The fiber-dissection technique provides unique 3-dimensional anatomic knowledge of the white matter.
To better identify the frontostriatal pathways in the human brain, we used a fiber-dissection technique to reconstruct neural connections between the frontal cortex and the nucleus accumbens (NAcc), which is the most ventral extent of the striatum.
Thirty previously frozen, formalin-fixed human brains were dissected under the operating microscope using a modified fiber-dissection technique, primarily reported by Klingler.
Our serial dissections of 30 human brain specimens clearly demonstrated that projection fibers form a connection between the NAcc and the frontal lobe. We evidenced this newly described subcortical tract as an accumbofrontal fasciculus. This focal projection was concentrated at the level of the ventromedial part of the NAcc and characterized by an elective and specific projection from the orbitomedial prefrontal cortex, particularly the gyrus rectus and the medial orbital gyrus situated between the H-shaped and the medial orbital sulcus.
The accumbofrontal fasciculus is an elective and specific projection from the orbitoprefrontal cortex. This fasciculus is part of a corticostriatothalamocortical loop and a putative target for deep-brain stimulation in the treatment of obsessive-compulsive disorder and major depression. The analysis of in vivo diffusion tractography, used today as a standard in the investigation of many brain disorders, could potentially take advantage of complementary anatomic correlations and functional extrapolations, as described in this study.
Corticostriatal connections in humans delineated with diffusion tensor imaging axonal tracking are organized into discrete circuits, similar to the organization described with tracing studies in monkeys.1 This was the hypothesis initially proposed by Alexander and colleagues,2 largely based on primate-to-human extrapolations. Anatomic studies in animals have already described multiple striatal circuits and suggested that subcomponents of the striatum project to distinct cortical areas.2 These circuits are subdivided into subchannels conveying sensorimotor, association, or limbic information.
The limbic compartment of the striatum (ventral striatum including the head of the caudate nucleus and the nucleus accumbens [NAcc]) has highly reproducible connections across subjects in humans. Tracing studies in monkeys have shown that the ventral striatum receives afferents from structures associated with the limbic system,3 including the orbitomedial prefrontal cortex (OMPFC),4 the amygdala, the hippocampus, the anterior cingulate cortex, and the temporal pole.
Today it remains clear that the use of deep-brain stimulation (DBS) for psychiatric indications confirms case series data to suggest that DBS in the frontostriatal pathway shows interesting preliminary results.5–8
The selection of sites for implantation of DBS electrodes for psychiatric conditions should be directed at modulating the activity of strategically placed nuclei or white matter tracts. A clear consensus on one optimal target for stimulation is not yet achievable but evidence exists to support the notion that DBS of the ventral caudate/ventral striatum and NAcc may be promising targets for treatment-resistant depression and obsessive-compulsive disorder.
This choice of potential targets for DBS should find a rationale in human anatomic studies since most clinical protocols are based on morphofunctional extrapolations from animal studies. Research efforts are currently focusing on brain functional imaging techniques (ie, PET and functional MRI) but only the analysis of gray matter involvement has been explored thus far. Diffusion tensor imaging (DTI) takes advantage of new tools capable of extracting white matter anatomy from the brain and makes it possible to describe the morphological connections between functional cortical regions.9 Nevertheless, its main limit remains its capability to select the appropriate region of interest for data processing, which requires perfect anatomic knowledge of the underlying neural networks, the subject of this analysis.10 As a consequence, the description of white fiber anatomy in current articles using DTI is often imprecise, because it is based on sketches of the tracts and not on consistent data provided by dissected material.
To better identify the neural connections between the frontal cortex and the NAcc in the human brain, we used a modified fiber-dissection technique previously described by Klingler.11,12 This technique appears to be less time-consuming for an equivalent dissection quality. The fiber-dissection technique involves peeling away the white matter tracts of the brain to display its 3-dimensional anatomic organization. The difficulties preparing the brain and the execution of fiber dissection have resulted in this method being neglected, particularly since the development of microtome and histological techniques.
Nevertheless, the fiber-dissection technique remains a very relevant and reliable method for neurosurgeons to study the details of the brain's anatomic features.
