Appearance time of blood in the brain as a possible indicator of oxygen extraction fraction: a feasibility study

Abstract

Background

Imaging examination of cerebral blood flow (CBF), oxygen extraction fraction (OEF), and metabolic rate of oxygen is crucial for understanding the normal functioning and pathophysiology of the brain. A recently developed method estimates the appearance time of cerebral blood (ATB) pixel-wise from the imaging examination of CBF alone. In this study, we aimed to test the potential of ATB as an indicator of OEF.

Results

We retrospectively reviewed patients (n = 62) with suspected cerebrovascular disorders including steno-occlusive disease who underwent positron emission tomography (PET) with ¹⁵O-labelled tracers. Regarding the generated OEF and ATB images, a visual assessment was performed to test the consistency of the elevated OEF and delayed ATB. The OEF and ATB values and the absolute differences between their ipsilateral and contralateral sides were extracted and obtained for the entire hemisphere and the middle, anterior, and posterior cerebral arterial regions. Consistency was observed in 52 PET scans (83.9%) in visual assessment. The OEF and ATB values were moderately correlated (r = 0.553, p < 0.001), and the differences between their ipsilateral and contralateral sides were weakly correlated (r = 0.276, p < 0.001).

Conclusion

Our results indicate the potential of ATB as an indicator of OEF.

Background

Imaging examination of cerebral blood flow (CBF), oxygen extraction fraction (OEF), and metabolic rate of oxygen (CMRO₂) is crucial to understanding the normal functioning and pathophysiology of the brain. The images can be obtained using positron emission tomography (PET) with ¹⁵O-labelled radiotracers, such as water (H₂¹⁵O), carbon dioxide (C¹⁵O₂), or oxygen (¹⁵O₂) [1]. Notably, such as bolus administration using autoradiography (ARG) [2, 3] and dynamic protocols [4, 5], have been developed to obtain these images for clinical assessment. The ARG protocol involves the integration method [2, 3, 5]. In contrast, the dynamic protocol involves optimisation procedures, such as a basis function method (BFM) [6], allowing the simultaneous estimation of multiple parameters, including the tracer’s uptake and washout rate constants and vascular volume. These protocols are useful for examining cerebrovascular disorders; however, the measurement procedure takes 30–60 min because of the multiple scans using multiple tracers. Furthermore, the procedure is labour-intensive because multiple PET scans are accompanied by tracer synthesis and administration. Therefore, obtaining a set of similar images using a single tracer administration would be helpful by reducing patients’ tolerance due to long examination time as well as laborious procedures for clinical staff.

A recent study developed a method for estimating the appearance time of cerebral blood (ATB) (or tracer), which is similar to arterial transit time (ATT), pixel-wise using a dynamic image from a single CBF examination scan data [7]. The study discovered that the region with reduced CBF was superimposable with the region with significant delay, suggesting that they were affected by occlusive lesions [7]. Notably, two other studies have also demonstrated that the delayed appearance of blood and OEF correlated significantly [8, 9]. These findings suggested that occluded areas could be identified by measuring ATB, and ATB could be a pathophysiological indicator in patients with occlusive lesions. This implies that the difference in ATB between hemispheres, which can be estimated only from CBF scan without ¹⁵O₂ data, could indicate an elevated OEF.

In this present study, we aimed to test the potential of ATB as an indicator of OEF by comparing OEF images with those of ATB in brain regions, generated using a pixel-wise method based on a previously developed method [7] from sets of C¹⁵O₂ and ¹⁵O₂ administration studies in patients with cerebrovascular disease including steno-occlusive disease.

