Principles of Functional MRI

Background: Recent studies have demonstrated that arterial spin labeling magnetic resonance imaging (ASL-MRI) and fluorodeoxyglucose positron emission tomography (FDG-PET) identify similar regional abnormalities and have comparable diagnostic accuracy in Alzheimer's disease (AD). The agreement between these modalities in the AD continuum, which is an important concept for early detection and disease monitoring, is yet unclear. Objective: We aimed to assess the ability of the cerebral blood flow (CBF) measures from ASL-MRI and cerebral metabolic rate for glucose (CMRgl) measures from FDG-PET to distinguish amyloid-β-positive (Aβ+) subjects in the AD continuum from healthy controls. Methods: The study included asymptomatic, cognitively normal (CN) controls and patients with early mild cognitive impairment (MCI), late MCI, and AD, all with significant levels of cortical Aβ based on their florbetapir PET scans to restrict the study to patients truly in the AD continuum. The discrimination power of each modality was based on the whole-brain patterns of CBF and CMRgl changes identified by partial least squares logistic regression, a multivariate analysis technique. Results: While CBF changes in the posterior inferior aspects of the brain and a pattern of CMRgl changes in the superior aspects of the brain including frontal and parietal regions best discriminated the Aβ+ subjects in the early disease stages from the Aβ- CN subjects, there was a greater agreement in the whole-brain patterns of CBF and CMRgl changes that best discriminated the Aβ+ subjects from the Aβ- CN subjects in the later disease stages. Despite the differences in the whole-brain patterns of CBF and CMRgl changes, the discriminative powers of both modalities were similar with statistically nonsignificant performance differences in sensitivity and specificity. Conclusion: The results comparing measurements of CBF to CMRgl add to previous reports that MRI-measured CBF has a similar diagnostic ability to detect AD as has FDG-PET. Our findings that CBF and CMRgl changes occur in different brain regions in Aβ+ subjects across the AD continuum compared with Aβ- CN subjects may be the result of methodological differences. Alternatively, these findings may signal alterations in neurovascular coupling which alter relationships between brain perfusion and glucose metabolism in the AD continuum.

Regulation of cerebral blood flow (CBF) during hypoglycemia (HYPO) is critical for maintaining adequate glucose delivery to the brain; however, specific changes in regional CBF during HYPO are not well defined. We measured CBF in 14 T1DM (mean ± SD age 31 ± 7 yrs, DM duration 16 ± 5 yrs, A1c 7.2 ± 1.0 %) and 13 healthy controls (CON; age 35 ± 11 yrs, A1c 5.2 ± 0.5 %) using 3.0 T MRI continuous arterial spin labeling during fasting baseline (BAS), a hyperinsulinemic (2 mU/kg/min) euglycemic clamp (EUG; 90 mg/dl), followed by a hypoglycemic clamp (HYPO; 50 mg/dl). Global and regional changes in CBF were measured in 15 regions of interest (ROI). At BAS, gray matter CBF was similar in CON and T1DM (58 ± 16 vs. 61 ± 11 ml/100 gm brain/min). No significant changes in CBF were observed from BAS to EUG in any brain region in either CON or T1DM. However, during HYPO in T1DM, CBF increased by 10% in gray matter and 16% in brainstem, both p <0.05. CBF also increased by 10% - 18% in specific ROIs involved in awareness of internal states, reward and emotion (orbitofrontal cortex, insula, anterior cingulate cortex, thalamus, and substantia nigra; p<0.01 to p<0.05). Minimal increases in CBF (1% - 6%) were observed in regions primarily involved in sensory or motor function, (occipital cortex, motor cortex, cerebellum, caudate, putamen) or in the posterior cingulate (all p > 0.30). In CON, no significant global or regional changes in CBF occurred during HYPO. In all subjects, higher A1c correlated with greater increases in CBF during HYPO (p<0.05). Conclusions: 1) Hyperinsulinemia per se does not alter CBF in T1DM or CON, 2) During HYPO, brain blood flow increases more in T1DM vs. CON, especially in regions involved in integration of internal awareness, reward and emotion, compared to sensory or motor function, and 3) Greater increases in CBF in subjects with higher A1c suggest that compensatory physiologic mechanisms enable subjects with chronic exposure to hyperglycemia to maintain adequate cerebral glucose supply when plasma glucose levels are suddenly reduced. Disclosure D. C. Simonson: Stock/Shareholder; Spouse/Partner; Phase V Technologies, Inc. N. R. Bolo: None. J. Theberge: None. G. Musen: None. A. M. Jacobson: None. Funding National Institutes of Health (DK073843)

How far is arterial spin labeling MRI from a clinical reality? Insights from arterial spin labeling comparative studies in Alzheimer's disease and other neurological disorders.

