Simultaneous monitoring of the middle cerebral and basilar arteries to detect right-to-left shunts using transcranial Doppler by agitated saline administration

Transcranial Doppler (TCD) is an easy, non-invasive, and real-time monitoring device for detecting right-to-left shunts (RLS). Nonetheless, it has limited benefits in patients with poor temporal windows. Therefore, we aimed to investigate whether the basilar artery (BA) window was as effective as the middle cerebral artery (MCA) in detecting RLS during TCD monitoring. Overall, we enrolled 344 patients with stroke, transient ischemic attack, headache, or dizziness. MCA and BA were monitored using a modified headset. To investigate the feasibility of the suboccipital window in detecting RLS, we instituted an evaluation tool with three tiers to evaluate microembolic signals (MESs) during TCD monitoring. Tier 1: TCD monitoring of the MCA (bilaterally) in the resting state, tier 2: TCD monitoring of the MCA (bilaterally) while performing the Valsalva maneuver, and tier 3: TCD monitoring of the index MCA and BA while performing the Valsalva maneuver. In tiers 2 and 3, a high agreement rate of 0.808 and 0.809 (p < 0.001), respectively, on the weighted kappa index, and a high intra-class correlation coefficient of 0.982 and 0.986 (p < 0.001), respectively, were observed on detecting MESs. Our data suggests that the BA window is as effective as the MCA window for detecting RLS on TCD.


Results
Demographic characteristics. The mean age of the patients was 58 ± 16 years, and the male: female ratio was 165:179. Among the patients, 129 (37.5%) had hypertension, 70 (20.3%) had diabetes mellitus, and 92 (26.7%) had dyslipidemia. The reasons for the test were stroke (n = 229), headache (n = 53), transient ischemic attack (TIA) (n = 22), white matter hyperintensities (WMHI) (n = 20), transient global amnesia (TGA) (n = 11), and dizziness (n = 9). Among these patients, 53 (15.4%) had PTWs; 41 had PTWs on both sides and 12 had it on one side only (eight on the right side and four on the left side) detail the demographic characteristics of the patients in this study are shown in Table 1.
Outcomes of the TCD monitoring parameters in both of the MCAs while performing the Valsalva maneuver in Tier 2. A total of 344 patients were measured simultaneously in both the MCAs of TCD monitoring with Valsalva maneuver in Tier2. In 41 patients with bilateral PTWs, bilateral internal carotid artery (ICA) was used instead of MCA on TCD examination. Also, in 12 patients with unilateral PTW, unilateral ICA and contralateral MCA were used on TCD monitoring (Table 1). When measured simultaneously in both of the MCAs, the average number of microembolic signals (MESs) were 14.97 ± 31.18 (range 0-298) in the right MCA, and 14.94 ± 32.41 (range 0-303) in the left MCA in tier 2. The measured MESs was categorized using the International Consensus Criteria or Spencer's Logarithmic Scale (SLS) in both the MCAs, and we compared the concordance of right and left MCA using weight kappa index. The weighed kappa index was 0.687 (p < 0.001) in both the MCAs when assessed using the International Consensus Criteria (Table 2). In addition, the weighed kappa index was 0.808 (p < 0.001) when both the MCAs were evaluated using the SLS (Table 3). Moreover, when we had compared the degree of agreement on count of MESs in both the MCA using the intraclass correlation coefficient (ICC), the ICC was 0.982 (p < 0.001) on comparing the number of MESs measured in both MCAs (Fig. 1).
Outcomes of the TCD monitoring parameters in both the index MCA and BA while performing the Valsalva maneuver in Tier 3. The index MCA(iMCA) and BA were measured in 303 patients, after excluding 41 out of 344 patients due to PTWs on both sides. Out of a total of 303 patients, 182 (60.1%) on the right and 121 (39.9%) on the left were measured simultaneously with the BA. When MESs was measured simultaneously, the average number of bubbles was 14.15 ± 36.32 (range 0-505) in the iMCA and 14.35 ± 38.85 (range 0-559) in the BA.
When iMCA and BA were categorized using the International Consensus Criteria to evaluate the concordance, The weighted kappa index was 0.812 (p < 0.001) when estimated (Table 2). Also, the weighted kappa index was 0.807 (p < 0.001) when iMCA and BA were evaluated using the SLS (Table 3). Also, when we had compared www.nature.com/scientificreports/ the degree of agreement on count of MESs in the iMCA and BA using the ICC, the ICC was 0.986 (p < 0.001) in the iMCA and BA (Fig. 1). The measurements for both the MCA and BA were similar when compared with the MCA measurements bilaterally.

