The Pedunculopontine Nucleus
One source of confusion about the pedunculopontine nucleus has been the variable terminology used. Most often it has been described as pedunculopontine nucleus, but the abbreviation PPTg (Pedunculopontine tegmental nucleus) has also been used in both the human and the rodent atlas of Paxinos and colleagues (Paxinos and Huang, 1995; Paxinos and Watson, 1998). The abbreviation Tg.pdpo (nucleus tegmenti pedunculopontinus) has been used in the human brain atlas of Schaltenbrand and Wahren (1977). Moreover, in some cases, confusion arose when the peripeduncular nucleus was used for the pedunculopontine nucleus (Mazzone et al., 2005, 2009; Stefani et al., 2007) since the peripeduncular nucleus is a distinct region separated from the pedunculopontine nucleus by the medial lemniscus (Yelnik, 2007; Zrinzo et al., 2007).
The pedunculopontine nucleus occupies a strategic position within the core of the brainstem tegmentum, at the crossroad of several major fibre systems. It receives direct input from the cerebral cortex and is reciprocally connected with several basal ganglia components and limbic areas, and gives further rise to ascending projections to the thalamus (Steininger et al., 1992; Matsumura et al., 2000). It is also directly connected to other motor nuclei in the brainstem and spinal cord (Garcia-Rill et al., 2001; Takakusaki et al., 2004, 2008). Additionally, clinical and experimental investigations have established that the basal ganglia and the cerebellum are concerned with complementary aspects of body posture and motor functions and interesting anatomical analogies have been made between these two sets of structures (Mehler and Nauta, 1974). The pedunculopontine nucleus may therefore act as an integrative interface between these structures. It is also possible that the cerebellum may influence the function of various basal ganglia components and the entire thalamus via a relay in the pedunculopontine nucleus (Hazrati and Parent, 1992). The pedunculopontine nucleus is, together with the cuneiform and the subcuneiform nucleus, the major component of the mesencephalic locomotor region. Both the mesencephalic locomotor region and the mesencephalic reticular formation comprise a population of different types of neurons with different physiological functions that extend from the meso-diencephalic junction down to the medulla oblongata, just above the junction to the spinal cord, as first described by August Forel (1877). Therefore, it would be more appropriate to name specific areas within the mesencephalic locomotor region or mesencephalic reticular formation. Confusion arises with respect to the naming of the localization and implantation zones in deep brain stimulation. The terms mesencephalic reticular formation or mesencephalic locomotor region could be used to avoid over generalization as this nomenclature has been adopted in animal studies of lesioning or stimulation of the cuneiform or pedunculopontine nucleus (Steeves et al., 1975; Garcia-Rill et al., 1987b; Skinner et al., 1990).
The pedunculopontine nucleus forms a cluster of cells that is located in the caudal mesencephalic tegmentum, extending from the caudal border of the red nucleus to the parabrachial nucleus. In mammals, including humans, the pedunculopontine nucleus is bordered medially by fibres of the superior cerebellar peduncle and the peduncular decussation and laterally by the medial lemniscus. Rostrally, the anterior portion of the pedunculopontine nucleus contacts the substantia nigra and is adjacent to the retrorubral field; the most dorsal aspect of the pedunculopontine nucleus is bound caudally by the cuneiform and subcuneiform nuclei and ventrally by the pontine reticular formation. The most caudal pole of the cuneiform is adjacent to neurons of the locus coeruleus (Pahapill and Lozano 2000). The whole pedunculopontine nucleus is encircled by the mesencephalic locomotor region; however, its exact boundaries are disputable. Those that propound an extended localization of the pedunculopontine nucleus (Garcia-Rill, 1991) expand the nucleus from the posterior end of the substantia nigra to the laterodorsal tegmental nucleus, which is caudally adjacent to the pedunculopontine nucleus within the central grey matter. The laterodorsal tegmental nucleus is bordered medially by the dorsal tegmental nucleus of Gudden, rostrally by the dorsal raphe nucleus, and laterally by the locus coeruleus. Similar to the pedunculopontine nucleus, it contains cholinergic neurons intermingled with non-cholinergic (mainly glutamatergic) neurons (Monti and Monti, 2000). Some of the cholinergic neurons in the pedunculopontine nucleus and laterodorsal tegmental nucleus contain α1- and α2-adrenoceptors, which have been proposed to be active during waking and are excited by noradrenaline from the locus coeruleus acting at α1-adrenoceptors (Hou et al., 2002).