Reproducibility of BOLD, perfusion, and CMRO2 measurements with calibrated-BOLD fMRI

Functional MR imaging was performed in sixteen healthy human subjects measuring both regional cerebral blood flow (CBF) and blood oxygen level dependent (BOLD) signal when visual and auditory stimuli were presented to subjects in the presence or absence of anesthesia. During anesthesia, 0.25 mean alveolar concentration (MAC) sevoflurane was administrated. We found that low-dose sevoflurane decreased the task-induced changes in both BOLD and CBF. Within the visual and auditory regions of interest inspected, both baseline CBF and the task-induced changes in CBF decreased significantly during anesthesia. Low-dose sevoflurane significantly altered the task-induced CBF-BOLD coupling; for a unit change of CBF, a larger change in BOLD was observed in the anesthesia condition than in the anesthesia-free condition. Low-dose sevoflurane was also found to have significant impact on the spatial nonuniformity of the task-induced coupling. The alteration of task-induced CBF-BOLD coupling by low-dose sevoflurane introduces ambiguity to the direct interpretation of functional MRI (fMRI) data based on only one of the indirect measures-CBF or BOLD. Our observations also indicate that the manipulation of the brain with an anesthetic agent complicates the model-based quantitative interpretation of fMRI data, in which the relative task-induced changes in oxidative metabolism are calculated by means of a calibrated model given the relative changes in the indirect vascular measures, usually CBF and BOLD.

The most widely-used functional magnetic resonance imaging (fMRI) technique is based on the blood oxygenation level dependent (BOLD) effect, which requires at least partial uncoupling between cerebral blood flow (CBF) and oxygen consumption changes during increased mental activity. To compare BOLD and CBF effects during tasking, BOLD and flow-sensitive alternating inversion recovery (FAIR) images were acquired during visual stimulation with red goggles at a frequency of 8 Hz in an interleaved fashion. With the FAIR technique, absolute and relative CBF changes were determined. Relative oxygen consumption changes can be estimated using the BOLD and relative CBF changes. In gray matter areas in the visual cortex, absolute and relative CBF changes in humans during photic stimulation were 31 +/- 11 SD ml/100 g tissue/min and 43 +/- 16 SD % (n = 12), respectively, while the relative oxygen consumption change was close to zero. These findings agree extremely well with previous results using positron emission tomography. The BOLD signal change is not linearly correlated with the relative CBF increase across subjects and negatively correlates with the oxygen consumption change. Caution should be exercised when interpreting the BOLD percent change as a quantitative index of the CBF change, especially in inter-subject comparisons.

Detecting changes in human cerebral blood flow after acute exercise using arterial spin labeling: Implications for fMRI

Cerebrovascular reactivity measured with arterial spin labeling and blood oxygen level dependent techniques

Early postoperative changes in cerebral oxygen metabolism following neonatal cardiac surgery: Effects of surgical duration

Hypothermia effects on neurovascular coupling and cerebral metabolic rate of oxygen

