Quantification of MAO-B activity with [11C]L-deprenyl-D2

[11C]L-deprenyl-D2 (DEP-D, [11C]DED) is used to measure the activity of monoamine oxidase B (MAO-B) in the brain. MAO-B is present in the outer mitochondrial membrane occurring in the brain predominantly in glial cells and in serotonergic neurons (Fowler et al., 2005). The expression of MAO-B in glial cells is thought to be predominantly, if not exclusively, within astrocytes (Ekblom et al., 1994; Gulyás et al, 2011), but not microglia. Astrocytosis is observed in the early stage of AD pathology, seen as increased [11C]DED uptake (Rodriguez-Vieitez et al., 2016).

MAO-B oxidises amines (for example dopamine and noradrenaline) from both endogenous and exogenous sources. MAO-B is selectively and irreversibly inhibited by L-deprenyl (selegiline). When the enzyme-substrate complex is formed, the rate-limiting step of MAO-B catalyzed oxidation creates a highly reactive intermediate, which forms a covalent bond with the enzyme, thus irreversibly inactivating it (Fowler et al., 2005). When L-deprenyl is labelled with 11C, the active enzyme MAO-B becomes labelled, and it can be imaged in vivo with PET.

Compartmental model

The brain concentrations of [11C]L-deprenyl-D2 peak at about 5 min after injection, and after a washout phase the concentrations reach a plateau about 30 min after injection (Fowler et al., 1995). An irreversible three-compartment model (two-tissue compartment model) can be applied to the time-activity curves (TACs) of labelled L-deprenyl in the brain and plasma to estimate MAO-B activity in the brain. In the model, K1 represents the plasma-to-organ transfer constant, k2 is the transfer rate of radiotracer from organ back to plasma, and k3 describes the rate of binding to MAO B. Under the PET study conditions, k3 is proportional to the functionally active free enzyme concentration. Because binding is irreversible, it is assumed that k4=0. From these model constants, K1 and k2 are dependent on blood flow, but k3 is not. Due to the high extraction of L-deprenyl, K1 is dominated by blood flow instead of capillary permeability (Fowler et al., 1988 and 1995). Also the net influx rate Ki (Ki=K1×k3/(k2+k3)) is dependent on blood flow (Lammertsma et al., 1991), as well as the more directly radioligand uptake related parameters, like SUV.

The very high rate of binding of labelled L-deprenyl to MAO-B creates difficulties in applying the compartmental model. The estimates of k2 and k3 tend to be highly correlated, and the rate of radioligand binding (k3) and delivery (K1) are hard to separate (Fowler et al., 2005). Especially in the brain regions of high MAO-B concentrations (basal ganglia, thalamus and cingulate gyrus), and in the elderly people who have reduced blood flow, the rate limiting step is the radioligand delivery instead of binding (Fowler et al., 1993). To reduce the problem with correlating k2 and k3 estimates, a combination model parameter λk3 has been introduced as an index of MAO-B activity; λ is the K1 over k2 ratio (Fowler et al., 1993 and 1995). K1 and k2 are both dependent on the blood flow, but λ and λk3 are independent of blood flow. Because this index contains the ratio k3/k2, the effect of their (positive) correlation is expected to be smaller. Reproducibility in test-retest studies was improved by using λk3 instead of k3 (Logan et al., 2000).

For instance, Fowler et al 1988 and 1995), Logan et al (2000), and Sturm et al., 2017 have estimated the three-compartment model parameters using traditional nonlinear least squares approach. In a addition, graphical analysis methods have been used.

Deuterium substitution

The rate-limiting step of MAO-B catalyzed oxidation involves cleavage of a certain carbonhydrogen bond in L-deprenyl. A carbon-deuterium bond is more difficult to cleave than the carbon-hydrogen bond, which leads to reduced rate of reaction when this hydrogen is substituted with deuterium (so called deuterium isotope effect ). Because the very high binding rate of [11C]L-deprenyl has been found to be problematic in quantification of MAO-B activity, the deuterium-substituted L-deprenyl, [11C]L-deprenyl-D2 is therefore preferred as PET radiopharmaceutical (Fowler et al., 1988, 1995, and 2004).

Confounding factors


Unlike most enzymes, MAO-B activity (λk3) increases clearly with normal ageing, which is accompanied by decreasing blood flow (decreasing K1) (Fowler et al., 1997). Therefore, the age must be controlled in the PET studies of MAO-B activity. If the age effect is studied, the measured index of MAO-B activity must not be flow dependent. The age-related and disease-associated increase of MAO-B has been attributed to neuron loss and gliosis (increase in glial cells).


