Measurement of Cerebral Perfusion Indices from the Early Phase of [18F]MK6240 Dynamic Tau PET Imaging

Nicolas J. Guehl, Maëva Dhaynaut, Bernard Hanseeuw, Sung‐Hyun Moon, Cristina Lois, Emma G. Thibault, Jessie Fanglu Fu, Julie C. Price, Keith A. Johnson, Georges El Fakhri, Marc D. NormandinJournal of Nuclear Medicine 2023 · DOI

This study, published in the Journal of Nuclear Medicine, demonstrates that the early phase (first 5–10 minutes) of a dynamic [$^{18}$F]MK6240 tau PET scan can serve as a reliable surrogate for cerebral perfusion. By extracting this "free" blood-flow information, researchers can simultaneously assess both tau pathology and neurodegeneration in a single imaging session, streamlining the diagnostic process for Alzheimer’s disease.

TL;DR

The early signal from [$^{18}$F]MK6240 PET scans provides a robust index of cerebral perfusion that correlates highly with established amyloid-based perfusion measures. This allows a single PET tracer to provide a "two-for-one" biomarker, capturing both specific tau pathology and general neurodegenerative changes.

What Problem Does This Paper Address?

In the study of Alzheimer’s disease (AD), clinicians use the A/T/N framework: Amyloid, Tau, and Neurodegeneration. While PET scans are excellent for "A" and "T," measuring "N" (often via cerebral blood flow or perfusion) usually requires a separate scan, such as [$^{15}$O]H$_2$O PET or arterial spin labeling (ASL) MRI.

These additional scans increase costs, radiation exposure, and patient burden. Researchers have previously shown that the early "wash-in" phase of amyloid tracers can proxy for blood flow. This paper investigates whether [$^{18}$F]MK6240—a high-affinity second-generation tau tracer—can do the same, effectively turning a standard tau scan into a multi-modal diagnostic tool.

Methods

The researchers analyzed 49 subjects: 25 cognitively normal (CN), 17 with mild cognitive impairment (MCI), and 7 with AD.

  • Imaging: All subjects underwent dynamic [$^{18}$F]MK6240 (tau) and [$^{11}$C]PiB (amyloid) PET scans, plus structural MRI.
  • Modeling:
    • Invasive Gold Standard: In 24 subjects, arterial blood was sampled to create a metabolite-corrected input function. They used a 1-tissue-compartment model (1T2kv) to estimate $K_1$ (the rate of tracer transfer from plasma to brain).
  • Non-invasive Proxy: They used the Simplified Reference Tissue Model 2 (SRTM2) to calculate $R_1$, the relative delivery rate of the tracer compared to a reference region (the cerebellum).
  • Simulations: They performed computer simulations to ensure that the "stickiness" of the tracer (specific binding to tau) didn't bias the perfusion measurements during the first few minutes of the scan.
  • Validation: They compared the MK6240-derived $R_1$ values against $R_1$ values from [$^{11}$C]PiB, which is already an established surrogate for perfusion.

Datasets

The study utilized a primary clinical dataset of 49 human subjects recruited through Massachusetts General Hospital. This included:

  • CN (Cognitively Normal): 25 subjects.
  • MCI (Mild Cognitive Impairment): 17 subjects.
  • AD (Alzheimer’s Disease): 7 subjects.
  • Sub-cohort: 24 subjects provided arterial blood samples for kinetic modeling validation.

Key Insights and Ideas

  • The "Wash-in" Window: Perfusion information is most accurate in the first 5 minutes. Simulations showed that while specific binding to tau increases over time, its influence on the $K_1$ and $R_1$ estimates remains under 2% if only the first 5 minutes of data are used.
  • High Extraction Fraction: [$^{18}$F]MK6240 appears to have a higher extraction fraction (the percentage of tracer that crosses into the brain on the first pass) than amyloid tracers like PiB. This potentially makes it a more sensitive tool for detecting subtle changes in blood flow.
  • Regional Sensitivity: The study found that perfusion deficits were most prominent in the temporal and parietal cortices of AD patients—regions known to be heavily affected by neurodegeneration.

Key Results

  • Internal Consistency: Non-invasive $R_1$ estimates (SRTM2) agreed almost perfectly with blood-based $R_1$ calculations ($r = 0.99$; mean difference $0.024 \pm 0.027$).
  • Head-to-Head Validation: $R_1$ measurements from [$^{18}$F]MK6240 correlated strongly with those from [$^{11}$C]PiB ($r = 0.93$; mean difference $-0.001 \pm 0.068$).
  • Clinical Differentiation: Statistically significant differences in regional $R_1$ were observed between CN, MCI, and AD groups, particularly in the temporal and parietal cortices, mirroring the expected patterns of hypoperfusion in AD.
  • Stability: Perfusion indices remained stable whether 5 or 10 minutes of data were used, though 5 minutes minimized bias from specific binding.

