Subarachnoid hemorrhage (SAH) is an uncommon cause of stroke that disproportionately contributes to increased mortality and morbidity. Globally, SAH affects ~8:100,000 people annually, with a mortality rate of ~30% (1, 2). One of the common complications of non-traumatic SAH is delayed cerebral ischemia (DCI) which can occur as early as post-bleed day 3, with the most frequent incidence occurring within 2 weeks (3). DCI affects up to 30% of patients with SAH and leaves them with a reduced quality of life due to motor deficits and/or cognitive dysfunction (4). The pathophysiology of DCI is complex, but it often occurs in the setting of vasospasm, i.e., aberrant constriction of the cerebral vasculature. Vasospasm is a maladaptive response that can contribute to a decrease in cerebral blood flow (CBF) (5, 6). In turn, reduced CBF is associated with spreading cortical depolarization (7) and DCI (8). As ischemic conditions can lead to poor functional outcomes, treating cerebral vasospasm and avoiding DCI after SAH remain a major unmet clinical need.

Calcium channel blockers (CCBs), which act on smooth muscle cells to dilate the vasculature, have been long investigated for their potential role in the treatment of large-artery vasospasm and the prevention of DCI. Indeed, the only guideline-recommended treatment for the prevention of DCI is the oral administration of dihydropyridine CCB nimodipine (9). Interestingly, the protective effect of nimodipine occurs without altering proximal vasospasm; thus, the effect of this CCB-based treatment is thought to occur on the microvascular level (10). Beyond prophylactic nimodipine, once vasospasm or DCI develops, there is no proven treatment to address them.

Recent data suggest that the intrathecal (IT) delivery of the CCB nicardipine may have a role in a reactive (rather than a preventative) treatment approach to address cerebral vasospasm and avoid DCI (11). This approach has the advantage of avoiding deleterious systemic effects, i.e., decreased blood pressure, that is common with oral and intravenous CCBs (12). An intermittent and titratable IT nicardipine regimen was shown to be associated with proximal vessel vasodilation, reduced risk for DCI, and improved long-term functional outcomes (11, 13). However, the macrovascular vasodilation itself, as quantified by a reduction in daily TCD blood flow velocity, was not associated with outcomes (11). Therefore, the underlying mechanism driving the effect of IT nicardipine on patient outcomes remains unclear.

Both DCI and, to a more controversial extent, vasospasm is thought to occur on the microvascular level (13–16). However, the efficacy of IT nicardipine has only been assessed to date in humans at the macrovascular level (11, 13, 17–19). Although the arteriolar tone and capillary diameter are governed by calcium-dependent processes (the latter via pericytes regulating capillary size) (20, 21), IT delivery does not guarantee that nicardipine successfully diffuses out of the subarachnoid space and reaches the smaller blood vessels within the brain tissue (20, 21). Further investigation into the influence of IT nicardipine on the microvasculature is needed to complement the existing body of work on its macrovascular effects (11, 13) and to improve our mechanistic understanding of the treatment's effect.

Current neuroimaging modalities capable of characterizing the microvascular hemodynamic response to IT nicardipine are limited in the critical care environment. While perfusion magnetic resonance imaging (MRI), positron emission tomography (PET), and/or computed tomography (CT) can quantify microvascular CBF, the cost, need for patient transport, and a lack of continuous monitoring limit their feasibility for capturing transient hemodynamic changes at the bedside. Brain tissue oxygen monitors (e.g., Licox^®) have the inherent disadvantage of their invasive nature. Other indirect measures of microvascular flow, like non-invasive near-infrared spectroscopy measurements of hemoglobin oxygen saturation, are poor surrogates of CBF (22).

In this study, we employ a non-invasive, well-validated optical technique known as diffuse correlation spectroscopy (DCS) to measure regional microvascular CBF (23). In DCS, a near-infrared light source is used to illuminate the tissue, and a detector is placed some distance away to collect light that has randomly scattered throughout the underlying tissue. Light scattering of moving red blood cells leads to temporal fluctuations in detected light intensity. These temporal fluctuations are quantified through a normalized temporal intensity autocorrelation function, g₂ (τ), that is fit to simple analytical models, to extract a blood flow index (BFI) (23). Although the units of BFI (cm²/s) differ from the traditional units of blood flow (ml/min/100 g), numerous validation studies demonstrate that BFI correlates with CBF measured by other “gold-standard” modalities (e.g., MRI) (24, 25).