As described above, the tissue uptake of [11C]L-deprenyl-D2 and especially [11C]L-deprenyl is strongly dependent on blood flow. Therefore, to quantification of MAO-B activity, the compartment model and a perfusion-independent model parameter or index, like k3 or λk3 must be used. Otherwise, for example, the increase of MAO-B activity in epileptogenic region might be underestimated because of subsequently reduced blood flow.


MAO-B is highly and variably inhibited in smokers (Fowler et al., 2003). An overnight abstinence for smokers does not produce any recovery of MAO-B activity. However, smoking a single cigarette does not produce a measurable decrease in MAO-B activity in non-smokers (Fowler et al., 2003).

Decreased K1 in later scans

In the test-retest setting, Logan et al (2000) noticed a decrease of K1 in the second scan (-7.7 ± 13.2%), although the decrease was not statistically significant (n=5). This may be caused by familiarization with the PET procedure and decreased anxiety (Logan et al., 2000).

Corrections applied to the PET data

Blood volume correction

Fowler et al (1995) subtracted from the brain PET data an approximate 4% blood volume before analyzing the data using the three-compartment model or graphical analysis for irreversible systems. Lammertsma et al (1991) included the vascular volume fraction in the model equations, noting that whole blood curve must be used instead of (total) plasma curve.

Plasma protein binding

Free fraction in plasma was 6.0% (Fowler et al., 2004). Considering the high K1 estimates in the brain, dissociation rate of the radiotracer from plasma protein may be high, so that most of protein bound radiotracer is also available for transport to the tissue. Plasma protein binding will affect K1/k2, and thus also λk3.

Blood-to-plasma transformation

For [11C]L-deprenyl, Lammertsma et al (1991) measured the blood-to-plasma ratio from discrete samples between 5 and 90 min. Based on in vitro experiments, they assumed that the plasma-to-blood ratio is 1.126 at the time of the arrival of the tracer in the blood, taken to be the time where the blood curve increased above 1% of the peak value (Lammertsma et al., 1991). They fitted a multi-exponential function to the ratios, determining the number of exponentials based on AIC and SC. The multi-exponential function was then used to calculate the total plasma curve from arterial blood curve which had been measured on-line. Fowler and Logan et al do not give details on this transformation in their publications.

Plasma metabolite correction

For [11C]L-deprenyl, Lammertsma et al (1991) measured the fraction of plasma metabolites from four samples at 5, 10, 15 and 20 min, and fitted a single exponential function to the fractions, assuming no metabolites at time 0. The exponential function was used to calculate the concentration of unchanged radiopharmaceutical in the plasma. Fowler and Logan et al do not give details on this correction in their publications.

Time delay correction

Lammertsma et al (1991) corrected for the time delay between blood curve and whole brain PET data by including the delay as one of the fitted model parameters. The TACs of smaller ROIs were then fitted with the delay fixed to this value.

Graphical analysis

Fowler et al (1995) have used graphical analysis for irreversible systems to calculate the net influx rate Ki, using metabolite corrected plasma curve as input function. Ki was taken as an average of slopes of the Patlak plot between 6 and 45 min and 6 and 55 min (Fowler et al., 1995).

Graphical method without plasma sampling

Graphical method (Patlak plot) can be used to estimate the net MAO-B uptake using either metabolite corrected plasma or reference tissue. Kumlien et al (1995) used cerebellar grey matter as reference region in epilepsy study because if its high perfusion and relatively low MAO-B activity. Later, to compensate the significant amount of MAO-B in cerebellum, the cerebellar time-activity curves were multiplied by a mono-exponential function to correct the deviation of the plot from linearity (Bergström et al., 1998; Kumlien et al., 2001). However, cerebellar MAO-B activity is probably not constant between individuals, and even less so in MAO-B inhibition studies.

Note also that the results of graphical method are not independent from perfusion, although the blood flow effects may be smaller than with SUV method.

Linear method

Fowler et al (1997 and 1999) and Logan et al (2000) estimated the three-compartment model parameters K1 and λk3 in a procedure which involves also graphical analysis:

  1. Ki is calculated using the average of Patlak graphical analysis slopes between 6-45 min and 7-55 min
  2. K1 (and k2+k3) are estimated with bilinear regression using a modification of the general method of Blomqvist (1984), shown in equation 1. Logan et al (2000) describe this process in more detail: Several values for K1 were estimated by successively increasing the maximum time T from 5 to 18 min, because K1 is more sensitive to data at earlier time points; an average K1 was used in the next step.
  3. The λk3 is calculated by solving it from the equation that relates Ki to the three-compartment model parameters (Eq. 2).

The estimates of K1 and λk3 from this linear method correlated very well with the estimates from the nonlinear method, and no noise-induced bias was noticed in the linear method, and the repeatability was also similar (Logan et al, 2000). Note that the equations in the original articles from years 1997 and 1998 do not contain the necessary square brackets.

