The 5-HT_1A receptor is one of the best characterised receptors in the serotonergic family; its role as an inhibitory autoreceptor in the raphe nuclei and the possible implications of this role for the treatment of depression and anxiety with serotonin reuptake inhibitors are well known (King et al. 2008).

Many of the 5-HT_1A radioligands were based on WAY-100635 (N-[2-[4-(2-methoxyphenyl)piperazin-1-yl]ethyl]-N-pyridin-2-ylcyclohexanecarboxamide), which in its carbonyl–¹¹C-labelled form is widely used for 5-HT_1A receptor imaging. Currently, four radioligands are used for PET studies of the 5-HT_1A receptor in humans: [carbonyl–¹¹C]WAY-100635, [¹⁸F]MPPF, [¹⁸F]FCWAY and [¹¹C]CUMI-101. [carbonyl–¹¹C]WAY-100635 is so far the most widely used 5-HT_1A receptor radioligand. It has a high target to background ratio, but its fast systemic metabolism makes it difficult to quantify accurately. [¹⁸F]MPPF has the advantage of the longer lived ¹⁸F-label, and it also selectively labels the 5-HT_1A receptors with a low non-specific binding. Its major disadvantage is its low brain uptake. [¹⁸F]FCWAY is rarely used, probably because of issues with defluorination of the parent compound which leads to high bone uptake of radioactivity (Ryu et al. 2007). [¹¹C]CUMI-101 is a high-affinity 5-HT_1A (partial) agonist radioligand that displays high specific binding and seems suitable for imaging the high-affinity site within the human brain (Milak et al. 2008; Pinborg et al. 2012).

The 5-HT_1B receptor is of particular interest in relation to obesity (Halford et al. 2007) and migraine (Tfelt-Hansen 2012). No less than two PET radiotracers for imaging the 5-HT_1B receptor have recently been introduced for use in humans: [¹¹C]AZ10419369 and [¹¹C]P943. [¹¹C]AZ10419369 (5-methyl-8-(4-[¹¹C]methyl-piperazin-1-yl)-4-oxo-4H-chromene-2-carboxylic acid(4-morpholin-4-yl-phenyl)-amide) is a 5-HT_1B partial agonist used in humans (Varnas et al. 2011); it has a slow systemic metabolism. The high-affinity 5-HT_1B antagonist radioligand, [¹¹C]P943 (R-1-[4-(2-methoxy-isopropyl)-phenyl]-3-[2-(4-methyl-piperazin-1-yl)benzyl]-pyrrolidin-2-one), also shows good properties for quantification in the human brain (Gallezot et al. 2010). For both radiotracers, it seems that cerebellum constitutes an acceptable reference region, and from studies in non-human primates, there is some evidence that the radiotracers are sensitive to displacement by endogenous 5-HT (Finnema et al. 2010; Ridler et al. 2011). The first normative healthy individual studies are now emerging (Savli et al. 2012), and an 8 % decline in 5-HT_1B receptor binding per decade, but no gender-related differences, has been reported (Matuskey et al. 2012).

5-HT_2A receptors are of interest for many reasons: they are a primary target of psychedelic compounds (Lee and Roth 2012), contribute to the efficacy of antipsychotic medications and are involved in the aetiology or treatment of various psychiatric disorders (Leysen 2004).

Several different radioligands for imaging the brain 5-HT_2A receptor have successfully been used in human PET studies, e.g. [¹⁸F]altanserin; [¹⁸F]setoperone; [¹¹C]NMSP; [¹¹C]MDL 100,907; and [¹¹C]Cimbi-36. In addition, the SPECT-tracer [¹²³I]-R91150 is used in imaging studies, but the radiotracer displays a lower signal-to-noise ratio compared to the available PET radioligands.

Despite its lipophilic radiometabolite, [¹⁸F]altanserin continues to be the most widely used PET radiotracer. One of the reasons for this is that ¹⁸F-labelling – because of the longer half-life – facilitates the application of a bolus/infusion paradigm that in turn enables subtraction of the lipophilic brain metabolite (Pinborg et al. 2003). Imaging data obtained from [¹⁸F]altanserin binding in the human brain are highly reproducible (Haugbol et al. 2007), and the large number of publications based on this radioligand provides a convenient reference for new findings.