MATERIALS AND METHODS
We dissected 30 previously frozen, formalin-fixed human brains under the operating microscope using a modified fiber-dissection technique originally described by Klingler12 and more recently by Türe et al.13 The brains were obtained from fresh autopsy specimens no later than 12 hours postmortem and were fixed in a 10% formalin solution for at least 24 hours. To maintain the normal contours of the brain, the basilar artery was ligated and used to suspend the brain in the formalin solution. The specimens were washed for several hours to remove the formalin and refrigerated at −18°C for 24 hours. The brains were then thawed and refrigerated again for another 24-hour cycle. This procedure was repeated 4 times with the same time intervals. The specimens were stored in a 4% formalin solution at room temperature before and between dissection sessions. They were dissected with use of the operating microscope with ×6 to ×30 magnification. The dissection tools were handmade by carefully shaping wooden spatulas with various tip sizes. The specimens were dissected in a stepwise manner, beginning with a lateral surface dissection of a cerebral hemisphere, from the cortex to the deep tissues. Using an identical method, a second procedure also included a dissection of the same hemisphere's medial surface. We decided to perform this combined approach to demonstrate NAcc connecting fibers, which could only appear after removal of adjacent fiber systems.
After removal of the arachnoid membrane and the leptomeningeal vessels, the superficial gray matter was removed, exposing the underlying white matter fibers.
To visualize the uncinate fasciculus, the anterior segment of the superior temporal gyrus, a portion of the middle temporal gyrus and the temporal pole were partially removed by the first step of fiber dissection (Figure 1). The cortex of the insula and parts of the extreme capsule, the claustrum, the external capsule, and the orbital gyri were dissected away. The fiber bundles of the uncinate fasciculus originated in the white matter of the temporal lobe, coursed around the M1 segment, entered the extreme and external capsules, connecting the temporal pole with the orbitofrontal region.
The uncinate fascicle is a large bundle that forms a double-fan-like structure consisting of a ventromedial part and a dorsolateral part.14 The ventromedial part connects the gyri on the orbital surface of the frontal lobe with the parahippocampal and other gyri on the medial surface of the temporal lobe. This fasciculus is connected to the medial frontal areas (subcallosa, gyrus rectus). The dorsolateral bundle united gyri on the superolateral part of the frontal lobe with the cortex of the more lateral temporal gyri near the temporal pole (Figure 1A).
The inferior occipitofrontal fasciculus, which forms the most dorsal part of the limen insulae, connected the frontal opercula and prefrontal region with the posterior temporal and occipital regions (Figure 1B). It passed through the basal portion of the insula, immediately superior to the uncinate fasciculus. Both fasciculi formed a double fan connected by a narrow isthmus deep to the limen insulae.
The gray matter of the anterior segment of the uncus was then removed to expose the amygdala. At this level, the ventral claustrum extended toward the base of the frontal lobe below the putamen, formed by islands of gray matter intermixed with and fragmented by fibers of the uncinate fasciculus (Figure 2). Medially, the anteroinferior part of the putamen and the head of the caudate nucleus blended in with the NAcc.
A constant white matter tract connected the amygdala and the NAcc. This fasciculus was found to arise from the anterolateral extension of the amygdala. It quickly took on an archiform incurvation, concave on the outside, with its lateral side against the uncinate fasciculus, partially covering the narrowest part of the fasciculus. This fasciculus was flattened to follow the inferolateral concavity and spread out on its inferior side where it became entirely spindly. This “amygdaloaccumbens” fasciculus was found systematically during the dissections. It joined with the uncinate fasciculus as it emerged from the amygdala.
The frontostriatal radial projections stemming from the caudate nucleus (Figure 3) were then isolated. The most caudal fibers of the anterior limb projected on the anterosuperior part of the anterior prefrontal cortex. The most rostral fibers, or the lower anterior part, became horizontal, taking an anteroinferior direction and approaching the orbitofrontal cortex, but did not reach it. At the anteroinferior extremity of the caudate nucleus, fibers could be seen individuating from the rest of the frontostriatal projections going toward the medial part of the orbitofrontal cortex (Figure 4). These fibers rose in part from the NAcc. To describe them more precisely, the dissection was pursued on the medial side of the hemisphere.