Methods

Study design and participants

We retrospectively analysed images of patients with suspected cerebrovascular diseases, including steno-occlusive diseases, who underwent PET with C¹⁵O₂ and ¹⁵O₂ administration between April 2015 and June 2023. All patients were diagnosed with either normal or steno-occlusive disease based on magnetic resonance imaging (MRI), except four, who were diagnosed solely on cervical ultrasound. This study enrolled 62 patients (51 males and 11 females; mean age, 71.1 ranging from 21 to 86 years). Notably, three patients underwent multiple PET examinations; two underwent a second examination after carotid stenting, whereas one underwent three examinations during antiplatelet therapy and after carotid stenting. We analysed 49 scan data; only the first scan data in multiple examination cases was used, and 13 scan data were excluded due to bilateral stenosis or occlusion because it was not clear which of the bilateral lesions contributed more to elevate OEF and prolong ATB.

This study was conducted in accordance with the ethical standards of the Declaration of Helsinki of 1964 and its later amendments. The Ethics Committee of Kagawa University, Japan, approved this study (approval numbers: 2020 − 197) and waived the requirement for written informed consent due to the retrospective observational study design.

PET measurement protocol

All scans were performed in three-dimensional list mode using the Biograph mCT64-4R scanner (Siemens Healthcare, Erlangen, Germany) through a previously described rapid protocol [6], except for the scan duration and time lag between ¹⁵O₂ and C¹⁵O₂ administration (670 s and 420 s, respectively). After computed tomography (CT) scan for attenuation correction (120 kV, Effective mAs 40, 3-mm slice thickness, and 1.0-s rotation), emission scans were performed with ¹⁵O₂ (1.4 GBq/min) inhalation for 60 s and, after 420 s, with C¹⁵O₂ (1.4 GBq/min) inhalation for 60 s. During the emission scan, blood was manually sampled (~ 1 mL) through a catheter inserted in the right radial artery at 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 120, 140, 160, 180, 200, 230, 260, 290, 320, 350, 410, 470, 480, 490, 500, 510, 520, 530, 540, 560, 580, 600, 620, 640, and 670 s from the PET start time. Radioactivity concentration in the blood samples was measured using an ARC-8001 well counter (Aloka, Musashino, Tokyo, Japan).

Data processing

We summed the list-mode data acquired using PET into sinogram data with 54 frames, consisting of 10 × 10 + 5 × 20 + 9 × 30 + 20 × 5 + 10 × 10 s. Next, the tomographic images were reconstructed using the filtered back-projection method with a Hann filter involving attenuation correction using the CT data. A reconstructed image comprised 54 frames with a matrix size of 200 × 200 × 74 and a pixel size of 2.0 mm × 2.0 mm × 3.0 mm. The measured arterial blood time-activity curves were calibrated to the PET scanner, separated into oxygen and water content after correcting for delay and dispersion [10–12], and used as arterial input functions. Our in-house program, coded in C language compiled with GCC ver 8.3.1, computed CBF, OEF, and ATB images using the reconstructed dynamic image and the generated input functions [12], using a standard PC installed with GNU/Linux (fc29.x86_64 64 bit; CPU, Intel Core i7-3770 3.40 GHz; memory, 16 GB). The ATB was estimated by accounting for ratios in computed CBF values by applying the three different analyses, ARG and BFM, with and without fixing the partition coefficient. The time of the input function was adjusted so that the ratios were close to 1. Then, the adjusted time was chosen as the ATB [7]. Negative and positive ATB values indicated earlier and later appearances, respectively.

Data analysis

Visual assessment of OEF and ATB images

Based on visual assessment of OEF and ATB images, the following groups were defined: Group A, patients without elevated OEF and delayed ATB; Group B, patients without elevated OEF but with delayed ATB; Group C, patients with elevated OEF and delayed ATB; and Group D, patients with elevated OEF but without delayed ATB. In this procedure, we did not consider the overlapping of OEF elevation and ATB delay.

One board-certified nuclear medicine physician and one nuclear medicine physician independently visually evaluated the OEF and ATB images for elevated OEF and delayed ATB, respectively. Disagreements were resolved through consensus.