Discussion
This study shows that the suboccipital window for insonation of the BA is as effective as the temporal window for MCA in detecting RLS using the TCD. Approximately 5-37% of stroke patients have PTWs that prevent sufficient TCD insonation of cerebral vessels around the circle of Willis [19][20][21][22][23] . The PTW rates are higher in Asians than in Caucasians 17,19 . It was difficult to  www.nature.com/scientificreports/ evaluate RLS using TCD in patients with PTWs. Therefore, some studies had attempted to detect RLS using the suboccipital windows, submandibular ICA, or the orbital windows rather than the temporal windows [30][31][32][33] . The advantages of using agitated saline while performing the TCD in a suboccipital window (for insonation of the BA) instead of the submandibular window or the orbital windows can be explained by several reasons in our study.
First of all, this study is meaningful in that the suboccipital window using BA has the similar diagnostic value as the temporal window using MCA to quantitatively detect RLS. One study identified MESs using the orbital window for ICA and the temporal window for MCA to detect RLS, and not the suboccipital window in a TCD study 33 . Another study compared the monitoring of cervical arteries (submandibular ICA and vertebral artery [VA]) to that of the MCA. The study addressed that the TCD of the cervical submandibular ICA and VA is a valid screening method to detect RLS due to a PFO 32 . However, most studies have replaced the temporal window due to a PTW with a suboccipital window to detect RLS using the TCD. The studies had demonstrated the usefulness of the suboccipital window (for monitoring the VA) in detecting RLS using the TCD 30,32 . Although these studies had determined that the suboccipital window had a diagnostic ability that was equal to that of the MCA, there are no quantitative studies detecting the extent of RLS, or comparing the temporal and suboccipital windows using correlation coefficients and reliable statistics.
In addition, the suboccipital window was similar to the temporal window in evaluating the PFO grade in a wide range (min 0 to max 509 MESs) as well as identifying the measured MESs at the same time 34,35 . Since the blood flow amount of the BA is quite equal to that of the MCA, the diagnostic ability for detecting RLS may be similar in the TCD monitoring. The amount of blood in the vertebrobasilar system is lower than that (about 20%) in the carotid circulation 36 . A study had evaluated the flow in the proximal and distal cerebral arteries using high-resolution phase-contrast magnetic resonance imaging. The study showed that the total cerebral blood flow (717 ± 123 mL/min) was distributed (on each side) as follows: middle cerebral artery, 21%; distal MCA, 6%; anterior cerebral artery (ACA), 12%; distal ACA, 4%; ophthalmic artery, 2%; posterior cerebral artery, 8%; and basilar artery, 20% 37 , which may be consistent with our investigations of a similar blood flow amount between MCA and BA. Therefore, the suboccipital window might be a better alternative than the submandibular ICA or orbital window in patients with PTWs. Several studies have shown that ischemic lesions are more common in the posterior cerebral circulation of patients with RLS found using the Valsalva maneuver 29 . However, we did not show that MESs was more common in the posterior cerebral circulation of patients during the Valsalva maneuver.
The suboccipital window using BA is as convenient as the temporal window when diagnosing RLS in TCD monitoring. A quantitative study conducted by Guo et al. that compared the left MCA and the left VA for the detection of RLS showed no significant differences between the two (including constant and provoked RLS) 30 . However, Guo's study had also several limitations. First, the probe for VA monitoring was manually positioned. The VA might have been missed while testing if its course had changed. Second, The VA has many variations including hypoplasia 38,39 . VA hypoplasia (VAH) is not an uncommon congenital variation of the VAs leading to asymmetry. VAH with caliber discrepancies of more than 1:1.7 was observed in up to 10% of normal individuals 38,39 . As it is difficult to monitor the VA using the TCD under the same condition, it can be difficult to detect RLS. As in this study using BA, it is easy to fix the probe through the headset and there is no need to consider VA hypoplasia. It was much easier to use as the sonographer did not have to fix the probe manually during TCD study in the suboccipital window. In this study, MESs were measured only on the supine position, our method allows us to investigate further patients with prone and sitting postures. Some studies have shown www.nature.com/scientificreports/ that the upright sitting position is the best position for detecting RLS 40,41 . This was relatively more comfortable and easily achieved. Therefore, we chose the BA instead of VA and compared it to the MCA in our study. Previous randomized controlled trials did not reveal any statistical significance between PFO closure and medical therapy for reducing stroke rates [42][43][44] . However, some recent studies have suggested that PFO closure is more effective than medical therapy for reducing recurrent stroke rates [4][5][6][7] . The benefits observed in recent trials following PFO closure is probably related to several factors involved in the selection of patients with PFO features, such as a large shunt size, those with a more severe presentation or presence of an atrial septal aneurysm 4,6,7 . Therefore, it is important to evaluate the size and the severity of the PFO with RLS. TCDs could be an alternative to TEE for assessing the severity of PFO and a useful follow-up test for patients after PFO closure. We believed that TCD monitoring for BA can also be a method for quantitatively assessing the severity of PFO.