[¹¹C]NMSP (N-methylspiperone, 8-[4-(4-fluorophenyl)-4-oxobutyl]-2-methyl-4-phenyl-2,4,8-triazaspiro[4.5]decan-1-one) is a dual D₂/5-HT₂ receptor ligand. As [¹⁸F]setoperone, NMSP has high affinity for both receptors, but since the density of D₂ receptors is low and that of 5-HT₂ receptors is high in the cortical brain regions, then the majority of specific binding in neocortex is due to 5-HT₂ receptor binding. For subcortical brain regions, the reverse is true (Lyon et al. 1986). [¹¹C]NMSP has been used primarily as an imaging tool to visualise D₂ receptor binding in the striatum but was also used in early PET studies to estimate changes in cortical 5-HT₂ receptor binding in, for example, aging (Wong et al. 1984). At that time, the more selective PET radioligand, [¹⁸F]altanserin, had not yet been fully developed.

The radioligand [¹⁸F]setoperone (6-[2-[4-(4-[¹⁸F]fluorobenzoyl)piperidin-1-yl]ethyl]-7-methyl-2,3-dihydro-[1,3]thiazolo[3,2-a]pyrimidin-5-one) is less selective than [¹⁸F]altanserin and is being used less and less, perhaps due to its lack of specificity for the 5-HT_2A receptors.

Following the positive validation studies, [¹⁸F]altanserin was used to determine changes in 5-HT_2A receptor density in relation to aging (Erritzoe et al. 2009), and a database of 5-HT_2A receptor binding in healthy volunteers was published (Adams et al. 2004), and it has been reported that binding in healthy subjects correlates with the body mass index (Erritzoe et al. 2009) but does not vary with gender (Frokjaer et al. 2009). Furthermore, twin studies have shown that [¹⁸F]altanserin binding is strongly genetically determined (Pinborg et al. 2008).

Based on in vitro data (Kristiansen et al. 2005), [¹¹C]MDL 100,907 is more selective for the 5-HT_2A receptor than [¹⁸F]altanserin, but it is questionable if this has any practical implications because of the scarcity of 5-HT_2B and to some extent 5-HT_2C receptors, and in any instance, [¹¹C]MDL 100,907 is much less widely used as a 5-HT_2A receptor radioligand than [¹⁸F]altanserin. The reason for this could be that arterial blood sampling is required for correct quantification of [¹¹C]MDL 100,907 (Talbot et al. 2012). Another promising radioligand is [¹¹C]Cimbi-36 (Ettrup et al. 2011) which is the first 5-HT_2A receptor agonist radioligand that has proven successful in humans (personal communication).

To date, it is questionable if 5-HT_2B receptors are expressed in the brain in sufficient amounts to allow for imaging. This is not the case for 5-HT_2C receptors, but currently, all radiolabelled 5-HT_2C receptor ligands have shared pharmacology with other receptors. No radiotracers have been developed for SPECT or PET imaging of 5-HT_2B or 5-HT_2C receptors.

Despite a number of research centres undertaking a concerted effort to develop 5-HT₃-selective PET and SPECT tracers, it seems that the very discrete localisation and relatively low levels of 5-HT₃ receptors that are localised with highest densities in the brainstem (Parker et al. 1996) make it a very difficult target to image in vivo.

5-HT₄ receptors are involved in learning and memory and are potential targets for the treatment of Alzheimer’s disease (for a review, see (Bockaert et al. 2004)). So far, there is one PET ligand, [¹¹C]SB207145 (8-amino-7-chloro-(N-[¹¹C]methyl-4-piperidylmethyl)-1,4-benzodioxan-5-carboxylate), that has been successfully evaluated in humans.