Quantitative assessment of OEF and ATB

For quantitative assessments, OEF and ATB values were extracted for the ipsilateral and contralateral sides of the region of interest (ROI) for the entire hemisphere and the middle (MCA), anterior (ACA) and posterior cerebral arterial (PCA) regions. ROI templates in NEUROSTATS were applied to set these regions [13]. For an automatic extraction of regional value, image templates for left and right hemisphere and cerebellum in NEUROSTATS were integrated and normalised to the PET-based templated image in SPM12 [14] and generated pixel size of 2 × 2 × 2 mm³. The templates of left and right MCA, ACA and PCA territory regions were also normalised simultaneously. Then, CBF, OEF, and ATB images from the same examination were normalised to the same brain template in SPM12. The ROI values of OEF and ATB in the normalised ACA, MCA and PCA ROIs were extracted. Figure 1 shows the representative ROIs on normalised images for ACA, MCA and PCA.

Fig. 1.

Representative ROIs on normalized image for ACA (green), MCA (yellow) and PCA (red)

The difference between ipsilateral and contralateral values of OEF and ATB were indicated as ΔOEF and ΔATB, respectively.

Statistical analysis

All statistical analyses were performed using the RStudio 2023.6.0.421 environment [15]. Data normality was assessed using the Shapiro-Wilk test. Statistical significance for the relationships between ATB and OEF and between ΔOEF and ΔATB were tested using Spearman’s rank correlation coefficient. The Kruskal–Wallis test incorporating the Steel-Dwass test was applied for group comparison. Statistical significance was set at p < 0.05.

Results

Visual assessment of OEF and ATB images

Of the 62 PET scan data, 35 (56.5%), 9 (14.5%), 17 (27.4%), and 1 (1.6%) were classified in Groups A, B, C, and D, respectively (Table 1). Visual assessment showed consistency in 52 scan data (83.9%; Groups A and C) and inconsistency in 10 scan data (16.1%; Groups B and D). Figures 2, 3, 4 and 5 show representative images of Groups A–D.

Table 1.

Clinical characteristics of and affected arteries in patients with suspected cerebrovascular diseases including steno-occlusive disease

	Group
	A	B	C	D
Number of patients	35	9	17	1
Sex, male/female	27/8	8/1	15/2	1/0
Age (years, mean ± SD)	70.3 ± 13.1	73.0 ± 8.9	71.5 ± 10.3	77
Number of PET scans	35	9	17	1
Affected artery
Bilateral ICA stenosis	7	–	4	–
Unilateral ICA stenosis	23	5	9	–
Unilateral ICA occlusion	–	2	1	1
Bilateral CCA stenosis	–	1	–	–
Unilateral MCA stenosis	2	1	2	–
Bilateral ICA and Unilateral stenosis	–	–	1	–
None	2	–	–	–

ICA, internal carotid artery; CCA, common carotid artery; MCA, middle cerebral artery The number of PET scans performed during or after treatment is shown in parentheses

Fig. 2.

PET images of a 41-year-old woman with no stenosis in the carotid or brain arteries in Group A. The CBF image shows no decrease (a). The OEF image shows no elevation (b). The ATB image shows no delay (c)

Fig. 3.

PET images of a 78-year-old man with stenosis in the bilateral common carotid arteries in Group B. The CBF image shows a decrease on the left side (a). The OEF image shows no elevation (b). The ATB image shows a delay on the left side (c). During antiplatelet therapy, the patient underwent a second PET scan due to stroke symptoms and was classified into Group C. The CBF image shows a decrease on the left side (d). The OEF image shows elevation on the left side (e). The ATB image shows a delay on the left side (f). The third PET scan was performed after stenting for the left common carotid artery, and the patient was classified into Group A. The CBF image shows a decrease on the left side (g). The OEF image shows no elevation (h). The ATB image shows no delay (i)

Fig. 4.