Overall, the anatomical and morphological structure of the pedunculopontine nucleus is similar in rodents, cats, non-human primates and humans (Figs 1–4). However, architecture or circuitry distribution of cholinergic, glutamatergic or GABA-ergic neurons seems to vary, as well as the phylogenetic distribution of afferent and efferent projections. Cytoarchitecturally the pedunculopontine nucleus consists of two parts: a compact part, pars compacta with a higher density of cholinergic neurons dorsolaterally; and a diffuse part, pars dissipata with glutamatergic, cholinergic and other neuron types, situated at the rostrocaudal axis of the pedunculopontine nucleus (Inglis and Winn, 1995).
Location of the human pedunculopontine nucleus pars compacta (PPTgC), pedunculopontine pars dissipata (PPTgD) and the cuneiform nucleus (CnF) and major surrounding structures along the axis of the brainstem. (A) 34 mm and (B) 35 mm rostral to the obex. (Reprinted from Paxinos and Huang, 1995, with kind permission from Elsevier.)
The coronal sections which are 15.5 mm (A) and 16.5 mm (B) posterior to the midcommissural point (3.5 and 4.5 mm posterior to the posterior commissure) in a human brain show the pedunculopontine nucleus, which has been marked with a circle. The anatomical abbreviation in the Schaltenbrand-atlas is tagged as Tg.pdpo (nucleus tegmenti pedunculopontinus). (Reprinted from Schaltenbrand and Wahren, 1977, with kind permission from Thieme.
Anatomical location of pedunculopontine tegmental nucleus (PPTg) and cuneiform nucleus (CnF) in rat midbrain region. 3A and 3B are a Schematic coronal plane of rat brainstem shown at −8.00 to −8.30 posterior to bregma, 1.8–2 mm lateral to the midline and 6.8–7.4 mm ventral to the skull surface according to Paxinos and Watson (1998).
Anatomical location of reticular formation (FR) and nucleus raphe (NR) in cat midbrain region. (A) Schematic coronal planes of the cat brainstem, where cuneiform nucleus (CnF) and pedunculopontine (PPTg) are lying in the coordination P 2.0, LR 3 to 4 mm H 0.5 to −5, according to sterotaxic atlas of the cat brain by Snider and Niemer, 1970. (B) Sketch of A, showing the position of PPTg.
Interspecies differences in Afferent and Efferent Connections
Anatomical Aspects With respect to afferent and efferent connections, phylogenetically related differences exist in the pallidotegmental pathway in rats and cats as compared with monkeys and humans. In rats, the pedunculopontine nucleus receives most of its afferents from the substantia nigra (Moon-Edley and Graybiel, 1983; Steininger et al., 1992). However, in the monkey the pedunculopontine nucleus receives its most extensive afferents from the globus pallidus medialis (De Vito et al., 1980; De Vito and Anderson, 1982). Additionally, retrograde axonal tracing studies showed that efferents of the globus pallidus medialis target a broader area of the pedunculopontine nucleus than efferents from the entopeduncular nucleus, i.e. the equivalent to the primate globus pallidus internus in rodents or cats (Lee et al., 2000). Some authors have even suggested that the entopeduncular nucleus is not linked with the pedunculopontine nucleus, but rather with a brainstem region located just medial to it, which they referred to as the 'midbrain extrapyramidal area' (Rye et al., 1987; Lee et al., 1988; Steininger et al., 1992). Similar to rats, projections of the entopeduncular nucleus appear to be less in cats than in monkeys (Vilensky et al., 1985).
The distribution of dopaminergic nerve terminals of the thalamus is markedly denser and more widespread in primates as compared with the rat (Garcia-Cabezas et al., 2009). Likewise, it has been reported that feline pallidothalamic projections are more restricted than those of primates (Larsen and McBride, 1979). Instead, the differentiation of afferent and efferent projections of substantia nigra pars reticulata to the pedunculopontine nucleus is more distinct in the monkey than in the rat. Additionally, the spatial relationship and morphological features of substantia nigra pars reticulata efferent cell populations exhibit a pattern in the monkey that does not compare well with that seen in the rat (Beckstead and Frankfurter, 1982). It has also been observed that in the rat there is a great overlap in the nigrotectal, nigrotegmental and nigrothalamic projecting neurons, with as many as 50% of the cells projecting to the thalamus and tectum (Deniau et al., 1978; Bentivoglio et al., 1979). In contrast, little overlap in these projection cells exists in the monkey substantia nigra pars reticulata (Parent et al., 1983), which allows a greater topographical discrimination of distinct output behaviour in different species.
