Blood markers in remote ischaemic conditioning for acute ischaemic stroke: data from the REmote ischaemic Conditioning After Stroke Trial

Remote ischaemic per‐conditioning (RIC) is neuroprotective in experimental ischaemic stroke. Several neurohumoral, vascular and inflammatory mediators are implicated. The effect of RIC on plasma biomarkers was assessed using clinical data from the REmote ischaemic Conditioning After Stroke Trial (RECAST‐1).


INTRODUC TI ON
Remote ischaemic per-conditioning (RIC)-inducing ischaemia/ reperfusion distant to the brain-in experimental ischaemic stroke is neuroprotective [1][2][3]. Several neurohumoral, vascular and inflammatory mediators are implicated, but the underlying mechanisms have not been elucidated. Animal models have demonstrated several inflammatory, neuroprotective and vascular biomarkers that may be influenced by RIC, primarily in the setting of myocardial ischaemia and infarction [1].
There is a paucity of data on the effects of RIC on potential mechanistic pathways in human patients with acute ischaemic stroke. Plasma heat shock proteins (HSP) 27, 60 and 70, C-reactive protein, S100-β, matrix metalloproteinase-9 (MMP-9), troponin T and endocannabinoid data have been reported in acute ischaemic stroke patients undergoing RIC in the REmote ischaemic Conditioning After Stroke Trial (RECAST-1) [4]. There were significant increases in total and phosphorylated HSP 27 in the RIC group, whilst the other biomarker levels did not change. Whether other inflammatory, neuroprotective or vascular markers are influenced by RIC in acute ischaemic stroke is unknown.
Several inflammation-related biomarkers including adipsin [5] α2-macroglobulin (A2 M), serum amyloid protein (SAP) [6] and E-selectin [7] have been associated with clinical outcome in stroke patients. Whether these or other elements of inflammatory pathways are influenced by RIC is unclear. Similarly, biomarkers contributing to vascular or neuroprotective pathways-vascular endothelial growth factor (VEGF) [8] von Willebrand factor (vWF) [9] or nitric oxide (NO)-may have roles in RIC [2,10] but have not been assessed in humans with acute ischaemic stroke.
Given the limited data regarding RIC and biomarkers in ischaemic stroke patients, in exploratory and hypothesis-generating analyses the effect of RIC on plasma biomarkers and the correlation of these biomarkers with clinical outcomes in acute ischaemic stroke patients in RECAST-1 were assessed.

ME THODS Population
RECAST-1 was a pilot, single-centre, randomized, blinded, sham-controlled trial in 26 patients with ischaemic stroke, who were randomized to receive four 5-min cycles of RIC or sham within 24 h of onset. The main results of the trial have been published: RIC was well tolerated, appeared safe and feasible in ischaemic stroke patients, and increased plasma HSP 27 [4]. In brief RECAST-1 recruited 26 adult patients from Derby Teaching Hospitals NHS Foundation Trust, UK, with ischaemic stroke within 24 h of ictus with arm and/or leg weakness. The intervention comprised four cycles of intermittent upper limb ischaemia with 5 min inflation 20 mmHg above the systolic blood pressure and 5 min deflation using a manual blood pressure cuff on the non-paretic arm.
The sham group received cuff inflation to 30 mmHg only. The trial was registered (ISRCTN 86672015).

RE SULTS
In total, 26 participants were recruited with a mean age of 76 (10.5) years, 35% female, with moderate severity ischaemic strokes (NIHSS 6) and were randomized at 15.8 h after stroke onset. Bloods were taken before the intervention, immediately after and on day 4. Baseline characteristics were well balanced between treatment groups, although the sham group had more diabetes mellitus (Table 1).

Inflammation-related blood biomarkers
Serum amyloid protein levels reduced from pre-to post-intervention in those randomized to RIC (n = 13, two-way ANOVA, MD 10.47, 95% CI 0.30-20.64, p < 0.05) but did not in those randomized to sham ( Figure 1). Levels of TNF-α also fell from pre-to post-intervention in those randomized to RIC (n = 13, two-way ANOVA, MD 3.58, 95% CI 0.40-6.77, p < 0.05). No significant changes over time, or by treatment, were seen for adipsin, A2 M, GCSF-A or E-selectin ( Figure 1). IL-1RA, an immunomodulatory cytokine that inhibits the actions of IL-1α and IL-1β, fell from pre-to post-intervention in the sham group (n = 13, two-way ANOVA, MD 17.11, 95% CI 2.46-31.77, p < 0.05) but did not change in those randomized to RIC ( Figure 2).
No significant changes pre-to post-intervention or to day 4, or by treatment allocation, were seen for the other interleukins measured ( Figure 2).