The 5-HT₄ receptor antagonist SB207145 was initially radiolabelled with C-11 (Gee et al. 2008) and evaluated for its potential as a PET radioligand for 5-HT₄ imaging. [¹¹C]SB207145 was subsequently successfully quantified for use in human brain studies (Marner et al. 2009). In this study, a comprehensive quantification of the binding of [¹¹C]SB207145 to cerebral 5-HT₄ receptors in the human brain in vivo was further provided. Distribution volumes and binding potentials of [¹¹C]SB207145 showed good test–retest reproducibility and time stability. The blocking study with piboserod confirmed that the cerebellum is a suitable reference region devoid of specific binding and that reference tissue models apply. Subsequently, it was shown that [¹¹C]SB207145 is not to any significant degree displaceable by acutely increased levels of endogenous 5-HT (Marner et al. 2010), but cautions need to be taken to ensure that the injected mass of SB207145 does not exceed 4.5 μg (Madsen et al. 2011c). That is, [¹¹C]SB207145 can be used for quantitative PET measurements of 5-HT₄ receptors in the human brain, and normative data on age- and sex-related variations have been published (Madsen et al. 2011b).

The 5-HT₅ receptor has two subtypes, the 5-HT_5A and the 5-HT_5B receptors. The 5-HT_5A receptor has been identified in the human brain, but the 5-HT_5B receptor is not expressed in humans because the coding sequence is interrupted by stop codons (Nelson 2004). The 5-HT_5A receptor shows a particularly high presence in raphe and other brainstem and pons nuclei (Volk et al. 2010). There are no available radioligands for either of the 5-HT₅ receptors.

5-HT₆ receptors are found exclusively in the CNS and are predominantly expressed in the striatum, limbic system and cortex (Woolley et al. 2004). They are of particular interest because of their involvement in learning and memory (King et al. 2008). So far, only one non-selective radioligand for PET imaging of the 5-HT₆ receptor has made its way into humans, namely, [¹¹C]GSK215083 (3-[(3-fluorophenyl)sulfonyl]-8-(4-[¹¹C]methyl-1-piperazinyl)quinoline (Parker et al. 2012). GSK215083 has high affinity for 5-HT₆ (in vitro K _i, 0.16 nM) but also has high 5-HT_2A affinity (in vitro K _i, 0.79 nM). However, the differential localisation of 5-HT_2A and 5-HT₆ receptors (predominantly cortical and striatal, respectively) combined with the ~5-fold difference in affinity means that discrimination between these two receptor types can be done to some extent.

Several research centres are attempting to develop 5-HT₇-selective PET and SPECT tracers, but so far, these efforts have not resulted in a radiotracer that has been taken into humans.

SERT is the serotonin transporter and it receives considerable interests, not the least due to the success of its inhibitors in the treatment of depression and anxiety disorders. Several PET and SPECT ligands have been developed for this purpose and a detailed review of SERT imaging by PET and SPECT can be found in Huang et al. (2010).

Initially, images of SERT in the human brain came from the non-selective cocaine derivative SPECT ligand [¹²³I]β-CIT and later with the selective but kinetically irreversible PET ligand [¹¹C](+)McN5652. Today, the most successful line of SERT radioligands are those developed from diarylsulfides such as [¹²³I]ADAM, and especially [¹¹C]MADAM, and [¹¹C]DASB.

The currently most appropriate SPECT radioligand for SERT imaging is [¹²³I]ADAM (2-((2-((dimethylamino)methyl)phenyl)thio)-5-iodophenylamine), which is potent, selective and has a high target to background ratio in human studies (Newberg et al. 2004). Quantification of the SERT binding with [¹²³I]ADAM SPECT is most often done with a ratio method, based on data acquired from 200 to 240 min. This has, however, been shown to overestimate the specific binding by about 10 %. The most reliable outcome is based on a 0–120 min SPECT acquisition followed by Logan non-invasive modelling (Frokjaer et al. 2008).