Physiological Aspects In line with this, physiological experiments have suggested that the pathway from the globus pallidus to the pedunculopontine nucleus may be larger in the monkey than in the cat. Antidromic activation experiments in the cat suggest that 8% (Larsen and Sutin, 1978; Larsen and McBride, 1979) to 50% (Filion and Harnois, 1978) of the entopeduncular nucleus neurons project to the pedunculopontine nucleus, whereas >80% of globus pallidus internus neurons that project to the pedunculopontine nucleus send axon collaterals to the ventrolateral nucleus of the thalamus in monkeys (Harnois and Filion, 1982; Parent and De Bellefeuille, 1982). With respect to efferent connections, electrical stimulation of the pedunculopontine nucleus in rats increases the firing rates of neurons in the entopeduncular nucleus (Scarnati et al., 1988), suggesting a strong efferent connection to this nucleus. Instead, in the monkey the pedunculopontine nucleus-efferent fibres to the subthalamic nucleus and substantia nigra pars reticulata are much larger than those to the pallidal complex (Lee et al., 2000).
Behavioural Aspects The subthalamic nucleus exerts diverse behavioural output, either directly or via projection of the pedunculopontine nucleus. There is evidence that the connectivity of neurons in the subthalamic nucleus differs between monkeys and rats. For example, the response to stimulation and blockade of GABA receptor antagonists in the monkey subthalamic nucleus is not predicted by the neural circuitry model derived from studies in cats or rats. While in rats unilateral microinjection of a GABAA receptor agonist into the subthalamic nucleus produced a site-dependent contralaterally directed postural asymmetry without locomotor activation (Dybdal and Gale, 2000; Périer et al., 2002), in cats microinjections of the GABAA receptor agonist muscimol into the subthalamic nucleus elicited circling behaviour to the contralateral side, while the GABAA receptor antagonists bicuculline and picrotoxin produced ipsilateral turning (Murer and Pazo, 1993). Additionally, inhibition of the subthalamic nucleus using focal injection of the GABA agonist muscimol was without effect in the monkey (Dybdal et al., 1997). On the other hand, a recent study has shown that microinjections of muscimol in the subthalamic nucleus induced movement disorders such as hemiballism and laterocollis (Karachi et al., 2009), whereas blockade of GABA inhibition in the subthalamic nucleus by focal application of bicuculline evoked dyskinetic and postural responses similar to those evoked by inhibition of the substantia nigra (Crossman et al., 1984; Dybdal et al., 1997; Dybdal and Gale, 2000). This suggests that, unlike the rat subthalamic nucleus, the monkey subthalamic nucleus may exert a net inhibitory influence on the substantia nigra either directly or via projection of the globus pallidus or pedunculopontine nucleus.
Diffusion Tractography Aspects Recently, probabilistic diffusion tractography image tracing showed differences in the organization of connectivity in the pedunculopontine nucleus between humans and monkeys. This study has confirmed the connection between the pedunculopontine nucleus and substantia nigra in monkeys, but lends credence to the surprising finding that the human pedunculopontine nucleus does not have a strong connection with the substantia nigra. Further, the cerebellum and spinal cord are more strongly connected in humans compared with monkeys (Aravamuthan et al., 2009). These organizational differences might be important to understand the electrophysiological output of spiking in connection with the pathology of gait in rodent, monkey and human patients.
The Pedunculopontine Nucleus Area in Clinical and Animal Studies
Lesion Studies in Humans In human Parkinson's disease, the destruction of cholinergic neurons in the pedunculopontine nucleus parallels the progression of the destruction of dopaminergic neurons in the substantia nigra pars compacta, but there is no evidence for the destruction of GABA-ergic neurons in the pedunculopontine nucleus. Additionally, a previous study of human autopsies in patients with Parkinson's disease showed a significant reduction in the total number of substance P-neurons in the pedunculopontine nucleus (decrease of 43%), which was corroborated by findings in the laterodorsal tegmental nucleus (decrease of 28%) and the oral pontine reticular nucleus (decrease of 41%), as well as in the median raphe nucleus (decrease of 76%; Gai et al., 1991).