Our study has several limitations. First, the MCA and BA were compared only in provoked RLS. A comparative study of the MCA and BA is needed in constant RLS. Second, this study has not proven that ischemic lesions are located predominantly in the vertebrobasilar circulation, especially in patients with provoked RLS 29 . A previous study showed that the time taken for appearance of the bubble after administering the injection was higher in the vertebrobasilar circulation 31 . In future studies, the number of bubbles in addition to the time of occurrence of the bubbles needs to be considered while comparing the MCA and BA. Third, the BA may also present with hypoplasia and have lower blood flow than that in the carotid circulation. This may be associated with bilateral VAH or bilateral fetal-type posterior circle of Willis. We identified 21 patients with BA hypoplasia (diameter < 2 mm) in our study. Even with the basilar artery hypoplasia, the weighted kappa index was 0.896 when categorized using the SLS in Tier 3. Additionally, there were no patients in whom the BA was not detected using the TCD.
In conclusion, the suboccipital window using BA in RLS diagnosis is as valuable as the temporal window and sonographers can easily perform TCD monitoring in detecting RLS in patients with PTW.

Methods
Participants. This retrospective study was reviewed and approved by the Institutional Review Board (IRB) of Ajou University Hospital, Suwon, Korea (AJIRB-MED-MDB-21-349). The Declaration of Helsinki, was followed. The requirement for informed consent was waived by IRB of Ajou University Hospital owing to the retrospective nature of this study. The stepwise TCD method was our routine protocol for detecting RLS in our hospital.
Patients tested using agitated saline TCD from December, 2015 to September, 2020 were recruited in this retrospective study. A total of 816 patients with headaches, TGA, WMHI, TIA, or stroke were tested using the TCD at Ajou University Hospital. The study excluded patients who did not reveal any MESs on TCD while in the resting state or on performing the Valsalva maneuver (n = 472). Therefore, a total of 344 patients were included in the study. Figure 2 shows the stepwise TCD method for the recruitment of this study population.
The TCD procedure. The tests were performed in the following manner. An evaluation tool with three tiers was instituted to evaluate MESs during TCD monitoring. In tier 1, the TCD test was performed in both the right and left MCAs (or internal carotid arteries [ICAs] with a PTW) in the resting state. In tier 2, the TCD test was carried out after administering agitated saline for > 3 min while performing the Valsalva maneuver. In tier 3, if one or more MESs were detected from tier 1 or 2, MCA on one side (termed as the index MCA [iMCA]) was tested along with BA for > 3 min after performing the Valsalva maneuver. However, patients with bilateral PTWs were excluded 41 in tier 3 (Fig. 2).
Patients were advised to limit coffee and tobacco during 4 h before the examination, and also recommended to stop anticholinergic drugs during 48 h in advance to the TCD examination. In addition, to reduce the effect of the drug, patients who scheduled an examination were tested in the morning. The patients were asked to exhale into a manometer and maintain a pressure of 40 mmHg for 10 s to trigger the Valsalva maneuver. The iMCA was defined as MCA on one side, either right or left that had better visibility or detectability for MES. Microbubbles were provoked by administering 9 mL of 0.9% saline mixed with 1 mL of the patient's blood and 1 mL of air. All 344 patients underwent TCD examinations as described in tiers 1 and 2. TCD as defined in tier 3, was performed in a total of 303 patients after excluding those with bilateral PTWs (n = 41).
A TCD examination was performed by sonographers using the PMD 150 M (Spencer Technologies, Seattle, WA) or TC8080 (Pioneer TC 8080; Viasys Healthcare, Madison, WI) with a 2 MHz probe. MES were recorded using probes mounted on the modified headset (Phifix; Astron, Suwon, Republic of Korea) that was monitored for both MCA and BA at the same time in detecting RLS using TCD (Fig. 3). There are two probes on the headset. One probe is placed on the temporal window to detect the MCA and the other probe is placed on the suboccipital window to detect the BA (Fig. 3). When the iMCA and BA were monitored simultaneously for MES using TCD, the patients were instructed to wear the headset and perform the Valsalva maneuver. During examination, the patient was placed in a supine position, and when measuring MES as BA and MCA using TCD, the patient turned his head to a comfortable direction and the test was performed. MES was defined as visible (short duration, high-intensity and unidirectional) and audible (click, chirp, or whistles) signals in the Doppler flow spectrum. All the MES were reviewed by two neurologists and sonographers of more than 10 years' experience. Statistical analysis. Statistical analyses were performed using the R Statistical Software (version 3.6.3.; R Foundation for Statistical Computing, Vienna, Austria). To verify the consistency of the agreement tests in various tiers, the analysis was performed in two ways. First, the weighted kappa index was used to compare the agreement between the right and left MCA, and the agreement between the iMCA and BA after classification based on the International Consensus Criteria and SLS in tiers 2 and 3. Second, MES identified quantitatively in the left and right MCA were verified and compared using the intraclass correlation coefficient (ICC) in tier 2. Similarly, the microbubbles identified in the iMCA and BA were verified, and compared using the ICC in tier 3. P < 0.05 was considered statistically significant.