The amyloid cascade is the leading current explanation of the etiology of Alzheimer's disease (AD) (Hardy and Selkoe, 2002; Jack et al., 2010). The hypothesis holds that amyloid-β (Aβ) has a primary role in the biochemical, histological, and pathological changes that happen in the brain when AD evolves. In this process, the deposition of Aβ is considered an early event, which implies that biomarkers of Aβ may detect the presence of disease at the earliest stages of onset (McKhann et al., 2011).

Modification of the amyloid-binding dye thioflavin-T made non-invasive PET imaging possible with the labeled form of the Pittsburgh compound B, the tracer [¹¹C]PiB. Although other tracers currently are under evaluation (Wong et al., 2013), [¹¹C]PiB is the most extensively examined marker of Aβ in human studies with PET (Klunk et al., 2004; Price et al., 2005; Mintun et al., 2006; Lockhart et al., 2007; Ikonomovic et al., 2008). Compared to cognitively intact controls, AD patients exhibit greater [¹¹C]PiB retention in areas known to have high Aβ deposits such as the frontal and parietal cortices, whereas brain areas relatively unaffected by Aβ pathology, such as the cerebellum, show equivalently modest [¹¹C]PiB retention (Klunk et al., 2004; Price et al., 2005; Ikonomovic et al., 2008). However, the specific transport and binding properties of [¹¹C]PiB remain uncertain when the tracer passes from the circulation into the tissue of the brain.

Klunk et al. (2004) compared the brain uptake of [¹¹C]PiB with the uptake of fluorine-18-labeled fluorodeoxyglucose ([¹⁸F]FDG), a marker of glucose consumption in brain of patients with Alzheimer's disease. The authors found that [¹¹C]PiB retention at late times correlated inversely with cerebral glucose metabolism traced by [¹⁸F]FDG. In the brain of patients with Aβ deposits, compared to patients without Aβ deposits, significant [¹¹C]PiB retention correlated well with variable levels of amyloid deposition, as determined by later postmortem analysis (Klunk et al., 2004). However, at earlier circulation times, Rostomian et al. (2011) found comparable [¹¹C]PiB and [¹⁸F]FDG uptakes in patients, consistent with substantial initial clearance of [¹¹C]PiB from brains of patients, as also found in brains of healthy control subjects studied by Blomqvist et al. (2008). These results indicate that the phases of initial uptake and subsequent retention of [¹¹C]PiB differ in human brain, with initial deposit of [¹¹C]PiB limited by delivery and hence also by cerebral blood flow, coupled to metabolism, and later retention limited by binding to tissue components such as Aβ (Forsberg et al., 2012).

To determine the extent to which the initial uptake of [¹¹C]PiB is flow-limited, and to convert the clearance to a measure of permeability, it is necessary to compute the permeability-surface area products of microvascular transport of [¹¹C]PiB in different brain regions, as well as the fractions of extracted [¹¹C]PiB. To obtain these values, we must know the values of the cerebral blood flow (CBF) that delivers [¹¹C]PiB to brain tissue. Here, to test the hypothesis of initial flow-limited brain uptake of [¹¹C]PiB in humans, we determined CBF values and [¹¹C]PiB uptake in five patients with AD, and six age-matched healthy volunteers, by means of PET.

In previous publications, we determined the variability of cerebral blood flow rates in patients with Alzheimer's disease (Rodell et al., 2012), and we determined the accumulation of tracer [¹¹C]PiB in regions of the brains of the same patients in relation to cerebral blood flow, compared to healthy control subjects.

The issue resolved in the present report is the question of whether [¹¹C]PiB uptake into the brains of these patients and healthy volunteers during the early phase of blood-brain transfer is flow-limited such that the initial uptake of [¹¹C]PiB can be used reliably as a marker of blood flow. The evidence from the study supports the claim that [¹¹C]PiB uptake during the first 4 min of circulation is not completely flow-limited, consistent with substantial but not unlimited permeability in the cerebral vasculature of patients with Alzheimer's disease and healthy control subjects. The PS-products were less than the values of cerebral blood flow, establishing unidirectional extraction fraction values at 0.63 or less. A tracer with this property yields flow values that are lower than the true values, and the flow fluctuations are attenuated. Thus, the estimated PS-products may not be sufficiently high to secure unidirectional clearance rates that always differ significantly when blood flow rates decline.

The search for tracers with initial flow-limited uptake sufficient to accurately estimate blood flow rates has been ongoing for a long time. The issue always concerned the duration of the flow-limited accumulation, and the determination of the moment when the accumulation effectively turns from flow-limitation to significant limitation by diffusion, binding, or metabolism. This transition happens for all tracers, given enough time. The use of the marker to trace blood flow therefore is limited to the period before the transition.