In the entire specimen, we dissected the medial aspect of the cerebral hemispheres (Figure 5). After removing the cortex, the medial portion of the corpus callosum and the fornix, we exposed the frontal horn of the lateral ventricle. The removal of the ependyma of the frontal horn exposed the subcallosal stratum, which is located between the caudate nucleus and the radiation of the corpus callosum. The head of the caudate nucleus was dissected on the anterior and inferomedial part and we progressively demonstrated the junction between the NAcc and the anteromedial part of the head of the caudate nucleus (Figure 6). At this level, the lower anterior part of the lentiform nucleus blends into the lower part of the head of the caudate nucleus in the area below the anterior limb of the internal capsule.
We then demonstrated the constant fiber projections arranged in a fasciculus connecting the NAcc and the orbitomedial prefrontal cortex (Figure 7). This accumbofrontal tract arose ribboned with the inferomedial part of the NAcc, whence it emerges on the lateral edge to then bend, taking an anterior and more medial direction, where it is generally flattened. It is then divided into several “subfasciculi,” which feed the axial fan resulting from the flattening fibers. Of the 30 anatomic specimens, 6 fasciculi were divided into 3 groups, 16 fasciculi were divided into 2 groups, 3 fasciculi remained solitary, spindled, and filiform, and 5 fasciculi were unexploitable.
Three types of frontostriatal projections were finally evidenced, as follows:
-A medial projection, described as “the accumbofrontal tract.”
-A lateral projection corresponding to the orbitolateral prefrontal fibers, evidenced near the lateral junction between the NAcc and caudate nucleus. These fibers were oriented laterally toward the inferolateral part of the prefrontostriatal cortex.
-An intermediate projection rising from the most anteromedial and lower part of the caudate nucleus head, inconstantly individualized between these 2 fascicles (Figure 6).
The accumbofrontal tract seems to be characterized by an elective and specific projection to the most orbitomedial part of the prefrontal cortex including the gyrus rectus and the medial orbital gyrus.
The corticostriatal system comprises broad fan-shaped projections that are topographically organized, stemming from the motor and prefrontal cortical areas, converging toward the striatum. Herein is individuated a triple anatomic and functional subdivision (motor, associative, and limbic), reflecting the anatomic organization of the cortical afferences and supporting information transfer from the cerebral cortex toward the basal ganglia. The projections passing from the cerebral cortex to the basal ganglia, which return to the cortex via the thalamus, form several parallel and functionally segregated loop systems.2,15–17 Prominent among these are a motor loop involving the motor and premotor cortices, an associative or cognitive loop, involving the dorsolateral prefrontal cortex, and a limbic loop, involving the orbital and medial prefrontal cortex.
Previous studies have already emphasized separate terminal fields from medial and orbital prefrontal cortex networks in the striatum.18,4 Recent human imaging studies have divided the prefrontal cortex into the dorsal anterior cingulated cortex, the ventral, medial prefrontal cortex, the orbital prefrontal cortex (OPFC), and the dorsolateral prefrontal cortex based on specific roles for mediating different aspects of error prediction and decision making.19–21 The OPFC can functionally be separated into its medial (OMPFC) and lateral (orbitolateral prefrontal cortex) parts.
Anatomical Considerations on the Accumbofrontal Tract
Our serial dissections of 30 brain specimens clearly demonstrated that we could individualize, among projection fibers of the prefrontal cortex, a connection between the ventromedial part of the NAcc and the frontal lobe, in particular, the OMPFC. We describe this neuroanatomical connection as the accumbofrontal fasciculus. This focal projection of the OMPFC is concentrated at the level of the NAcc and corresponds to observations made in adult macaque monkeys after injection of anterograde tracers into the ventral, medial prefrontal cortex around the gyrus rectus area.22 The injection site used actually seems to correspond to the OMPFC, whose associative fibers were demonstrated during the dissections. It is of particular interest that this projection remained confined to a strictly limited ventromedial striatal position at the level of the NAcc, within and dorsal to its shell. Several physiological studies in monkeys23–25 have confirmed the involvement of the ventral portion of the striatum in a functional axis with the OPFC.