Herein, we use DCS to test the hypothesis that IT nicardipine increases microvascular CBF from pre-administration levels in the absence of systemic changes. We also explore factors that are associated with the microvascular flow response, including patient characteristics, systemic hemodynamic response, baseline macrovascular flow velocities, and the development of DCI. Finally, we determine the correlation between microvascular CBF measured by DCS and macrovascular flow velocities measured by transcranial Doppler ultrasound (TCD).

Between 1 April 2019 and 31 October 2021, 57 patients with non-traumatic SAH (aneurysmal or diffuse pattern angiogram negative) admitted to the neurocritical care unit at Emory University Hospital consented. Of 57 patients, 37 developed clinically significant large-artery vasospasm that required treatment with IT nicardipine, as determined by the treating-attending physician. The CBF response to the first dose of IT nicardipine was monitored in 27 out of 37; the remaining 10 were excluded due to the availability of the research-based DCS device. Of those 27 monitored with DCS, 20 (74%) were included in our final analysis; the remaining 7 datasets were excluded due to measurement duration of <15 min. Of these seven discarded datasets, premature discontinuation was caused by patient discomfort related to IT administration and EVD clamping (n = 3), failure to properly secure the DCS sensor on the forehead (n = 3), or co-administration of oral nimodipine at the time of DCS monitoring that induced confounding effects on CBF (n = 1). The average duration of DCS monitoring post-nicardipine was 82.7 ± 23.2 min.

Patients were mostly female (70%), with a mean ± standard deviation age of 48.5 ± 10.1 years (Table 1). Aneurysm location and grade severity were heterogeneous, as described in Table 2. An IT nicardipine dose of 4 or 5 mg was administered every 6–8 h with an average of 36 ± 20 doses. Treatment was initiated on post-bleed day 5 ± 1. One patient had evidence of DCI on CT prior to nicardipine initiation. Five out of 20 developed new DCI on the recent imaging available (up to 6 weeks post-bleed); the mean time between the first dose of IT nicardipine to DCI development was 5 ± 4 days. Supplementary Figure 1 shows the DCI-related infarction of each patient diagnosed with DCI.

The average MAP prior to IT nicardipine administration was 103 ± 16 mmHg; the average HR was 80 ± 14 bpm. After administration, no significant trends with time were observed for MAP or HR (both p > 0.05, Figure 2). Average ICP at the start of IT nicardipine and 30 min post-administration was 9 ± 4 mmHg and 11 ± 5 mmHg, respectively (p > 0.05); ICP remained unchanged from IT nicardipine administration during the monitoring session (p = 0.8, data not shown).

Microvascular CBF measured with DCS significantly increased with time after IT nicardipine administration (slope = 0.64%/min, SE = 0.09, p < 0.001), although appreciable response heterogeneity was noted across subjects (Figure 3A). Using a latent class mixture model, 19 out of 20 patients were classified into two distinct classes of CBF response; Class 1 (n = 6, in magenta in Figure 3B), on average, showed a minimal change in rCBF post-administration, whereas Class 2 (n = 13, in green in Figure 3B), on average, exhibited a steady increase in rCBF over the course of monitoring. Notably, for this analysis, visual inspection revealed 1 distinct trajectory that the LCMM model was overly sensitive (shown in blue in Figure 3B); this outlier trajectory was excluded from the LCMM analysis. No significant differences in age, SAH grade, number of doses, depth to the brain, baseline MAP, baseline HR, or baseline ICP were seen between the LCMM-identified classes of CBF response to nicardipine (Table 3, all p > 0.05). Patients in Class 1 were mostly female (100% vs. 54% in Class 2, p = 0.044) and tended to have a higher pretreatment MCA velocity in the hemisphere of DCI monitoring (151 vs. 126 cm/s in Class 2, p = 0.02). Most strikingly, the frequency of DCI was significantly different between classes (p < 0.001); 5 out of 6 subjects in Class 1 had DCI, 4 of whom developed new DCI, despite treatment with IT nicardipine, on the recent imaging available up to 6 weeks post-bleed, and 1 of whom started with DCI prior to nicardipine, while only 1 out of 13 subjects in Class 2 had DCI (Figure 3C). Further review of the excluded outlier revealed that the CBF trajectory was correlated with these trends; this patient had a sharp increase in CBF, similar to those in Class 2, and did not develop DCI.