The first-pass extraction of [11C]L-deprenyl-D2 is high, and, assuming that it is 1, then K1 would equal plasma flow, which is about 40% of blood flow. Therefore, the K1 estimates by Logan et al ( 2000), ranging from 0.476 to 0.845, are quite high. The step 2) may lead to an overestimation of K1, if the blood volume in tissue is not considered: blood volume correction was not mentioned by Fowler et al (1997 and 1999) and Logan et al (2000).

Analysis method in TPC

Preprocessing of the plasma input

For a detailed description on preprocessing of blood data, read the report TPCMOD0033 Appendix A in PET intranet.

Make sure that you have all the necessary data files:

  1. On-line blood sampler data file
  2. Dynamic PET image file from this study
    This file is used to correct for possible start time mismatch between PET scanner and blood sampling. If you are sure that both were started simultaneously, then you do not need the dynamic image file yet.
  3. Count-rate curve
  4. Plasma curve from manual sampling
  5. Blood curve from manual sampling
  6. Plasma parent fractions

Calculate the DEP-D plasma and total blood TACs, corrected for time delay. The CLI script S:\petsoftware\bin\windows\DEP-D_input.vbs is unfortunately not functional any more.

Regional MAO B activity

Weights should be added to regional tissue data.

Fit irreversible two-tissue compartmental model to the regional data using PMOD or fitk3. Because of the relatively large variation in regional estimates for fitted vascular blood volume fraction (VB), we suggest that its value should be constrained in the fit, for example to 2.7%, that was the population average (n=15) in TPC. With fitk3 this can be done with option -Vb=2.7.

The most reliable model parameter for describing the activity of MAO B is λ×k3, where λ = K1/k2 (independent from perfusion), and k3 is proportional to the association constant kon (Fowler et al., 1995; Arakawa et al., 2017).

MAO B inhibition percentage

If two PET studies have been performed for each subject, one before (baseline) and one after (medication) dosage of a drug that occupies MAO B, the MAO B inhibition (EI) percentage can be computed as described here.

MAO B activity maps

Fowler and Logan have suggested a method to compute λ×k3 images.

To reduce the number of parameters to be estimated, a clustering algorithm which groups voxels with similar kinetics could be applied prior to the voxel analysis. In estimating the model parameters for each voxel, λ could then be fixed at the cluster value (Shumay et al., 2012).

See also:


Arakawa R, Stenkrona P, Takano A, Nag S, Maior RS, Halldin C. Test-retest reproducibility of [11C]L-deprenyl-D2 binding to MAO-B in the human brain. EJNMMI Res. 2017; 7:54. doi: 10.1186/s13550-017-0301-4.

Analysis of [11C]L-deprenyl-D2 (DEP-D) brain PET studies. TPCMOD0033.

Analysis of [11C]L-deprenyl-D2 (DEP-D) brain PET studies: processing of blood data. TPCMOD0033 Appendix A (in TPC intranet).

Analysis of [11C]L-deprenyl-D2 (DEP-D) brain PET studies: compartment model analysis of regional data. TPCMOD0033 Appendix C (in TPC intranet).

Fowler JS, Wang GJ, Logan J, Xie S, Volkow ND, MacGregor RR, Schlyer DJ, Pappas N, Alexoff DL, Patlak C, Wolf AP. Selective reduction of radiotracer trapping by deuterium substitution: comparison of carbon-11-L-deprenyl and carbon-11-deprenyl-D2 for MAO B mapping. J Nucl Med. 1995; 36(7): 1255-1262.

Hirvonen J, Kailajärvi M, Haltia T, Koskimies S, Någren K, Virsu P, Oikonen V, Sipilä H, Ruokoniemi P, Virtanen K, Scheinin M, Rinne JO. Assessment of MAO-B occupancy in the brain with PET and [11C]-L-deprenyl-D2: a dose-finding study with a novel MAO-B inhibitor, EVT 301. Clin Pharmacol Ther. 2009; 85(5): 506-512. doi: 10.1038/clpt.2008.241.

Sturm S, Forsberg A, Nave S, Stenkrona P, Seneca N, Varrone A, Comley RA, Fazio P, Jamois C, Nakao R, Ejduk Z, Al-Tawil N, Akenine U, Halldin C, Andreasen N, Ricci B. Positron emission tomography measurement of brain MAO-B inhibition in patients with Alzheimer's disease and elderly controls after oral administration of sembragiline. Eur J Nucl Med Mol Imaging 2017; 44: 382-391. doi: 10.1007/s00259-016-3510-6.

Tags: ,

Updated at: 2023-09-11
Created at: 2014-05-19
Written by: Vesa Oikonen