Limitations and Future Work

  • Sample Size: The AD group was relatively small ($n=7$), which may limit the statistical power for some regional comparisons.
  • Scanner Variability: Data were collected on two different PET systems (ECAT EXACT HR+ and Discovery MI). While the authors calibrated the systems, this is a common real-world source of noise.
  • Missing "True" Gold Standard: While PiB is a good proxy, the authors note that a direct head-to-head comparison with [$^{15}$O]H$_2$O PET (the absolute gold standard for blood flow) is still needed to precisely determine the extraction fraction of MK6240.
  • Future Direction: This method opens the door for "dual-phase" imaging protocols where a single injection provides a perfusion map (early phase) and a tau pathology map (late phase).

Summary updated

Abstract

6-(fluoro-<sup>18</sup>F)-3-(1H-pyrrolo[2,3-c]pyridin-1-yl)isoquinolin-5-amine ([<sup>18</sup>F]MK6240) has high affinity and selectivity for hyperphosphorylated tau and readily crosses the blood-brain barrier. This study investigated whether the early phase of [<sup>18</sup>F]MK6240 can be used to provide a surrogate index of cerebral perfusion. <b>Methods:</b> Forty-nine subjects who were cognitively normal (CN), had mild cognitive impairment (MCI), or had Alzheimer's disease (AD) underwent paired dynamic [<sup>18</sup>F]MK6240 and [<sup>11</sup>C]Pittsburgh compound B (PiB) PET, as well as structural MRI to obtain anatomic information. Arterial blood samples were collected in a subset of 24 subjects for [<sup>18</sup>F]MK6240 scans to derive metabolite-corrected arterial input functions. Regional time-activity curves were extracted using atlases available in the Montreal Neurologic Institute template space and using FreeSurfer. The early phase of brain time-activity curves was analyzed using a 1-tissue-compartment model to obtain a robust estimate of the rate of transfer from plasma to brain tissue, <i>K</i> <sub>1</sub> (mL⋅cm<sup>-3</sup>⋅min<sup>-1</sup>), and the simplified reference tissue model 2 was investigated for noninvasive estimation of the relative delivery rate, <i>R</i> <sub>1</sub> (unitless). A head-to-head comparison with <i>R</i> <sub>1</sub> derived from [<sup>11</sup>C]PiB scans was performed. Grouped differences in <i>R</i> <sub>1</sub> were evaluated among CN, MCI, and AD subjects. <b>Results:</b> Regional <i>K</i> <sub>1</sub> values suggested a relatively high extraction fraction. <i>R</i> <sub>1</sub> estimated noninvasively from simplified reference tissue model 2 agreed well with <i>R</i> <sub>1</sub> calculated indirectly from the blood-based compartment modeling (<i>r</i> = 0.99; mean difference, 0.024 ± 0.027), suggesting that robust estimates were obtained. <i>R</i> <sub>1</sub> measurements obtained with [<sup>18</sup>F]MK6240 correlated strongly and overall agreed well with those obtained from [<sup>11</sup>C]PiB (<i>r</i> = 0.93; mean difference, -0.001 ± 0.068). Statistically significant differences were observed in regional <i>R</i> <sub>1</sub> measurements among CN, MCI, and AD subjects, notably in the temporal and parietal cortices. <b>Conclusion:</b> Our results provide evidence that the early phase of [<sup>18</sup>F]MK6240 images may be used to derive a useful index of cerebral perfusion. The early and late phases of a [<sup>18</sup>F]MK6240 dynamic acquisition may thus offer complementary information about the pathophysiologic mechanisms of the disease.
BibTeX
@article{guehl2023measurement,
  author = {Nicolas J. Guehl and Maëva Dhaynaut and Bernard Hanseeuw and Sung‐Hyun Moon and Cristina Lois and Emma G. Thibault and Jessie Fanglu Fu and Julie C. Price and Keith A. Johnson and Georges El Fakhri and Marc D. Normandin},
  title = {{Measurement of Cerebral Perfusion Indices from the Early Phase of [<sup>18</sup>F]MK6240 Dynamic Tau PET Imaging}},
  journal = {Journal of Nuclear Medicine},
  year = {2023},
  doi = {10.2967/jnumed.122.265072},
}
Figure