In this study, we demonstrated that 3 types of projections arising from the OPFC could be individuated: (1) a focal projection from the most medial part of the OMPFC to the NAcc, which is delineated herein as the accumbofrontal fasciculus, (2) an intermediate fasciculus from the rest of the OMPFC to the inferomedial part of the head of the caudate nucleus, and (3) a more lateral fasciculus corresponding to orbitolateral prefrontal cortex projection fibers, classically described in the literature (Figure 6).
The accumbofrontal fasciculus is characterized by an elective and specific projection from the OMPFC, in particular, the gyrus rectus and the medial orbital gyrus situated between the H-shaped and the medial orbital sulcus. Brodmann26 used the number 11 to refer to a region that included the gyrus rectus and more laterally a large part of the rostral orbitofrontal cortex. It is important to note that the boundaries between the various association areas of the prefrontal cortex do not show a close correspondence to those of the cytoarchitectonic fields, as delineated by Brodmann and other authors.27
Morphological Anatomy of the Ventral Striatum Revisited
The limbic or ventral striatum28 occupies the dorsomedial to ventral part of the caudate nucleus as well as the precommissural part of the putamen.29 It therefore comprises the nucleus accumbens, the ventromedial putaminal-caudate part, and the olfactory bulb.30 The borders of the ventral striatum seem to blend with the borders of the ventral pallidum. Some authors explain this with the human striatopallidal ventral complex concept16,31 in which the ventral striatum is said to interdigitate with ventral pallidal components and be directly contiguous with extensive ventral striatal areas (including the NAcc further rostrally, at the anterior and medial junction of the ventral striatum), as described in this dissection.
Initially defined in relation to its amygdalostriatal afferences as well as those stemming from the allocortex (olfactory and hippocampal cortex),31 the ventral striatum is also the seat of afferences emanating from the cingulate. It finally receives afferences coming from the median and parafascicular nuclei of the thalamus as well as a modulating innervation from the tegmental-ventral area.
The connections of the ventral striatopallidal complex with the limbic cortex thus make a cortical-subcortical reentry loop, the limbic loop.
The limbic system (lobe) is not only defined by the association of one part of the OMPFC with the olfactory, hippocampal, and parahippocampal cortex, but also with the basolateral nucleus of the amygdala and a part of the insula.32 The basolateral nucleus receives its afferences from the thalamus, the orbitofrontal cortex, one part of the temporal lobe, and the cingulate gyrus. The efferent connections of the amygdala are for the most part reciprocal with its afferents. A major efferent pathway joins the hypothalamus via the stria terminalis, with also a connection with the mediodorsal nucleus. This provides an indirect connection between the amygdala and the prefrontal cortex, particularly its orbitofrontal portion. Through the basolateral nucleus, the amygdala is connected with the ventral striatum by amygdala-accumbens fibers. These were found in all the dissections (Figure 1B, Figure 2). Nieuwenhuys et al33 describes them as “fibers originating mainly from the basal amygdaloid nucleus passing via the longitudinal association bundle to the ventral striatum which includes the accumbens nucleus.” Experimental studies have shown that this amygdalostriatal projection is present in the monkey34 including the NAcc as a target area of the amygdala. In rats, numerous neurons in the basolateral amygdaloid nucleus project to both the prefrontal cortex and the nucleus accumbens.35 Finally, the amygdala is organized such that it parallels the striatal component of the limbic corticostriatothalamocortical loop.36 The basolateral amygdala has a distinct reciprocal relationship with the NAcc core and the cortical regions associated with this striatal structure. The centromedial amygdala together with the bed nucleus of the stria terminalis (considered together as the extended amygdala) has its relationship with the NAcc shell, area 25, and widespread areas of the hypothalamus.4
Morphological Data on the Nucleus Accumbens
The NAcc measures approximately 8 × 6 × 6 mm at its largest dimensions and has an ellipsoid shape with relatively clearcut boundaries on myelin-stained sections, with the most superior edge directed medially.37 Other recently published data contradict these mean values, reporting a length of 10.5 mm, a width of 14.5 mm, and a height of 7.0 mm.38 The NAcc extends dorsolaterally into the ventral putamen and dorsomedially into the ventral caudate without a real demarcation in most studies. Our study did not allow a detailed morphometric description despite a relatively clear microanatomical landmark between the NAcc and the caudate nucleus. This lack of morphological comparative aspects in this study stems from the tissue dehydration caused by formalin fixation using Klingler's technique, resulting in approximate morphometric data.
