kwiatkowski et al, 2009) and by age 14 years, 37% (Bernaudin et al, 2011). Children with SCI are at greater risk for overt stroke, recurrent or progressive SCI, and impaired cognitive function than the general SCA population. SCI is presumed, but not proven to represent infarctive injury. The mechanism underlying SCI in SCA is not fully understood, but is probably multifactorial. Anaemia and relative high blood pressure, endothelial cell activation, hypercoagulopathy, alterations in vasomotor tone secondary to decreased nitric oxide bioavailabilty, reperfusion injury, intravascular haemolysis and the impaired haemodynamics and resultant increased blood flow to the brain that result from anaemia have been postulated to contribute to the pathogenesis of infarction in SCA patients (DeBaun et al, 2012; Kato et al, 2007). Thrombospondin 1 (THBS1, TSP1) and L-selectin have been implicated in sickle cell-related complications, including vaso-occlusive crisis and stroke, but their role in SCI has not been evaluated. We tested the hypothesis that children with SCA and SCI would have altered plasma THSB1 and L-selectin levels when compared to children with SCA and no SCI. We used commercially available immunoassays to measure plasma THSB1 and L-selectin levels in 116 children with SCA (n=65 with SCI, n=51 without SCI, one sample per patient). These steady state plasma samples were obtained from the Silent Infarct Transfusion (SIT) Trial Biologic Repository at the Johns Hopkins University. This was a convenience sample, selected from 524 available undiluted plasma samples from patients who were enrolled in the trial between February 2007 and May 2009 and had plasma samples that were processed per protocol (Casella et al, 2010). Baseline characteristics were similar between the SCI and non-SCI groups; no differences in mean Hb, white blood cell or platelet counts were observed. Children with SCI had higher median THSB1 values than the non-SCI controls (8.4 versus 6.2 μg/ml for SCI and non-SCI, respectively, P=0.03; Fig 1A). Compared to without SCI, 14% (n=9/65) of children with SCI had THSB1 values above the 95th percentile for non-SCI THSB1 values (Fig 1B). L-selectin was significantly higher in children with SCI when compared to non-SCI controls (median values of 1.46 versus 1.35 μg/ml for SCI and non-SCI controls, respectively, P=0.03; Fig 2A). Fifteen percent (10/65) of children with SCI had L-selectin values above the 95th percentile for non-SCI values (Fig 2B). Baseline oxygen saturation, a reported risk factor for a number of sickle-related complications, correlated inversely with THSB1 in children with SCA (r=-0.4, P<0.001). Systolic blood pressure, a known risk factor for SCI, correlated directly with L-selectin in children with SCI (r=0.28, P=0.02). Fig 1 THSB1 concentrations obtained from children with sickle cell anaemia (SCA). (A) Box plot of THSB1 concentrations for children with silent cerebral infarction (SCI; n=65), non-SCI (nonSCI; n=51). The middle horizontal line inside box indicates median values ... Fig 2 L-selectin concentrations obtained from children with sickle cell anaemia. (A) Box plot of L-selectin concentrations for children with silent cerebral infarction (SCI; n=65) and non-SCI (nonSCI; n=51). The middle horizontal line inside box indicates median ... THSB1 and L-selectin were chosen for this study based on biological plausibility. Our results indicating differentially elevated plasma THSB1 in SCA children with SCI suggest that these patients are also at risk for an increased inflammatory response that is observed after recurrent ischaemia/hypoxia and reoxygenation. Furthermore, we observed a correlation between plasma THSB1 and baseline oxygen saturations, which support other studies that have reported increased risk for complications, such as overt stroke, in SCA patients with steady state oxygen desaturation (Quinn & Sargent, 2008). These findings may have practical implications for therapeutic intervention with hydroxycarbamide, which decreases the in vitro adhesion of sickle erythrocytes to THSB1 (Hillery et al, 2000), and aspirin, which has been shown to inhibit the release of THSB1 (Coppinger et al, 2007). The increased L-selectin levels observed in SCA suggest increased systemic inflammatory stress and provide evidence for concomitant immune activation. Whereas previous studies report associations of L-selectin with various clinical parameters, including decreased oxygen saturations, we report a direct relationship with SBP. We and others have reported an increased risk of SCI with increased SBP (DeBaun et al, 2012), strengthening the plausibility of elevated L-selectin as a potential risk factor for SCI. If this observation were validated, additional studies to evaluate for a potential neuroprotective role for selective- and pan-selectin inhibitors would be informative. Although we have identified the first two plasma proteins that are statistically significantly associated with SCI in children with SCA, our study was not designed to measure time-dependent correlations with brain injury. There is evidence to suggest that children with SCA are at risk for both chronic and intermittent acute cerebral ischaemia events (Savage et al., 2011; Quinn et al, 2013); however, it is not possible to assess whether increased plasma THSB1 and L-selectin levels reflect ongoing, past, or sporadic subclinical brain injury. We have previously reported that plasma levels of GFAP, an intracellular glial cell intermediate filament protein, are elevated in a significant fraction of children with SCA and are associated with acute cerebral infarction in SCA, with a weak association with SCI; inverse correlations with performance IQ were observed (Savage et al, 2011), as well as with elevated systolic blood pressure, a known correlate of SCI and stroke in SCD (DeBaun et al, 2012). The GFAP study was limited by the same time-dependency issues as the present study; however, the finding of correlations in both studies, despite the fact that samples were most probably obtained remotely from the event, suggests that detection of central nervous system events and ongoing injury in SCA may be feasible using biomarkers. In addition, patients with higher risk of stroke, including those with previous stroke and abnormal TCDs, were excluded from the SIT trial, which introduces concerns for generalizability. Similarly, the use of nonrandom sampling introduces the potential for sampling bias, and may limit our ability to make specific inferences; however, the lack of difference in clinical parameters between the two groups is reassuring in this regard. Future studies to examine how longitudinal changes in THSB1 and L-selectin and other potential markers of brain injury, such as GFAP, correlate with clinical outcomes and quantitative magnetic resonance imaging measures of lesion burden are needed in order to determine causality and to assess risk for SCI. In summary, our results indicate that THSB1 and L-selectin are associated with SCI in children with SCA. These findings, which may provide additional insight into the pathophysiology of disease for children with SCA, need to be substantiated in additional studies and clinical trials.