The first selective PET radioligand for imaging SERT in the human brain was [¹¹C](+)McN5652 (trans-1,2,3,5,6,10-β-hexahydro-6-[4-(methylthio)phenyl]pyrrolo-[2,1-a]isoquinoline) (Szabo et al. 1995). Its use has been limited by a relatively low target to background ratio in vivo, as well as a slow brain uptake and irreversible kinetics that complicate quantification in high binding regions (Frankle et al. 2004). Despite this, [¹¹C](+)McN5652 PET has been used to investigate SERT binding in humans in more than 20 published studies (Pubmed November 2012).

[¹¹C]DASB (3-amino-4-(2-dimethylamino-methyl-phenylsulfanyl)-benzonitrile) is one of a series of ¹¹C-labelled arylthiobenzylamines developed by Wilson and Houle (1999). [¹¹C]DASB displays good selectivity and high affinity and is to date the most widely used radiotracer for in vivo imaging of SERT, with more than 150 hits on Pubmed. [¹¹C]DASB PET can be quantified with reference tissue models (Ginovart et al. 2001); this is the most frequently used way of quantifying [¹¹C]DASB. Furthermore, test–retest data showed high reproducibility and reliability (Kim et al. 2006). [¹¹C]DASB PET normative data are available (Erritzoe et al. 2009, 2010), and apart from a multitude of patient studies, it has also been used to measure SERT occupancy at clinical doses of selective reuptake inhibitors, e.g. citalopram and paroxetine (Meyer et al. 2004).

[¹¹C]MADAM (N,N-dimethyl-2-(2-amino-4-methylphenylthio)-benzylamine) is yet another PET radioligand that has made its way into humans (Lundberg et al. 2005). When done several weeks apart, test–retest reproducibility of [¹¹C]MADAM is excellent (Lundberg et al. 2006). As for [¹¹C]DASB, [¹¹C]MADAM has been used to estimate relative SERT occupancy of citalopram and escitalopram (Lundberg et al. 2007) and to investigate the relationship between 5-HT_1A and SERT binding in healthy young men and women (Henningsson et al. 2009; Jovanovic et al. 2008). [¹¹C]DASB and [¹¹C]MADAM thus seem comparable in terms of their pharmacological and kinetic profile.

The primary reason why new radioligands are still being developed is that the low density of SERT-binding sites in the neocortex is challenging for accurate measurement and it would also be valuable to have an ¹⁸F-labelled SERT radioligand.

Using [¹⁸F]MPPF, Kepe and co-workers found significant reductions in 5-HT_1A receptor densities in both the hippocampus and raphe nuclei in patients with mild cognitive impairment (MCI) (24 %) and AD patients (49 %) (Kepe et al. 2006). They interpreted their findings as secondary to neocortical degeneration of synapses and neurons, because a positive association of 5-HT_1A receptor binding with glucose metabolism was found in the hippocampus. Further, 5-HT_1A receptor binding was negatively correlated with [¹⁸F]FDDNP, a marker of tau pathology (Kepe et al. 2006). Interestingly, using the same tracer, Truchot and co-workers found a biphasic change in 5-HT_1A receptor binding in the hippocampus, parahippocampus and inferior temporal cortex with upregulation in MCI and a marked reduction (approx. 50 %) in AD (Truchot et al. 2007) which may suggest a compensatory upregulation due to lower serotonin levels in the pre-dementia stage of AD. Following this upregulation, 5-HT_1A receptor binding may eventually decrease in early to moderate stages of AD due to neurodegeneration, as suggested by Kepe et al. (2006).

Several studies have found widespread reduction in 5-HT_2A receptor binding in mild to moderate AD. Using PET and [¹⁸F]setoperone in patients with AD, Blin and colleagues found a 35–69 % reduction in temporoparietal cortical 5-HT₂ receptor binding (Blin et al. 1993). In a similar patient group (mean MMSE 20), but using [¹⁸F]altanserin, Meltzer and colleagues found reductions of approximately 30 % in the anterior cingulate, prefrontal and sensorimotor cortices (Meltzer et al. 1999). In two consecutive [¹⁸F]altanserin studies in MCI and AD patients, these findings were corroborated by widespread neocortical reductions in receptor binding (MCI: 22–29 % and AD: 28–39 %) (Hasselbalch et al. 2008; Marner et al. 2011). Importantly, all three [¹⁸F]altanserin studies employed correction for partial volume error due to atrophy, and atrophy did not explain the findings.