PET images of a 75-year-old man with stenosis in the left internal carotid artery in Group C. The CBF image shows a decrease on the left side (a). The OEF image shows elevation on the left side (b). The ATB image shows a delay on the left side (c)

Fig. 5.

PET images of a 77-year-old man with occlusion of the right internal carotid artery in Group D. The CBF image shows a decrease on the right side (a). The OEF image shows elevation on the right side (b). The ATB image shows no delay (c)

Quantitative assessment of OEF and ATB

Figure 6a shows the relationship between the OEF and ATB values in 372 regions (the values of ACA, MCA, and PCA regions extracted from each hemisphere of the cerebrum), and both values were moderately correlated (r = 0.553, p < 0.001). Figure 6b shows the relationship between ΔOEF and ΔATB in 186 (the values of ACA, MCA, and PCA regions) regions, and both were weakly correlated (r = 0.276, p < 0.001).

Fig. 6.

Relationship between OEF and ATB values (a), indicating moderate correlation (r = 0.553, p < 0.001), and between ΔOEF and ΔATB (b), indicating weak correlation (r = 0.276, p < 0.001)

OEF and ATB based on the group classification

Of the 47 PET scans for the patients with unilateral vascular lesions, 26 (55.3%), 8 (17.0%), 12 (25.5%), and 1 (2.1%) were in Groups A, B, C, and D, respectively. The Shapiro-Wilk test did not show normality in groups A, B, or C for ΔOEF and ΔATB (p < 0.05). For ΔOEF, the difference was significant between Groups A and C (p = 0.007) but not between Groups A and B (p = 0.727) or B and C (p = 0.153) (Fig. 7a). For ΔATB, the difference was significant between Groups A and B (p = 0.029) and between A and C (p = 0.020), but not between Groups B and C (p = 0.921) (Fig. 7b).

Fig. 7.

Relationships between group classifications based on visual assessment of ΔOEF (a) and ΔATB (b) images of the hemisphere. Box plots show the median, 25th percentile, and 75th percentile of data, with the minimum and maximum values represented by whiskers

Discussion

In this study, we tested the potential of ATB in the brain as an indicator of OEF. ATB images can be generated from a dynamic image from CBF examination alone within a reasonable computation time of < 5 min [7]. The visual assessment of OEF and ATB images was identical in 83.9% of PET scans performed in this study. We observed moderate correlations between OEF and ATB values and weak correlations between ΔOEF and ΔATB. These results suggest that ATB may be an indicator of elevated OEF.

OEF is a physiological parameter that indicates an increasing demand for oxygen for energy metabolism in the brain due to reduced blood supply [16]. ATB is the time difference between the appearance of blood in the pixel location and the applied input function adjusted to the mean time for the whole brain [7]; therefore, it depends on the pathway and blood velocity. Regarding the pathway, collateral circulation develops at a stenosis or an occlusive region. For velocity, a decreased cerebral circulatory reserve drives an increase in cerebral blood volume (CBV) with or without a change in CBF and an increase in the transit time results in decreased velocity. Therefore, an association of delayed ATB with elevated OEF may be expected, although ATB is a measure of blood appearance time and not a direct macroscopic cerebral blood transit time. This could result in regions with elevated OEF and delayed ATB, which do not always overlap (Fig. 3e). In this regard, the measure of mean transit time, namely the ratio between CBV and blood flow, should directly indicate transit time; however, this still requires another scan administered with a contrast agent, such as carbon monoxide. In this study, visual assessment showed delayed ATB in 94% of PET scans (17/18) with elevated OEF on the ipsilateral side and regression analysis showed a moderate correlation, r = 0.553. In the statistical analysis between groups determined from the visual assessment, ΔOEF and ΔATB differed significantly between groups. Our previous study showed that the correlation between OEF and ATB was less as r = 0.45 [7], which might be due to smaller ROIs, the little overlap between OEF elevation and ATB delay, and PET scanner performance. Previous studies showed the prolonged arterial arrival time (ATT) in internal carotid artery (ICA) territory due to ICA occlusion in arterial spin-labeling (ASL) MRI studies [17, 18]. Our findings are similar to that of Islam et al. which showed significant correlation between OEF and delay time [9]. They estimated the delay time pixel-wise by shifting the input function between − 3 and 3 s by 0.5 s step. This method is unknown, but it might take much time than the present method (~ 1.5 min) [7]. Takeuchi et al. compared OEF and ATT parameters, which can be obtained from the ASL technique in magnetic resonance scanning, similar to the present ATB [8], and showed a significant correlation. All of the facts suggest that classifying each group based on visual assessment of the ATB image is possible.