The determination of the brain's microvascular PS-products rests on the measure of the marker's initial clearance, here symbolized by the term K₁, as shown in Figure 7, displaying both the initial idealized clearance, as derived from the actual uptake of [¹¹C]PiB into the frontal lobe of a healthy volunteer. Judging from this graph, [¹¹C]PiB uptake into brain tissue is effectively unidirectional for only a few minutes after administration the tracer. The accurate determination of this initial clearance is aided by regression to the measured monoexponential approach to a steady-state volume of distribution.

In previous attempts to obtain blood-brain clearance measures of [¹¹C]PiB, the authors generally concluded that clearances and hence uptake rates are “high” but values of permeability-surface areas were not reported (Blomqvist et al., 2008; Rostomian et al., 2011), although Blomqvist et al. did obtain unidirectional clearance estimates. Blomqvist et al. (2008) assumed continuous steady-state accumulation of [¹¹C]PiB in brain tissue, for which there is little evidence. The issue reflects the uncertainty of the number of compartments of [¹¹C]PiB distribution in brain tissue that separately and significantly can be identified kinetically. The present analysis gave no evidence of continuous steady-state accumulation of [¹¹C]PiB or its labeled metabolites in brain tissue and instead reflected the kinetics of a single tissue compartment.

There is no strongly correlated decline of brain energy metabolism and cerebral blood flow rates in healthy aging (Aanerud et al., 2012). Neurodegeneration, on the other hand, such as Alzheimer's disease, generally presents with declines of both variables in specific parts of the brain, most likely driven by the reduced energy turnover in these same parts of the brain of demented individuals (Johannsen et al., 2000; Rodell et al., 2012). The unidirectional blood-brain clearance term K₁ is an expression of the unidirectional influx of [¹¹C]PiB and as such is related to the blood flow rate by means of equation (6). Therefore, we expected to find a correlation between blood flow estimates and [¹¹C]PiB accumulation in the first tomography frames of [¹¹C]PiB uptake. This expectation was supported by the inverse correlation between blood flow and [¹¹C]PiB accumulation in the last tomography frames, as evidence of the presence of Aβ plaques in the tissue. Indeed, blood flow and [¹¹C]PiB accumulation at that time were highly correlated (R² = 0.999; P = 0.021).

To convert estimates of unidirectional clearance to blood flow values, the extraction fraction must be known, according to the equation, F = K₁/E. In the present study, the unidirectional extraction fraction averaged 0.53 for all subjects and regions, but differed significantly between the patients and the healthy control subjects. In a companion study, we used the unidirectional clearance and estimates of PS products to make accurate predictions of blood flow that correlated well with actual CBF measures. With an average extraction fraction of 0.53, the values of blood flow calculated from the unidirectional clearance respectively over- and underestimated the CBF values determined with labeled water in patients and healthy control subjects and did not differ significantly on average.

No non-gaseous cerebral blood flow indicator is fully extracted. Depending on the distribution volume, a quantity of tracer always remains in the vascular bed, implying that there is no perfect single compartment modeling that fully describes the kinetics (Ohta et al., 1992, 1996). The conversion of non-gaseous tracer uptake therefore depends on the validity of the extraction fraction used to calculate the blood flow from the unidirectional blood-brain clearance, determined kinetically or graphically. The extraction fractions for typical non-gaseous blood flow indicators vary from more than 90% for ethanol, iodoantipyrine, and butanol, to less than 90% for antipyrine and water (Gjedde et al., 1975, 1980; Sakurada et al., 1978). The lower the extraction fraction, the more the calculation itself depends on the unknown blood flow.

The relations among the clearance and extraction estimates and the measured flow values are illustrated in Figure 8. The similar regional PS-products in the two groups explain the numerically attenuated and hence insignificant differences between the extraction fractions. Thus, the comparatively low PS-products are consistent with extraction fractions that do not change very significantly with changing blood flow rates, as shown in Figure 9.

The information that can be gleaned from the [¹¹C]PiB uptake clearly depends on when the uptake is determined. After 10 min of circulation, among the two groups, the retention of [¹¹C]PiB and its possible metabolites differed significantly for the frontal lobe and cerebral cortex, but not for the gray matter of the cerebellum. In the cortical regions, the uptake of [¹¹C]PiB in the patients with Alzheimer's disease exceeded that of the healthy controls by 40–50%, indicating [¹¹C]PiB retention at this time. For the cerebellum, the accumulation of [¹¹C]PiB did not differ between the two groups, as shown in Figure 10. From 1 to 4 min of circulation, the brain uptake of [¹¹C]PiB relative to healthy controls correlated significantly with the differences of cerebral blood flow (R² = 0.999; P = 0.021), indicating the partially flow-limited uptake of [¹¹C]PiB during this period shown in Figure 10.

To claim that a tracer with the properties of [¹¹C]PiB with an average extraction fraction of 0.53 ± 0.10 (SD), is useful as a flow indicator at the very least requires a careful assessment of the variability of the results and the potential errors that may arise. In the present study, the correlation coefficient of measured and calculated flow values was R² = 0.20 with a significance of P < 0.01 and a slope of 58% of the line of identity for a total of 33 comparisons (3 brain regions from 11 subjects each).