Brain (re)organisation following amputation: Implications for phantom limb pain

Abstract

Following arm amputation the region that represented the missing hand in primary somatosensory cortex (S1) becomes deprived of its primary input, resulting in changed boundaries of the S1 body map. This remapping process has been termed 'reorganisation' and has been attributed to multiple mechanisms, including increased expression of previously masked inputs. In a maladaptive plasticity model, such reorganisation has been associated with phantom limb pain (PLP). Brain activity associated with phantom hand movements is also correlated with PLP, suggesting that preserved limb functional representation may serve as a complementary process. Here we review some of the most recent evidence for the potential drivers and consequences of brain (re)organisation following amputation, based on human neuroimaging. We emphasise other perceptual and behavioural factors consequential to arm amputation, such as non-painful phantom sensations, perceived limb ownership, intact hand compensatory behaviour or prosthesis use, which have also been related to both cortical changes and PLP. We also discuss new findings based on interventions designed to alter the brain representation of the phantom limb, including augmented/virtual reality applications and brain computer interfaces. These studies point to a close interaction of sensory changes and alterations in brain regions involved in body representation, pain processing and motor control. Finally, we review recent evidence based on methodological advances such as high field neuroimaging and multivariate techniques that provide new opportunities to interrogate somatosensory representations in the missing hand cortical territory. Collectively, this research highlights the need to consider potential contributions of additional brain mechanisms, beyond S1 remapping, and the dynamic interplay of contextual factors with brain changes for understanding and alleviating PLP.

Keywords: Cortical reorganisation; Multivariate analysis; Neuroimaging; Pain treatment; Phantom limb pain; Preserved function; Use-dependent plasticity.

Fig. 1

Somatotopic mapping of the entire body (A–C) and of the hand (D–E) in primary somatosensory cortex (S1) as revelaed by human task-based fMRI. Somatotopies are often studied using a travelling wave (also known as phase encoding) experimental design, where each of the body parts is stimulated sequentially in a set cycle (A,D). This techinque is designed to identify brain areas showing body part selectivity, such that each colour in the maps indicates selectivity to one body part in the sequence over all others (B – group map, E − sample participant). A further characterising feature of somatotopic representation is a gradient in selectivity, such that neighbouring body parts show greater overlap in cortical activity. This gradient can be observed using block- or event-related designs, where activity for each of the body parts can be assessed independently (C,F). Note that to avoid circular analysis, the activity gradients shown in (C,F) were extracted from independent regions of interest, based on the maps shown in (A,D). A-C was adapted from (Tal et al., 2017); D-E was adapted from Sanders et al. (2019).

Fig. 2

Shifted lip representation in the deprived cortex correlates with phantom limb pain (PLP). (A) Pictures illustrating the sensory stimulation applied over the thumb (top) and the lips (bottom). (B) fMRI activity during sensory stimulation applied over the intact thumb (blue), the lip on the deprived hemisphere (red) and the lip on the intact hemisphere (green) in amputees with PLP (n ​= ​10); projected to one hemisphere. The yelllow line shows a probabilistic deliniation of Broadmann areas 3b and 1 of S1. The stimulations were applied in different sessions in pseudorandomized order. The colored patches show the location of peak activity for the individual patients. The patches were sligthly enlarged (4 ​mm) for visualization purposes. The projections are carried out on a semi-inflated surface (all using surface-based analyses). C) Correlation between PLP severity (based on the Pain Intensity scale of the West Haven-Yale Multidimensional Pain Inventory adapted for PLP and Euclidean distances between the cortical representation of the thumb and the lips in the deafferented hemisphere (r ​= ​−0.79, p ​= ​0.006). These distances are measured between the lip representation (deafferented hemisphere) and intact thumb (with x-axis flipped to match the hemisphere of the lip). They are in mm and are calculated in the folded brain in standard space (i.e. standard brain). Based on HF’s unpublished data.

Fig. 3

Persistent representation of the missing hand. (A) Activity group maps in controls (left) and amputees (right) during movements of the nondominant (controls) or phantom hand (amputees). White circle indicates the position of the anatomical hand knob. (B) A finger-selectivity map (using a travelling wave paradigm) for individual phantom finger movements reveals a complete hand somatotopy in primary somatosensory cortex of an amputee, with specific and adjacent clusters showing selectivity to specific phantom fingers. (C) Centre of gravity of lip activity clusters in individual participants (amputees, orange; controls, purple) reveals a medial shift in amputees’ lip representation, localised to the face area. On average, lips in the deprived hemisphere were shifted medially by 8 ​mm, compared to the intact hemisphere (note that the hand area is located 63 ​mm medially to the lips in controls). Images adapted from: (A) (Makin et al., 2013b); (B) (Kikkert et al., 2016); (C) (Makin et al., 2015b).

Fig. 4

Multiple drivers of remapping in primary somatosensory cortex and phantom limb pain (PLP). These variables have been shown to modulate either S1 organisation, PLP, or both. These multiple factors, and any interactions between them, need to be considered in addition to injury-related changes in any future mechanistic analyses of S1 remapping and PLP. Abbreviations. ACC, anterior cingulate cortex; MCC, midcingulate cortex; PCC, posterior cingulate cortex; Nac, Nucleus accumbens; BG, basal ganglia; PAG, periaqueductal grey; PFC, prefrontal cortex; OFC, orbitofrontal cortex; M1, primary motor cortex; S1, primary somatosensory cortex; SMA, supplementary motor area; SII, secondary somatosensory cortex; PB, parabrachial nucleus; PPC, posterior parietal cortex; Hip, hippocampus; HT, hypothalamus.