Neuroprotective and vascular blood biomarkers
Plasma NO, vWF or VEGF levels did not change over time or differ between treatment groups ( Figure 3).

Association between absolute biomarker levels and clinical outcomes
Overall, higher S100-β levels at pre-intervention, post-intervention and day 4 were significantly moderately correlated with worse NIHSS scores at day 90, but not mRS or BI (Table 2, Figure   S1). Higher IL-1RA levels at day 4 were significantly moderately was mainly seen in those randomized to sham, not RIC (Tables   S1 and S2). Inconsistently, higher IL-6 levels at day 4 were significantly strongly correlated with worse NIHSS scores overall.

F I G U R E 1
Effect of RIC on inflammation-related blood biomarkers. A2 M, alpha-2 macroglobulin; GCSF, granulocyte colony-stimulating factor; SAP, serum amyloid protein; TNF, tissue necrosis factor. *p < 0.05 in RIC group over time, two-way ANOVA.  Table 3, Figure S3a,b). These significant correlations were seen in those randomized to sham but not in those randomized to RIC (Tables S3 and   S4). Overall, an increase in A2 M pre-intervention to day 4 was weakly and moderately correlated with improved day 90 mRS (r = −0.359, p = 0.07) and MMSE (r = 0.465, p = 0.019) respectively. An increase in HSP 60 pre-intervention to day 4 was moderately and strongly correlated with worse clinical outcomes at day 90 (NIHSS, r = 0.641, p = 0.007; mRS, r = 0.500, p = 0.049; BI, r= −0.523, p = 0.038; Figure S4).
Overall, an increase in IL-6 pre-to post-intervention was strongly correlated with improved clinical outcomes at day 90 (NIHSS, r = −0.584, p = 0.059; mRS, r = −0.677, p = 0.022; BI, r = 0.690, p = 0.019), which was mainly seen in those randomized to sham, not RIC (Tables S3 and   S4). In contrast, an increase in IL-6 over 4 days was strongly correlated with worse NIHSS and non-significantly weakly correlated with worse mRS and BI. A similar picture was seen for IL-10 with increasing levels over 4 days being associated with a worse NIHSS (r = 0.592, p = 0.026).
An increase in MMP-9 pre-to post-intervention was significantly moderately correlated with worse clinical outcomes at day 90 ( Figure S5); no such association was seen pre-intervention to day 4. An increase in S100-β pre-intervention to day 4 was moderately correlated with higher NIHSS scores at day 90. No significant correlations with outcome were seen for adipsin, GCSF-A, E-selectin, TNF-α or other HSPs (Table 3).
An increase in NO levels pre-intervention to day 4 had a tendency towards being weakly correlated with worse clinical outcomes at day 90 (NIHSS, r = 0.372, p = 0.062; mRS, r = 0.334, p = 0.10; BI, r = −0.355, p = 0.08, Table 3, Figure S6). Increases in VEGF levels pre-to post-intervention and pre-intervention to day 4 were non-significantly weakly correlated with better clinical outcomes.
No significant correlations were noted for vWF.