Of 20 patients, 18 had TCD measurements prior to IT nicardipine initiation (4.5 ± 1.9 h, time difference between TCD and DCS measurements). Average pretreatment MCA and ACA maximal mean blood flow velocities were 132.3 ± 34.2 cm/s and 98.9 ± 28.5 cm/s, respectively. In this subset of patients, we observed a significant positive correlation between mean resting-state blood flow velocity in the MCA and baseline, pretreatment BFI measured by DCS (r_s = 0.69, p = 0.0014) (Figure 4A). A similar trend was seen between mean flow velocity in the ACA and baseline BFI, although it did not reach statistical significance (r_s = 0.42, p = 0.083, Figure 4B).

Here, we used DCS to quantify the microvascular CBF response to IT nicardipine administration in SAH patients with large-artery cerebral vasospasm. On average, microvascular CBF increased over 90 min of monitoring. This CBF increase was observed in the absence of systemic changes in blood pressure or HR, suggesting that the cerebral hemodynamic response is likely due to a local drug effect rather than systemic influences and further strengthening the plausibility of microvascular effects on top of the macrovascular effects, as previously reported (11, 13). However, we observed substantial heterogeneity in this response across patients. A latent class mixture model was used to objectively identify two distinct classes of CBF response, i.e., Class 1, wherein a minimal-to-no change in CBF was observed, and Class 2, wherein a steady increase in CBF was observed. Interestingly, this differential CBF response was significantly associated with DCI. Notably, 1 out of 13 patients with increased CBF after IT nicardipine developed DCI, while 5 out of 6 patients whose CBF did not respond to IT nicardipine had DCI. Although these are preliminary observations, they suggest that the CBF response to the first dose of IT nicardipine could be a novel biomarker for treatment response and DCI risk, alerting clinicians to consider alternative interventions. Given the strong association between DCI and poor functional outcome (4, 34, 35), real-time identification of high-risk patients, or those who do not respond to nicardipine treatment, is of high clinical significance.

Several possible mechanisms may explain the lack of CBF response to nicardipine in the Class 1 subset. MCA velocity was higher in Class 1 compared to Class 2 (p = 0.02, Table 3), suggesting that the severity of MCA vasospasm may be worse in Class 1. Many vasospasm-independent mechanisms, including microthrombosis, venous outflow obstruction, and cortical spreading depolarization, may be present in these patients that are unaffected by nicardipine (36, 37). Alternatively, the additional volume of drug and diluent into the cranial vault could potentially transiently decrease cerebral perfusion, while ICP remains minimally changed. Furthermore, the timing of treatment may influence the CBF response. Using a rabbit SAH model, Vorkapic et al. (38) demonstrated two phases of large-artery vasospasms, i.e., an early pharmacologically reversible phase and a later irreversible phase. During the acute reversible vasospasm stage, cerebral vasoconstriction could be reversed by CCBs. However, as organic changes such as intimal proliferation, edema, and degeneration occur, the efficacy of pharmacological interventions is reduced (17, 38). Regardless of the mechanism, coupling DCS with a pharmacological challenge may be a valuable risk-stratification tool to measure a treatment effect in the microvascular bed alongside TCD assessment of the macrovasculature to better guide clinical decisions and minimize DCI after SAH.