Clinical case reports have shown that lesions or subacute infarctions in the midbrain region primarily present with gait instability and support the view that the mesencephalic locomotor region plays an important role in locomotion control in humans. Because of these reports, it was hypothesized that pedunculopontine nucleus damage can lead to gait disorders (Masdeu et al., 1994; Bhidayasiri et al., 2003; Hathout and Bhidayasire, 2005; Kuo et al., 2008). Clinical evidence hypothesizes the involvement of the mesencephalic locomotor region, in particular the pedunculopontine nucleus, in gait ataxia (Hathout and Bhidayasire, 2005) or in gait disturbances of patients with dystonia. The pedunculopontine nucleus, cuneiform and the periaqueductal grey may be involved in dystonia (McNaught et al., 2004; Shashidharan et al., 2005; Holton et al., 2008; Loher and Krauss, 2009). These studies provide a complex association between the pedunculopontine nucleus and transmitters in the brainstem and basal ganglia that are involved in the regulation of motor function and gait disturbance.
Lesion in non-human Primates Animal studies in non-human primates have shown that unilateral radiofrequency lesions of the pedunculopontine nucleus cause a temporary akinesia, and bilateral lesions of the pedunculopontine nucleus result in sustained akinesia (Aziz and Stein 1997; Munro-Davies et al., 1999). Additionally, studies in monkey have shown that unilateral excitotoxic lesions of the pedunculopontine nucleus with kainic acid produce hemiparkinsonism, which was characterized by relatively mild levels of flexed posture and hypokinesia in the contralateral limbs (Kojima et al., 1997; Matsumura and Kojima, 2001). A recent study has shown that in the monkey, bilateral preferential lesions in the pedunculopontine nucleus resulted in a significant and reproducible gait and posture instability, which could not be ameliorated by treatment with the dopaminergic agonist apomorphine (Karachi et al., 2010). This result is concomitant with the observation that gait and postural instability of late stage Parkinson's disease or symptoms of progressive supranuclear palsy in humans do not respond to treatment with dopaminergic agonists, possibly because these symptoms are due to advanced destruction of cholinergic neurons in the pedunculopontine nucleus.
Lesion Studies in Rodents Disturbed motor function, such as akinesia or gait disturbances i.e. deficits related to Parkinson's disease, have not been found after lesioning the pedunculopontine nucleus in rodents, whereas in monkeys or in humans, degeneration or lesion of the pedunculopontine nucleus may cause akinesia or gait disturbance. In rodents, however, mostly non-motor behaviour has been investigated after local manipulation of the pedunculopontine nucleus. The function of the pedunculopontine nucleus has been associated with sensorimotor gating, execution of externally cued reward motor, or place preference context and reinforcement, but not locomotion as such (Dellu et al., 1991; Koch et al., 1993; Inglis et al., 1994; Olmstead and Franklin, 1994). Local injection of the excitotoxin ibotenic acid in the pedunculopontine nucleus produced deficient sensorimotor gating, increased anxiety and disturbed working memory (Koch et al., 1993; Steiniger and Kretschmer, 2004). Another study has observed a reduction in anxiety-like behaviour after bilateral electrolytic lesions in the pedunculopontine nucleus (Homs-Ormo et al., 2003). Regional differences in the extent of damage to the pedunculopontine nucleus or to surrounding structures, or the differences in the degree of damage to cholinergic versus non-cholinergic cells, could be responsible for the contradictory results reported. Notably, these studies did not find changes in spontaneous ambulatory behaviour. On the other hand, it is remarkable that injection of GABA antagonist into the pedunculopontine nucleus abolishes haloperidol-induced catalepsy in rats (Miwa et al., 1996).