Group B included patients with delayed ATB but without an elevated OEF; we observed an inconsistency between the ATB and OEF in this group. ATB is an approximation of MTT as above, and MTT extends in two types of the pathophysiology of hemodynamic cerebral circulation disorders, namely stage I and II by Powers [19], a decrease of cerebral circulatory reserve derives an increase in CBV without change in CBF and an increase in MTT. Stage II, known as a misery perfusion, shows an increase in CBV with a decrease in CBF and an increase in MTT. The patients in Group B can be classified as Stage I, thus showing delayed ATB but without an elevated OEF. Those can be identified by a change in CBF. Only one of the nine patients underwent PET examination after 2 and 3 months (Fig. 3), showing decreased CBF in the ipsilateral region. The set of images obtained at 2 months showed elevated OEF and delayed ATB (Fig. 3), which disappeared at 3 months after treatment, suggesting that ATB delay with a decrease in CBF may predict an elevated OEF in the future.

In one patient in Group D, the OEF image of the right hemisphere showed an elevated OEF, whereas delayed ATB was absent. The patient had vascular anomalies (terminations in the right foetal posterior and left posterior inferior cerebellar arteries); however, no obvious collateral vessels were present. Why delayed ATB was absent despite an elevated OEF remains unclear.

ATB images might be generated using dynamic single-photon emission CT (SPECT) scanning. SPECT scanners are more widely used in hospitals than PET scanners, which would be more easily accessible for examination because of the tracer lifetime, such as ¹²³I with a half-life of 13 h. Possible limitations in analysing SPECT dynamic images are noise and longer frame durations. The issue may be resolved by incorporating artificial intelligence techniques, such as image denoising. However, further studies are required to confirm this hypothesis.

Conclusion

This study demonstrates the potential of ATB, a parameter derived using a dynamic image from CBF examination alone, as an indicator of OEF.

Abbreviations

ARG

Autoradiography

ASL

Arterial spin-labelling

ATB

Appearance time of cerebral blood

ATT

Arterial transit time

BFM

Basis function method

C¹⁵O₂

¹⁵O-labelled carbon dioxide

CBF

Cerebral blood flow

CMRO₂

Metabolic rate of oxygen

CT

Computed tomography

H₂¹⁵O

¹⁵O-labelled water

OEF

Oxygen extraction fraction

PET

Positron emission tomography

ROI

Region of interest

SPECT

Single-photon emission computed tomography

Author contributions

Concept and design: MM, NK, Y.Maeda, YY, and YN. Methodology: MM and NK. Data curation: MM, NK, KM, TN, Y.Manabe, Y.Maeda, YY and TH. Software: NK. Formal analysis: MM and NK. Writing–original draft: MM. Writing–review & editing: NK, YY, and YN. All authors read and approved the final manuscript.

Funding

This study was supported by a Grant-in-Aid for Scientific Research (KAKENHI) from the Japan Society for the Promotion of Science (grant number 19K17201; received by KM).

Data availability

The datasets used and analysed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the principles of Declaration of Helsinki. The Ethics Committee of Kagawa University, Japan, approved this study (approval number: 2020 − 197) and waived the requirement for written informed consent due to the retrospective observational study design.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.