The NAcc is functionally divided into 2 distinct components: the shell and the core,39–42 impossible to individuate in our dissections. Historically, the anatomic distinction between these 2 structures has been easier to detect in rats and lower primate models; however, recent human evidence points to the core being the dorsolateral component distinguished by a lower concentration of opiate receptors and a higher concentration of calcium-binding proteins. The core is virtually identical to the ventromedial caudate's histological structure and cannot be clearly separated from the rest of the striatal complex. The shell, the ventromedial portion of NAcc, is inversely devoid of calcium-binding proteins and replete with opiate receptors43 This shell is characterized by several atypical striatal features, some of which are reminiscent of the extended amygdala, which is directly continuous with the posteromedial shell.32 Distinct anatomic boundaries separating such functional units may not be apparent,44 as in our dissections. Pennartz et al45 pointed out that the NAcc is a collection of neuronal ensembles or groups with specific input–output relationships and with different functional and behavioral connotations.
Finally, the OMPFC projects to the medial and ventral portion of the striatum22 and especially to the NAcc via the accumbofrontal tract, described in this study. Output from this striatal region terminates in the medial and ventral pallidum and in the medial portion of the substantia nigra pars reticulata. 4,29 The pallidothalamic projections terminate in the magnocellular part of the mediodorsal thalami, which in turn project to the OMPFC.29 In addition, this relatively segregated corticosubcorticocortical loop is also connected to the amygdala28,34 via the amygdalar-accumbens fasciculus and can thus be considered a limbic loop (Figure 8).
Functional Extrapolation Data
Animal and human studies emphasize the essential role of the striatum in processing reward-related information.25,46,47 The ventral striatal loop, or limbic loop, involves different structures. The main center of this loop seems to be the NAcc and the ventral striatum, which receive projections from the prefrontal cortex and control the ventral pallidum. The orbitofrontal cortex can be functionally separated into a lateral and medial part; the activity in the medial part is related to the learning, monitoring, and memory of the reward value of positive reinforcers, whereas the lateral part's activity is related to the evaluation of negative reinforcers.48 Lesions or dysfunctions of the OMPFC can thus lead to an insensitivity to reward, which may in turn be responsible for a decreased number of voluntary actions and be associated with impulsive and reflexive responses to endogenous or environmental percepts.49 Several functional magnetic resonance imaging studies have reported activation in the ventral striatum and ventral or medial frontal regions during monetary reward or punishment.50 In addition, parts of the striatal area that receive the OFC inputs also receive inputs from the amygdala and the hippocampus, respectively, associated with emotional valence and memory.51,52,28 Finally, these connections appear to integrate information related to reward processing and memory to modulate striatal activity.53
An understanding of the thalamocortical/corticostriatothalamocortical loops appears to be important since DBS is now considered a major psychiatric disorder treatment option.5–7,15,54 Greenberg et al7 have confirmed the favorable effects of chronic bilateral stimulation of the ventral striatum in patients with obsessive-compulsive disorder (OCD). Similarly, bilateral DBS of the NAcc produces a marked improvement in patients experiencing treatment-resistant major depression.55 High-frequency stimulation of the NAcc reduces the firing rate of neurons in the orbitofrontal cortex in rats.56 Metabolic hyperactivity in this region has been consistently associated with OCD symptoms.51 However, several brain white matter tracts cannot be identified on MRI studies because they are indistinguishable from the surrounding white matter. Fibers of the uncinate fasciculus, the inferior occipitofrontal fasciculus, and similar white matter tracts are intermingled with other fibers that course in various directions. This may explain why they are not clearly visible on MR images.
