Venous plasma and blood data in PET

Quantitative analysis of PET data requires precise measurements of radioactivity concentration in the tissue, and, for accurate normalization of the tissue concentration, an input function (IF). The concentration of the unchanged (non-metabolized) tracer in arterial plasma (or blood, depending on the radioligand) is the gold standard IF in quantitative PET data analysis. Arterial input function (AIF) is often referred to as plasma time-activity curve (PTAC) or blood time-activity curve (BTAC). Arterial catheterization for blood sampling in PET is safe (Jons et al., 1997; Everett et al., 2009) and reliable. The invasiveness of arterial puncture has stimulated development of alternative methods for obtaining the input function, but also led to using semiquantitative analysis methods that do not require input function, for example SUV. For certain radioligands and tissues, arterial input function (AIF) can be substituted by a reference tissue as input function in quantitative models. Alternative methods must be compared to and validated against results from analysis applying arterial input function (Riaño Barros et al., 2014; McGinnity et al., 2017). Image-derived or population-based input function methods often require a few blood samples for accurate scaling of the blood or plasma curve. For most radioligands, metabolite correction is necessary, requiring arterial blood sampling even when BTAC can otherwise be derived from dynamic PET image. If the first-pass extraction of the radioligand is low, then it may be possible to sample venous or arterialized venous blood instead of the arterial blood.

Venous sampling

Venous blood sampling is less invasive and labour-intensive procedure than arterial sampling. However, venous concentrations of a certain compound or radioligand are never consistent with the arterial concentrations. Arterial plasma concentrations are independent of the site of blood sampling, while the venous concentrations are dependent of the clearance of compound in the vascular bed and thus the sampling site (Chiou, 1989a and 1989b; Gumbleton et al., 1994; Peters & Jamar, 1998; Zanotti-Fregonara et al., 2011) Venous curve is also delayed and dispersed compared to the arterial curve.

Venous blood samples are usually taken from intravenous cannula placed in antecubital vein of the arm. The antecubital vein concentration depends on the fraction of arm blood flow contributed by muscle, adipose tissue, connective tissue (including bone), skin, and arteriovenous shunts in the skin, and the distribution volume and kinetics of radioligand in each of these arm tissues. In physiologically based pharmacokinetic (PBPK) models, assuming a set of “standard arm” parameters, arterial concentration of drugs can be estimated via deconvolution of the antecubital vein concentration (Levitt, 2004; Musther et al., 2015). This approach would not be produce the required accuracy in PET studies, and could incline to circular reasoning. Extraction and perfusion in the arm and in the region of interest (for example brain) are different, and therefore arteriovenous difference and venous concentration are different, too; this prevent not only using venous BTAC as input function, but also its usage in correction of the vascular volume fraction. High inter-subject variability in arteriovenous concentration differences are common in both human and animal studies (Zanotti-Fregonara et al., 2011 and 2014).

After a bolus administration, equilibrium between artery and vein is present only transiently when the net uptake of the radioligand in the tissue is zero. The time and duration of this phase varies among the radioligands and sampling sites, or may be absent for the duration of the PET scan (Zanotti-Fregonara et al., 2011). During the arterial peak the net extraction into the tissue is very high and venous concentrations are usually much lower than arterial concentration, while at later times venous concentrations may exceed arterial concentrations. After intravenous bolus, peak arterial plasma concentrations are usually achieved at ∼0.5 min, while peak venous plasma concentrations occur at ∼1-5 min after injection (Chiou, 1989a). Models for A-V difference have been developed for pharmacokinetic and PET studies.

Intravenous catheter placed for the administration of the radioligand can be used for venous blood sampling (Hoekstra et al., 2000; Ponto et al., 2002). Initial part of the venous BTAC is lower than that of arterial BTAC. This can lead to large underestimation of AUC and subsequent overestimation of Ki (Nishizawa et al., 1998). In [18F]FDG studies the overestimation in myocardial Ki was ∼10±10%, half of which was cancelled out in calculation of MRglu because glucose concentration in arterialized samples was underestimated, too (van der Weerdt et al., 2002). Arterial whole blood radioactivity concentration can be measured with an automatic blood sampling system to record the curve precisely after radiotracer bolus infusion; this is not possible in venous or arterialized venous sampling. In the brain studies, venous sinuses can be used to extract image-derived venous BTAC (Wahl et al., 1999), which may need to corrected for PVE and plasma-to-blood ratio, but may still provide input function closer to the arterial input than the arterialized venous blood sampling (Wahl et al., 1999).