A secondary aim of this study was to investigate the relationship between DCS measurements of resting-state, pretreatment microvascular blood flow, and TCD measurements of macrovascular flow velocity. We observed a moderate positive correlation between the DCS-measured BFI which is measured prior to nicardipine administration and the TCD-measured mean blood flow velocity in the MCA which is measured on the morning of DCS monitoring. A similar correlation was trending toward significance in the ACA (p = 0.082). Of note, the sensor was positioned over the watershed area between the ACA and MCA territories. This positive correlation between resting-state microvascular and macrovascular flow is seen in healthy adults, including by DCS vs. TCD (39, 40), as well as in brain-injured cohorts, including SAH (41–43). In SAH patients with macrovascular vasospasm, we expected a negative or null correlation, as elevated TCD velocities are indicative of vasoconstriction that may have downstream effects of reducing microvascular blood flow. However, microvascular blood flow is reported to only be significantly reduced in the setting of severe large-artery vasospasm (44), while most of our cohort was classified as either mild or moderate vasospasm. These results may suggest that this macrovascular and microvascular correlation in the MCA territory remains intact at rest in our cohort of patients with mild-to-moderate large-artery vasospasm. We note that TCD and DCS were not measured concurrently during/post-administration of IT nicardipine due to technical challenges of simultaneous monitoring; thus, we cannot infer how the treatment influences macro-micro correlation. In the future, measurements of microvascular blood flow with DCS in concert with macrovascular TCD assessments may serve a role in dissecting the complicated CBF dynamics after SAH.

Our study has several limitations. First, by design, this was an observational study that involved dynamic monitoring of patients treated as part of their usual care without a placebo comparison. Limitations regarding the small sample size and study design necessitate confirmation of these results in the setting of a randomized controlled trial. Second, seven (26%) DCS monitoring sessions were discarded from the analysis. While this incidence is appreciable, we anticipate a significantly lower discard rate with continued future enrollment as we continually make improvements to the research device that enhances SNR. Third, based on previous reports of a short half-life of nicardipine (18, 45), we limited monitoring for up to 90 min, although monitoring was discontinued prior to 90 min in several patients due to instrument availability or the administration of therapeutic agents known to exert confounding influences on cerebral hemodynamics (e.g., oral nimodipine). With a larger sample size and extended monitoring, we will explore the necessary duration of monitoring required to robustly distinguish the dichotomy between the patients who did and did not respond favorably to IT nicardipine. Fourth, we were limited by SNR to a single measurement location for DCS, which we selected to be the hemisphere in which higher TCD velocities were measured (suggesting worse large-artery vasospasm). As a result, it is difficult to ascertain if our local measurement is indicative of changes that persist throughout the brain, or if there is regional heterogeneity in the response. In general, intrathecally administered nicardipine is expected to distribute within the CSF compartment across the subarachnoid space where blood vessels travel. Thus, the vasodilatory effects should be diffused as nicardipine permeates blood vessels within the subarachnoid space. However, the regional response accessible to DCS could differ from that of other vascular territories. In the future, increasing the number of laser sources and/or detectors may enable simultaneous monitoring of multiple locations around the head to determine if regional response heterogeneity is present as well as to explore factors associated with response heterogeneity (e.g., SAH and/or vasospasm location/severity). Finally, we employed the semi-infinite solution to the correlation diffusion equation (23) to analyze DCS data, which models the head as a homogenous medium. While the correlation with TCD suggests that this approach provides a suitable first-order approximation, future studies should employ multi-layer approaches (46) that enhance sensitivity to cerebral perfusion and minimize contributions from extracerebral layers.

In summary, we use DCS to characterize the cerebral hemodynamic response to IT nicardipine treatment in patients with large-artery cerebral vasospasm after SAH. We observed the following two distinct classes of microvascular CBF response: (1) minimal-to-no change and (2) a steady increase. We found that 83% of patients whose CBF failed to respond to nicardipine developed DCI, while 8% of patients whose CBF increased developed DCI. These results suggest that the acute (< 90 min) DCS-measured CBF response to IT nicardipine is associated with intermediate-term (up to 3 weeks) development of DCI. If these findings are corroborated with future, larger-scale prospective, randomized control trials, they suggest that DCS, alongside TCD, may provide a valuable biomarker to better inform clinical decisions, diminish the development of DCI, and improve functional outcomes following SAH.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Emory University Institutional Review Board (IRB00107474). The patients/participants provided their written informed consent to participate in this study.

Data acquisition: ES, KC, AL, SL, and ZF. Data analysis: ES. Data interpretation, discussion, and writing—reviewing and editing: ES, CM, FA, OBS, PK, OS, and EB. Statistical analysis and writing—original draft preparation: ES, CM, OS, and EB. Supervision: OS and EB. All authors have read and agreed to the published version of the manuscript.