DISCUSS ION
Here, in the first analysis assessing blood biomarkers in RIC in acute ischaemic stroke patients, it has been demonstrated that RIC reduced SAP and TNF-α levels. Overall, increasing SAP levels pre-to post-intervention and pre-intervention to day 4 were significantly moderately correlated with worse clinical outcomes after ischaemic stroke. Further, higher S100-β levels at pre-intervention, postintervention and day 4 were significantly moderately correlated with worse NIHSS scores at day 90. No other biomarkers had significant and consistent correlations with outcome at day 90.
In the primary RECAST-1 publication a significant increase in total and phosphorylated HSP 27 was reported in the RIC group, whilst other HSPs (60, 70) and C-reactive protein, S100-β, MMP-9, troponin T and endocannabinoid levels did not differ between treatment groups [4]. These data are added to by demonstrating that RIC was associated with reductions in SAP and TNF-α pre-to post-intervention but did not influence other inflammatory cytokines or putative neuroprotective biomarkers.  pro-inflammatory markers over a longer period. Whether repeated cycles of RIC over the first days prolong the reduction in these pro-inflammatory markers is a question for future studies.
In contrast to the present study, data from 50 healthy volunteers revealed that RIC was associated with elevated VEGF and four inflammation-related markers (transforming growth factor β1, leukaemia inhibitory factor, MMP-9 and tissue inhibitor of metalloproteinases 1) 1 h post-RIC [11]. These findings in healthy volunteers cannot be extrapolated to acute stroke patients. However, a recent study assessing RIC five times a week for 8 weeks after acute ischaemic stroke (post-conditioning) found that those who received RIC had lower levels of nuclear factor κB (an inflammatory transcription factor) and toll-like receptor 4 (a mediator of systemic inflammatory response) and improved cerebral collateral circulation scores on computed tomography angiography [12]. This anti-inflammatory effect of RIC was also seen in older patients with symptomatic intracranial arterial stenosis who received 180 days of bilateral arm ischaemic preconditioning resulting in reduced markers of inflammation including highly sensitive C-reactive protein and IL-6 [13]. Here, it has been demonstrated that RIC administered as a one-off dose of four 5-min cycles of limb ischaemia may reduce some pro-inflammatory markers whilst promoting HSP 27 which may prevent breakdown of the bloodbrain barrier [14].
An increase in SAP from pre-to post-intervention and pre-intervention to day 4 was associated with worse clinical outcomes at day 90, including cognition, across the trial population. SAP has an important role in inflammation and innate immunity, activating the classical complement pathway with increased SAP levels being associated with increased death at day 90 after acute ischaemic stroke [6]. SAP binds to all forms of amyloid fibrils, is present in amyloid deposits and prevents proteolysis of the amyloid fibrils of Alzheimer disease [15]; its role in vascular cognitive impairment, however, is not well understood. Nonetheless, proof-of-concept studies of vascular cognitive impairment and RIC demonstrate twice daily RIC for 1 year improved visuospatial and executive function and reduced white matter hyperintensities compared with sham in a small study of 30 patients with cerebral small vessel disease [16]. One putative mechanism is that RIC reduces pro-inflammatory biomarkers (such as SAP), thereby improving vascular health and building ischaemic tolerance, resulting in improved clinical outcomes including cognition. Indeed, repeated RIC (or chronic RIC) has beneficial effects on vascular and endothelial function in healthy volunteers [17] and on cerebral blood flow after stroke caused by intracranial stenosis [18] and may be of therapeutic use in treating chronic inflammation associated with atherosclerotic disease [19].  [20,21] in the context of mechanical thrombectomy [22] and in planned larger phase III efficacy trials including RECAST-3 (ISRCTN63231313) and Remote Ischemic Conditioning in Patients With Acute Stroke (RESIST, NCT03481777).
There are a number of limitations in our study. First, it was powered for tolerability of the intervention and not to detect changes in biomarkers; hence, the biomarker analyses may be underpowered.
Further, the associations seen here may represent chance, perhaps in part due to multiple testing, and require confirmation in larger cohorts. In these exploratory and hypothesis-generating analyses, two putative biomarkers were positively influenced by RIC and reassuringly there were no concerning safety signals from this dataset. Secondly, participants were recruited 16 h after stroke onset, which may be too late for RIC to exert any potential beneficial effect.
Other studies have randomized patients in the hyperacute period and blood biomarker comparison with the present study may prove illuminating and instructive. Thirdly, RECAST-1 performed one set of four cycles of RIC and it is unclear how many cycles of RIC and over what time period are optimal; the following RECAST-2 trial addressed this [20]. Last, advanced imaging data are not available to quantify lesion size, which may have potentially influenced our findings within or between treatment groups.
In summary, increases in plasma levels of SAP are associated with worse clinical outcomes after ischaemic stroke. RIC reduced SAP and TNF-α levels from pre-to post-intervention. These findings should be considered preliminary given the size of the study. Larger studies assessing biomarkers, safety and efficacy of RIC in acute ischaemic stroke are warranted; the RECAST-3 (ISRCTN63231313) and RESIST (NCT03481777) trials are addressing this question.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this paper are available from the corresponding author upon reasonable request.