The replacement of arterial sampling by venous sampling from a resting arm or hand has been investigated in [18F]FDG studies (Reivich et al., 1979).

The late blood sample that is needed for scaling image-derived, population-based, or model-based input function can be taken from the injection cannula, if it has been proven that venous blood equals arterial blood after a certain time. Venous samples could even be used in [15O]H2 PET to scale input function derived by measuring the radioactivity of exhaled air (Koeppe et al., 1985). In [18F]FDG studies, at least 40 min after administration the activity concentration in venous samples equals arterial blood (Wakita et al., 2000), or even after 10 min (Chen et al., 1998). Venous and arterial blood activities were similar at 60 min for [11C]ABP688, at 90 min for [11C]CUMI-101, and at 100 min for [11C]DASB (Bartlett et al., 2018). For [18F]altanserin the venous TAC reaches the level of arterial TAC at ∼25 min (B&oumldvarsson & Mørkebjerg, 2006). In a α-[11C]methyl-tryptophan study the venous curves coincided with arterial ones after ∼20 min (Nishizawa et al., 1998). For [carbonyl-11C]WAY-100635 venous samples were used after 10 min (Hahn et el (2012)). [18F]FLT metabolite analysis from venous samples tends to slightly overestimate the fraction of the unchanged tracer (Visvikis et al., 2004). Parent tracer fraction was similar in arterial and venous blood for [11C]CUMI-101 and [11C]DASB, but for [11C]ABP688 the parent fraction was higher in venous plasma until 60 min (Bartlett et al., 2018).

Arterialized venous sampling

Venous blood sampling has been improved by sampling blood from a hand heated to about 44 ℃ to “arterialize” the venous blood. Heating the hand leads to vascular dilatation, which increases blood flow as a cooling mechanism, but does not increase metabolism (Goldschmidt & Light, 1925; Weiner & Cooper, 1955), resulting in venous blood becoming more similar to arterial blood (Phelps et al., 1979; Brooks et al., 1989; Jensen and Heiling, 1991; Copeland et al., 1992; Liu et al., 1992; van der Weerdt et al., 2002). It should be noted that immersing hand in water at 44 ℃ will lead to a heat input of >50 W, which imposes substantial thermoregulatory load (Frayn & Macdonald, 1992). Brooks et al. (1989) proposed heating the hand only to 39 ℃ with heating pad.

Significant differences in substrate concentrations between arterial and arterialized venous samples may remain (Green et al., 1990; Nauck et al., 1992). Oxygen saturation from the sampled blood must be measured, and it should be >95% before accepting that the blood is adequately arterialized (Bramley et al., 1991). Although veins drain hand regions with different tissue compositions, the choice of the vein did not influence the oxygen saturation of the blood (Bramley et al., 1991). The difference between arterial and arterialized venous blood is often much smaller with other substrates than O2, including glucose, fatty acids, and amino acids (Frayn & Macdonald, 1992; Brooks et al., 1989), but after bolus administration of radiotracers marked differences in the initial distribution phases can be expected. Therefore arterialized venous blood sampling should be used only when absolute quantification is not necessary (clinical use), or is validated against arterial blood sampling, or arterial-venous concentration differences can be modelled.

The arterialized venous BTAC peak is delayed and more dispersed as compared to arterial BTAC, which may need to be taken into account, although for precise analysis these effects need to be corrected for AIF, too. In a [18F]fallypride, the peak became ∼30 s later (Mukherjee et al., 2002).


Usage of arterialized venous samples must be validated by comparison to the arterial samples for each radioligand; for example: in brain MAO-A activity measurements using [11C]befloxatone arterialized venous blood samples did not provide satisfactory input function for Logan plot analysis (Zanotti-Fregonara et al., 2013), but could be used for scaling a population-based input function, provided an optimal arterialization was achieved (Zanotti-Fregonara et al., 2014). In assessment of D2/D3 receptors using [18F]fallypride, arterialized venous sampling was used in place of arterial sampling in some subjects (Mukherjee et al., 2002). Arterialized venous samples were validated for brain [18F]FDG PET studies by Phelps et al. (1979), and for heart [18F]FDG PET studies by van der Weerdt et al. (2002).

Higher inter-subject variability has been observed in results obtained using arterialized venous blood. In a myocardial [18F]FDG study the increased variability could not be explained by different degrees of arterialization (van der Weerdt et al., 2002), although poorly performed arterialization certainly would lead to that. An additional cause of variability may be the inter-subject arteriovenous concentration differences, common in both human and animal studies (Zanotti-Fregonara et al., 2011 and 2014).