In contrast to the 5-HT_1A receptor findings above, the reductions in 5-HT_2A receptor are widespread and occur early in the course of the disease (Hasselbalch et al. 2008), but the reason for this reduction is largely unexplained. It is unlikely that loss of serotonergic neurons projecting to neocortex has any major impact on the postsynaptic 5-HT_2A receptor reductions, since reductions in 5-HT_2A receptors are not accompanied by similar reductions in presynaptic serotonin transporter binding (Marner et al. 2012). One plausible explanation for the diffused neocortical 5-HT_2A receptor reductions could be that local AD pathology, especially in the form of widespread beta-amyloid accumulation, is responsible for the reduction in 5-HT_2A receptors. Beta-amyloid accumulation has a spatial distribution similar to the reduction in 5-HT_2A receptor binding, whereas neurodegeneration in the form of tau accumulation follows a different pattern (Braak and Braak 1991). As an indirect support of the proposed beta-amyloid/5HT_2A receptor association, Marner and co-workers found a nonsignificant 5–10 % reduction in ¹⁸F-altanserin binding in neocortical regions in a 2-year follow-up study of 15 MCI patients (Marner et al. 2011). This reduction inversely mirrors the small increase in beta-amyloid accumulation found in early AD in most studies (Jack et al. 2010). Further, experimental studies have shown an association between increased beta-amyloid load and decreased 5-HT_2A receptor binding (Christensen et al. 2008; Holm et al. 2010). Thus, 5-HT_2A receptor may be especially sensitive to beta-amyloid pathology, but the nature of this interaction needs to be clarified in future studies.

The 5-HT₄ receptor is especially interesting in AD as it is both linked to memory function and beta-amyloid accumulation. There is ample evidence for pro-cognitive actions of agonists to the 5-HT₄ receptor. Thus, in animals, pre-task systemic injections of 5-HT₄ receptor agonists have shown to improve performance in a variety of memory tasks such as olfactory associative learning (Marchetti et al. 2000; Marchetti-Gauthier et al. 1997), place and object recognition (Lamirault and Simon 2001), the Morris water maze (Lelong et al. 2001), and matching-to-sample (Terry et al. 1998). In the only study in healthy humans, using [¹¹C]SB207145 and Rey’s Auditory Verbal Learning Test in healthy young subjects, Haahr and co-workers found significant negative associations between the immediate and delayed recall scores and hippocampal 5-HT₄ receptor binding (Haahr et al. 2013). This paradoxical finding was explained by upregulation of 5-HT₄ receptor levels to partially compensate for lower hippocampal 5-HT levels in subjects with poorer memory function (Haahr et al. 2013). In early AD, Madsen and co-workers found some support for this hypothesis: In mild stage AD patients, beta-amyloid imaging with [¹¹C]PiB and 5-HT₄ receptor imaging using [¹¹C]SB207145 was performed in the same subjects (Madsen et al. 2011d). PiB-positive individuals had 13 % higher 5-HT₄ receptor binding compared to PiB-negative individuals. The 5-HT₄ receptor binding was positively correlated to beta-amyloid burden and negatively to the MMSE score (Mini-Mental State Examination – a measure of global cognitive function) of the AD patients. These findings suggest that cerebral 5-HT₄ receptor upregulation starts at a preclinical stage of AD and continues while dementia is still at a mild stage, which contrasts other receptor subtypes. As in the study of healthy subjects mentioned above, it was speculated that the upregulation was a compensatory effect of decreased levels of interstitial 5-HT in early AD (Madsen et al. 2011d). Further, agonism of the 5-HT₄ receptor increases alpha-secretase activity and thus promotes non-amyloidogenic degradation of the amyloid precursor protein (APP) (Cachard-Chastel et al. 2007). Therefore, an upregulation of the 5-HT₄ receptor level in early AD may serve to counteract beta-amyloid accumulation, but further studies are needed to elucidate these rather speculative interactions.