The fiber-dissection technique provides unique 3-dimensional anatomic knowledge of the white matter. This technique is limited because the demonstration of one fiber system usually results in the destruction of other adjacent fiber systems. DTI is a powerful and noninvasive technique for visualizing white matter fiber tractography and can reveal complex relationships among different fiber systems at the same time. However, DTI can introduce continuity between the fibers where there is none and suffers from image distortions and artifacts due to magnetic susceptibility.57 Another source of inaccuracy in DTI studies is manual selection of the seed points, which limits the repeatability of such techniques. Small differences in region-of-interest placements may result in significantly different reconstructed fiber tract bundles.58 Finally, a precise understanding of the 3-dimensional anatomy of the white tracts on the basis of fiber dissections could increase the accuracy of the seed point selection process. These 2 complementary techniques should be combined for neuroanatomical research. Nevertheless, these techniques do not reveal the precise origin and termination of the fibers composing each fasciculus, which depends on gray matter and furthermore on brain functional explorations.
As suggested here, the analysis of in vivo diffusion tractography used in the investigation of brain disorders (for example, major depression, OCD, obesity) could become more precise, because it would be based on a solid anatomic foundation. This article illustrates the critical importance of complementarity between in vivo and ex vivo anatomic techniques in investigating neurological and psychiatric disorders, thus providing a reliable prerequisite for further functional correlations. Such extrapolations should only be established from human brain functional imaging and clinical study validations.
Dr Houeto and his colleagues performed an anatomic dissection of the white matter tracts that connect the ventral striatum and the orbital surface of the frontal lobe. They used a modification of the technique described by Klingler and resurrected by Türe and Yasurgil to separate the larger white matter tract in 30 brains. This technique requires great patience and stillness in order to demonstrate the smaller fiber tracts. Our understanding of the importance of the cortico-striatal-thalamal-cortical loops and the orbital surface of the frontal lobe has increased significantly over the past 10 years. The orbital surface of the frontal lobe not only assigns a reward value to a given stimulus but can modify the reward value. In broad strokes the medial orbital frontal lobe upgrades reward values, and the lateral orbital frontal lobe downgrades reward values. Loss of these functions rob the patient of the ability to respond appropriately to the environment and to learn from experience.
The authors compare their findings with those obtained using DTI. As the authors point out DTI is a technique that requires seeding. The tracts seen depend on the placement of regions of interest and are very sensitive to the parameters chosen by the operator. Tracts can be created or missed depending on those parameters. DTI is not very good at sorting out fiber tracts that cross within a single voxel. Thus, the size of the fiber tract can be grossly underestimated. I would assume that the technique described by Klingler would also have a hard time demonstrating crossing fiber bundles.
The proposal that the tract connecting the nucleus accumbens and the medial orbital frontal lobe would make a good target for treating depression and obsessive-compulsive disorders is interesting. Being a strictly anatomic article that proposal is highly speculative.
The author's anatomical dissections add to our understanding of an area which will certainly be of increasing interest to neurosurgeons in the future.
Allan H. Friedman
Durham, North Carolina
The authors use a modified fiber-dissection technique, primarily reported by Klinger, to study the white matter pathways between the frontal cortex and the nucleus accumbens in 30 human brain specimens. They identify a subcortical tract connecting the orbitomedial prefrontal cortex to the ventromesial part of the nucleus accumbens. The authors conclude that this accumbofrontal fasciculus, part of a corticostriatothalamocortical loop, could represent a target for deep brain stimulation in the treatment of obsessive-compulsive disorder and major depression.
This is a very interesting, well-written article. The rationale of this study is original, since the connectivity of the nucleus accumbens is poorly known. The methodology is robust and well exposed. The results are well presented, with excellent photographs of fiber dissections, and they are well discussed. Currently, very few data are available about the accumbofrontal fasciculus. The technique of white matter pathways should be more widely used to better understand the anatomofunctional connectivity of the brain, especially in combination with diffusion tractography. Thus, this article provides useful information for the neurosurgical community, because it may have both anatomic and clinical implications, especially for deep brain stimulation.
deep brain stimulation;
diffusion tensor imaging;
orbitomedial prefrontal cortex;
orbital prefrontal cortex
- brain diseases
- corpus striatum
- tissue dissection
- objective (goal)
- nucleus accumbens
- obsessive-compulsive disorder
- prefrontal cortex
- frontal lobe
- deep brain stimulation
- major depressive disorder
- white matter
- orbital gyrus
- diffusion tractography
- operating microscope
- biological neural networks
- structure of gyrus rectus