Even though arterialized venous blood sampling would provide similar BTACs as arterial sampling, it may still not be a valid method for taking samples for metabolite analysis. For example, for [18F]fallypride, the fractions of radioactive metabolites is marginally higher in arterialized venous blood than in arterial blood (Mukherjee et al., 2002). For [11C]acetate and [11C]palmitate venous blood samples can be used for the metabolite analysis (Ng et al., 2013).

Table 1. Positive and negative sides of arterial, heated-hand venous, and venous blood sampling.
Arterial Arterialized venous Venous
Accuracy Gold standard Moderate Low
Validation Gold standard Required Required
Success rate Very high High High
Risk for complications Low Very low Very low
Contraindications Common No No
Automatic sampling Possible Not possible Not possible

See also:


Brooks DC, Black PR, Arcangeli MA, Aoki TT, Wilmore DW. The heated dorsal hand vein: an alternative arterial sampling site. JPEN J Parenter Enteral Nutr. 1989; 13(1): 102-105. doi: 10.1177/0148607189013001102.

Copeland KC, Kenney FA, Nair KS. Heated dorsal hand vein sampling for metabolic studies: a reappraisal. Am J Physiol. 1992; 263(5 Pt 1): E1010-E1014. doi: 10.1152/ajpendo.1992.263.5.E1010.

Frayn KN, Macdonald IA. Methodological considerations in arterialization of venous blood. Clin Chem. 1992; 38(2): 316-317. PMID: 1541023.

Green JH, Ellis FR, Shallcross TM, Bramley PN. Invalidity of hand heating as a method to arterialize venous blood. Clin Chem. 1990; 36(5): 719-722 PMID: 2337978.

Jensen MD, Heiling VJ. Heated hand vein blood is satisfactory for measurements during free fatty acid kinetic studies. Metabolism 1991; 40(4): 406-409. doi: 10.1016/0026-0495(91)90152-M.

Jons PH, Ernst M, Hankerson J, Hardy K, Zametkin AJ. Follow-up of radial arterial catheterization for positron emission tomography studies. Hum Brain Mapp. 1997; 5(2): 119-123. doi: 10.1002/(SICI)1097-0193(1997)5:2<119::AID-HBM5>3.0.CO;2-6.

Levitt DG. Physiologically based pharmacokinetic modeling of arterial - antecubital vein concentration difference. BMC Clin Pharmacol. 2004; 4:2. doi: 10.1186/1472-6904-4-2.

Liu D, Moberg E, Kollind M, Lins PE, Adamson U, Macdonald IA. Arterial, arterialized venous, venous and capillary blood glucose measurements in normal man during hyperinsulinaemic euglycaemia and hypoglycaemia. Diabetologia 1992; 35(3): 287-290. doi: 10.1007/BF00400932.

Nauck MA, Liess H, Siegel EG, Niedmann PD, Creutzfeldt W. Critical evaluation of the ‘heated-hand-technique’ for obtaining ‘arterialized’ venous blood: incomplete arterialization and alterations in glucagon responses. Clin Physiol. 1992; 12(5): 537-552. doi: 10.1111/j.1475-097X.1992.tb00357.x.

Nauck MA, Blietz RW, Qualmann C. Comparison of hyperinsulinaemic clamp experiments using venous, ‘arterialized’ venous or capillary euglycaemia. Clin Physiol. 1996; 16(6): 589-602. doi: 10.1111/j.1475-097X.1996.tb00736.x.

Ponto LLB, Graham MM, Richmond JC, Ward CA, Clermont DA, Schmitt BA, Clark J, Conklin A, Weldon L, Watkins GL, Madsen MT, Hichwa RD. Contamination levels in blood samples drawn from the injection intravenous line. Mol Imaging Biol. 2002; 4(6): 410-414. doi: 10.1016/S1536-1632(02)00121-X.

van der Weerdt AP, Klein LJ, Visser CA, Visser FC, Lammertsma AA. Use of arterialised venous instead of arterial blood for measurement of myocardial glucose metabolism during euglycaemic-hyperinsulinaemic clamping. Eur J Nucl Med Mol Imaging 2002; 29(5): 663-669. doi: 10.1007/s00259-002-0772-y.

Tags: , , , , ,

Updated at: 2019-02-03
Created at: 2018-11-11
Written by: Vesa Oikonen