Although most of the studies that completely lesioned the pedunculopontine nucleus in rodents indicated no motor impairment (Winn, 2006), one study has shown that lesion within a restricted portion of the anterior part of the pedunculopontine nucleus or (pedunculopontine nucleus pars dissipata in rats) do produce motor deficits (Alderson et al., 2008). These results are consistent with the hypothesis that the anterior pedunculopontine nucleus, which is analogous to the pedunculopontine nucleus pars dissipata in rodents, has functions related to the motor process, while the posterior pedunculopontine nucleus or pedunculopontine nucleus pars compacta is less concerned with motor control processes (Olszewski and Baxter, 1954).
Interspecies differences and Animal Models Rodent and monkey studies have accumulated data about the involvement of the pedunculopontine nucleus in the development of Parkinson's disease. Experimental studies using the 6-hydoxydopamine lesion rat model have shown higher neuronal firing activity in the pedunculopontine nucleus (Breit et al., 2001, 2005; Jeon et al., 2003). On the other hand, unilateral lesions of the pedunculopontine nucleus have been shown to induce hemiparkinsonism in monkeys (Kojima et al., 1997; Aziz et al., 1998; Munro-Davies et al., 1999; Matsumura, 2001), which shows that in monkeys, hypoactivity of the pedunculopontine nucleus causes Parkinson's disease-like symptoms. This may be explained by a lesioning-induced hypoactivity of substantia nigra pars compacta neurons that leads to the appearance of parkinsonian motor symptoms. Neuronal metabolic activity in the pedunculopontine nucleus provides further evidence in favour of underactivation of pedunculopontine nucleus neurons in the Parkinson model of basal ganglia in monkeys (Gomez-Gallego et al., 2007). Therefore realization of the differences between species and differences in animal models needs to be taken into account when explaining the pathophysiology of movement disorders and the role of the pedunculopontine nucleus (Vilensky et al., 1985).
The different results of electrophysiological studies in rodent, non-human primate and human patients with Parkinson's disease may be explained on the basis of variable afferent and efferent connectivity to the basal ganglia nuclei. It is conceivable that degeneration of the pedunculopontine nucleus in Parkinson's disease in humans causes overactivity of the glutamatergic efferent projection from the subthalamic nucleus and disinhibits the GABA-ergic interneurons in the pedunculopontine nucleus, which in turn might decrease the outflow in both ascending and descending circuitries of the basal ganglia (Pahapill and Lozano, 2000; Mena-Segovia et al., 2004).
Validity of the 6-hydroxydopamine and 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine Models for Studying the Role of the Pedunculopontine Nucleus It should be noted that the two validated animal models that are used mostly for Parkinson's disease therapeutic experimental research, namely the 6-hydroxydopamine induced lesion in rodents and the chronic or acute 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) primate model, only show the pathology of dopaminergic destruction in the basal ganglia, but not the destruction of the pedunculopontine nucleus, even though the depletion of dopamine has a profound indirect and direct effect on the pedunculopontine nucleus (Karachi et al., 2010). Therefore the question remains open as to whether electrophysiological findings in the pedunculopontine nucleus are truly comparable.
To produce gait disorders or akinesia in rodents, the modulation of both dopaminergic neurons in the substantia nigra pars compacta and cholinergic and non-cholinergic neurons in the pedunculopontine nucleus, which are involved in the pathology of Parkinson's disease or progressive supranuclear palsy in humans, might be necessary. A previous study has proposed that, in order to show a sufficiently pronounced motor deficient behaviour in quadrupedal animals (rodents), it requires 50–70% dopamine depletion in the striatum, which resembles up to 50% of dopamine loss in the substantia nigra pars compacta (Fearnley et al., 1991). While acute or chronic lesions of dopaminergic neurons in the substantia nigra pars compacta by 6-hydroxydopamine or MPTP toxin apparently do not affect the integrity of the pedunculopontine nucleus in rodents, a recent study has shown that MPTP treatment in aged monkeys destroyed cholinergic neurons of the pedunculopontine nucleus (Karachi et al., 2010).
Chronic treatment with the environmental toxin rotenone in rodents produces a hypokinetic syndrome, postural instability and unsteady gait, which is accompanied by destroyed cholinergic neurons in the pedunculopontine nucleus and the presence of Lewy bodies in the striatum and substantia nigra pars compacta (Betarbet et al., 2000; Alam and Schmidt, 2002; Höglinger et al., 2003). This rodent model may therefore be more suitable to investigate the role of pedunculopontine nucleus stimulation on disturbed parkinsonian gait and other Parkinson's disease symptoms.