Thirty years ago, a publication by Seisjo, entitled ‘Cerebral metabolic rate in hypercarbia—a controversy' summarized the current state of knowledge related to the influence of hypercapnia (increased pCO2) on cerebral metabolic rate of oxygen consumption (CMRO2) (Siesjo, 1980). Seisjo quoted a number of papers that provided controversial results ranging from reduced, to unchanged, to increased CMRO2 and concluded his paper with the following comment: ‘Since hypercarbia is a common pathophysiologic condition, its effects on cerebral metabolism and blood flow are of obvious concern to many scientists and clinicians, anesthesiologists included. It is disconcerting that 30 years after the first quantitative report (Kety and Schmidt, 1948), we still do not know how hypercarbia affects cerebral metabolic rate.' Today, 30 years later, scientists still debate this matter. In the current issue of this journal, Jain et al (2011) report on the measurement of global CMRO2 in human brain during rest and hypercapnia. The authors developed a magnetic resonance imaging (MRI)-based technique that allows simultaneous measurements of cerebral blood flow (CBF) and venous blood oxygenation level (SvO2) with a temporal resolution of 30 seconds. Using the widely accepted principle that CMRO2 is proportional to the product of CBF and arterial–venous difference in blood oxygenation level, Jain et al found decreases in CMRO2 during mild hypercapnia (5% inspired CO2) that were small and not significant. At the same time, they found significant increases in CBF and SvO2—a result that is in agreement with practically all previous studies. A similar result was recently reported by Chen and Pike (2010), whose findings also suggested no significant change in global CMRO2 with mild hypercapnia. However, in another recently published study, Xu et al (2011) reported that mild hypercapnia resulted in a 13% suppression of CMRO2. This result is similar to (for example) previously published data in rhesus monkey (Kliefoth et al, 1979), but is opposite to reported increases in CMRO2 in rats (Horvath et al, 1994). Some of this inconsistency in results between human and animal studies can be attributed to the different physiological conditions under which the experiments were performed. However, the data in Jain et al (2011), Chen and Pike (2010) and Xu et al (2011) were obtained in normal awake humans, and one can only speculate that differences should be attributed to differences in experimental techniques. Substantial progress has been made in developing in vivo methods to study brain metabolism and hemodynamics since the initial publication (Kety and Schmidt, 1948), and the paper by Jain et al contributes significantly to this development. Yet, the accuracy of this and other methods must be further scrutinized before we can put narrow-enough error bars on the results to provide an accurate answer to the old question: How does hypercapnia influence brain metabolism? One more compelling reason to seek a definitive answer to this question lies in current attempts to use hypercapnia to tease out the effects of changes in blood flow and brain metabolism during functional brain activation (so-called calibrated functional MRI (Davis et al, 1998; Kim et al, 1999)). We hope that the paper by Jain et al will help in resolving this controversy as well.

Holding up under pressure: a complex interplay between cerebral blood flow and ventilatory responses to alterations in carbon dioxide.

To determine the short-term effects of antihypertensive therapy on cerebral blood flow (CBF). Prospective, observational study. A university-affiliated teaching hospital. Twenty-four patients (age range 53-85 yrs) with chronic hypertension, nine of whom had carotid artery occlusive disease (CAOD). The CBF (xenon-133 inhalation technique) and blood pressure were measured before and at 60 minutes after administration of antihypertensive therapy. Age was inversely related to the change in CBF in patients with CAOD (p < 0.01). In all patients, the change in CBF after taking antihypertensive drugs was significantly inversely associated with baseline CBF (p < 0.01). Changes in regional CBF, measured by asymmetry scores, were significantly greater in patients with CAOD than in those without CAOD (p < 0.05). Elderly patients with occlusive extracranial cerebrovascular disease are at risk of drug-induced changes in both mean and regional CBF, and may benefit from a CBF assessment before being prescribed antihypertensive therapy.

Functional neuroimaging is most commonly performed using the blood-oxygenation-level-dependent (BOLD) approach, which is sensitive to changes in cerebral blood flow (CBF), cerebral blood volume (CBV), and the cerebral metabolic rate of oxygen (CMRO(2)). However, the precise mechanism by which neuronal activity elicits a hemodynamic response remains controversial. Here, visual stimulation (14 secs flashing checkerboard) and breath-hold (4 secs exhale+14 secs breath hold) experiments were performed in alternating sequence on healthy volunteers using BOLD, CBV-weighted, and CBF-weighted fMRI. After visual stimulation, the BOLD signal persisted for 33+/-5 secs (n=9) and was biphasic with a negative component (undershoot), whereas CBV and CBF returned to baseline without an undershoot at 20+/-5 and 20+/-3 secs, respectively. After breath hold, the BOLD signal returned to baseline (23+/-7 secs) at the same time (P>0.05) as CBV (21+/-6 secs) and CBF (18+/-3 secs), without a poststimulus undershoot. These data suggest that the BOLD undershoot after visual activation reflects a persistent increase in CMRO(2). These observations support the view that CBV and CBF responses elicited by neuronal activation are not necessarily coupled to local tissue metabolism.

Calibrated fMRI in the medial temporal lobe during a memory-encoding task