Arcuate fasciculus’ middle and ventral temporal connections undercut by tract-tracing evidence

We are pleased to respond to the recent review article of Giampiccolo and Duffau1 published in Brain about the temporal terminations of the arcuate fasciculus. The arcuate fasciculus is regarded a major white matter pathway for language in the human brain.2 As such, there has been massive cross-disciplinary interest about the arcuate fasciculus since its first discovery in the early 19th century. However, as highlighted in our recent review article,3 the anatomical definition of the arcuate fasciculus remains highly controversial, as its exact course and terminations, and how it differs between human versus non-human primate brains remain intensely debated in the literature. Giampiccolo and Duffau1 had proposed a novel model of the arcuate fasciculus’s temporal connections based on recent evidence from clinical cases and direct electrical stimulation studies. Their model depicted more extensive arcuate fasciculus temporal terminations that reach anterior-ventrally beyond the posterior superior temporal cortex, into the middle temporal gyrus, and the inferior/basal temporal cortex (Fig. 1). While this model may have important clinical implications, it is based on methods that are limited in providing the ‘ground truth’ delineation of the arcuate fasciculus tract. Specifically, we contend that fundamental findings about the arcuate fasciculus’s temporal connections from monkey tract-tracing studies, the gold-standard methodology for detecting intercortical connections, are incongruent with Giampiccolo and Duffau’s model.1 Contemporary models of the arcuate fasciculus fibre pathways in humans. Left: Macaque tract-tracing based human tractography model. Figure adapted from Becker et al.,3 based on Petrides.4Right: Giampiccolo and Duffau’s model with extended temporal lobe terminations added onto the tract tracing model. AF = arcuate fasciculus; SFS = superior frontal sulcus; SPrS = superior precentral sulcus; IPrS = inferior precentral sulcus; IFS = inferior frontal sulcus; AS = ascending branch of Sylvian fissure; HS = horizontal branch of Sylvian fissure; IPS = intra-parietal sulcus; Tpt = temporal parietal area [Brodmann area (BA) 22]; aSMG = anterior supra-marginal gyrus (BA40); pSMG = posterior supra marginal gyrus (BA40); ANG = angular gyrus (BA39); SPL = superior parietal lobule (BA7); MTG = middle temporal gyrus; aITG = anterior inferior temporal gyrus; pITG = posterior inferior temporal gyrus. Precious knowledge about brain connections is derived from tract-tracing experiments in non-human primates, which many, including the authors themselves, acknowledge to be the gold standard in the field.4 Indeed, in contrast with other in vivo techniques, the tract-tracing approach has the advantage to reveal monosynaptic connections (i.e. direct, single-axonal connections from one neuron to another) between cortical areas, and is therefore, the only technique that can reveal the true course and terminations of a single continuous white matter tract, such as the arcuate fasciculus. This method involves the injection of anatomical tracers into a living brain, which are then transported by interconnecting axons from neurons in the injected site to other neurons. Critically, this is followed by histological analyses of the brain tissue post-mortem, in order to visualize the axons, and the cell bodies of the neurons that have been labelled by the injected tracers. Unfortunately, this technique, requiring both invasive in vivo cortical injections, and the eventual sacrifice of the subject for post-mortem histological analyses, is virtually impossible to conduct in humans. Based on tract-tracing experiments in macaque monkeys, the arcuate fasciculus’s temporal terminations rarely exceed the posterior third of the superior temporal gyrus anteriorly (i.e. at the level of the planum temporale), and the ventral bank of the superior temporal sulcus, ventrally (Fig. 1).4 Using diffusion MRI tractography,researchers have performed the careful dissection of the human arcuate fasciculus based on the macaque arcuate fasciculus obtained via tract-tracing. This work had demonstrated that, in the human brain, the arcuate fasciculus temporal terminations stay mostly in the posterior superior temporal gyrus/sulcus region5–8 (Fig. 1, left), which runs contrary to the revised model proposed by Giampiccolo and Duffau1 (Fig. 1, right). Without tract-tracing technique, the best modern-day approaches for delineating the human arcuate fasciculus include diffusion MRI tractography, blunt dissections, and direct electrical stimulation. Notably, Giampiccolo and Duffau1 had carefully derived their model from studies employing all these approaches. Nevertheless, we cannot ignore the fact that these methods do not allow the distinction of mono-synaptical from poly-synaptical connections (i.e. indirect connections that involve one or more interconnecting neurons). In other words, these methods are not able to disentangle the arcuate fasciculus fibres from the other temporal fibre pathways that build synapses with the arcuate fasciculus in the temporal lobe.4,9,10 It is thus possible that the more anterior and ventral temporal gyrus terminations proposed by Giampiccolo and Duffau,1 might not reflect monosynaptic connections from the arcuate fasciculus, but interactions between the arcuate fasciculus and fibres from other temporal pathways. As such, we feel the authors should nuance their position that their revised arcuate fasciculus model represents the true human arcuate fasciculus and discuss the potential disagreements between their model and other recent models of the human arcuate fasciculus that have been derived based on macaque tract tracing studies.5–8 In conclusion, the human arcuate fasciculus model proposed by Giampiccolo and Duffau1 is based on the best available methods (diffusion MRI tractography, electrical stimulation, blunt dissections) to study white matter connections in the human brain. These methods, unfortunately, do not warrant the detection of true monosynaptic connections, which is only possible using invasive tract-tracing techniques. Importantly, their model is incongruent with existing human arcuate fasciculus models that are based on macaque tract-tracing knowledge.4–8 As it stands, there is currently no method to determine which model depicts the true human arcuate fasciculus, since tract-tracing studies are not possible in humans. Until that technology surfaces, the field is confronted with two competing hypotheses: (i) the human arcuate fasciculus has indeed expanded anteriorly and ventrally into the middle and inferior temporal cortex as Giampiccolo and Duffau1 depicts; or (ii) the human arcuate fasciculus is homologous to that of the monkey arcuate fasciculus, and Giampiccolo and Duffau’s observations1 could be due to the fact that the human arcuate fasciculus is entangled with other temporal tracts [e.g. the inferior longitudinal fasciculus (ILF), inferior fronto-occipital fasciculus (IFOF), the middle longitudinal fasciculus (MdLF), uncinate fasciculus (UF), extreme capsule fasciculus (ECF), etc.]. Data sharing is not applicable to this article as no new data were created or analysed in this study. The project has received funding from the European Research Council under the European Union's Horizon 2020 research and innovation program grant agreement No. 716931-GESTIMAGE-ERC-2016-STG (P.I. Adrien Meguerditchian), as well as from grants ANR-16-CONV-0002 (ILCB), ANR-17-EURE-0029 (NeuroSchool) and the Excellence Initiative of Aix-Marseille University A*MIDEX AMX-19-IET-004 (NeuroMarseille). The authors report no competing interests.