On the other hand, observation of gait disturbance by lesioning the pedunculopontine nucleus in rodents cannot be fully elucidated in the activity box or from the catalepsy test. So far, exploratory behaviour in the activity box (a dynamic test) and catalepsy (a static test) have been investigated, but the kinetics of gait disturbance are a mixture of both static and dynamic motor behaviour (Alam et al., 2009). Therefore, in most studies that have investigated gait disturbance after pedunculopontine nucleus lesion in rodents, the appropriate behavioural assessment, such as rotarod or stepping assessment, was not used.
The Cuneiform and the Subcuneiform Nucleus
There is controversy and considerable confusion concerning the role of the pedunculopontine nucleus in the initiation of locomotion. Some experiments have related the effect of stimulation, drug injection or lesion on initiation of movement on the pedunculopontine nucleus, but their results may have involved activation in adjacent structures of the mesencephalic locomotor region, especially the cuneiform (Milner and Mogensen, 1988; Brudzynski et al., 1993). Histological verification of sites for electric stimulation and injection in cats, rats and guinea pigs has shown that these sites were invariably located in and around the cuneiform nucleus (Amemiya and Yamaguchi, 1984; Brudzynski et al., 1986; Milner and Mogenson, 1988; Coles et al., 1989).
The cuneiform nucleus is ventrally demarcated by the decussation of the superior cerebellar peduncle and the pedunculopontine nucleus. At caudal levels, it is dorsally bordered by the external cortex of the inferior colliculus, laterally by the dorsal nucleus of the lateral lemniscus and ventrally by the superior cerebellar peduncle (Fig. 1). Similar to the pedunculopontine nucleus, the cuneiform is not a homogeneous population of cells. It consists of two different types of cells, i.e. GABA-ergic and nitrergic neurons (Pose et al., 2000).
The pedunculopontine nucleus and the cuneiform nucleus are both important sites in the mesencephalic locomotor region and have been suggested to facilitate muscle tone during initiation of locomotion (Mori et al., 1987). Some human atlases further differentiate between cuneiform and subcuneiform nucleus, however, without giving a precise anatomical boundary of both the cuneiform and subcuneiform nucleus (Olszewski and Baxter, 1954). Clinical studies have suggested that the most suitable targets are in the subcuneiform nucleus, which is located slightly posterior to the pedunculopontine nucleus pars compacta, probably in the ventral part of the cuneiform nucleus since stimulation-induced locomotion has been reported in animals at this location (Takakusaki et al., 2003; Piallat et al., 2009; Ferraye et al., 2010). However, in most animal studies the subcuneiform nucleus is seen as an integral part of the cuneiform and not specifically outlined. Cholinergic neurons of the pedunculopontine nucleus extend dorsally to the classic pedunculopontine nucleus pars compacta, i.e. in the so-called subcuneiform. Neuronal populations in fact partly overlap in this region and a sharp delineation of separate individual regions does not reflect the exact reality correctly.
Effective sites for chemically evoked locomotion in freely moving rats appear to be located predominantly in the cuneiform and not in the pedunculopontine nucleus as defined by Milner and Mogenson (1988). However, lesions of the cuneiform in rats do not show differences in locomotion behaviour as compared with sham lesion (Allen et al., 1996), while electrical stimulation of the cuneiform nucleus elicited suppression of muscular tone and somatic reflexes (Mileikovsky et al., 1989, 1990; Mileikovsky and Nozdrachev, 1997). These effects may be mediated by premotor inhibitory neurons within the ventromedial medulla (Morales et al., 1999), which is one of the principal targets of cuneiform fibres. It is possible that cuneiform terminals modulate the activity of premotor neurons, since the cuneiform is related to modulation of both the sensory and the motor system (Zemlan and Behbehani, 1988). However, the cuneiform and the pedunculopontine nucleus are close in proximity and perhaps the effect of electrical stimulation or local drug injection always attribute to one or the other (Garcia-Rill and Skinner, 1987a, b; Garcia-Rill et al., 1991).
Brain. 2011;134(1):1-23. © 2011
Oxford University Press
Cite this: The Pedunculopontine Nucleus Area: Critical Evaluation of Interspecies Differences Relevant for its Use as a Target for Deep Brain Stimulation - Medscape - Jan 01, 2011.