Phantom limb pain is a real physiological condition. So why do we tolerate mystery and myth when it comes to treatment?

by Tamar R. Makin

This essay originally appeared in the July 2021 edition of the academic journal Brain (Volume 144, Issue 7), under the title “Phantom Limb Pain: Thinking Outside the (Mirror) Box.” It is republished here under Creative Commons Attribution License 4.0. The author is a professor of cognitive neuroscience at University College London.

Phantom limb pain (PLP) is a curious phenomenon. It raises challenging questions relevant to what it means for us to live inside our bodies, and has thus been a source of wonder and curiosity throughout modern culture. René Descartes liked to use PLP as a cautionary example for why the human senses cannot be trusted. Admiral Horatio Nelson, who lost his arm in 1797, took his phantom sensations as evidence for the existence of his eternal soul. More recently, millions of viewers of hit TV shows such as Grey’s Anatomy and House have sympathized with patients (and doctors) struggling to cure PLP. Phantom limb pain has even been the topic of an action-adventure stealth video game.

Although PLP was once considered to be rare, most recent accounts estimate that about two-thirds of amputees experience it. Despite being very common, PLP is notoriously difficult to treat with conventional medicine. The unusual challenge we are faced with is that the body part to be treated is not physically present. A mechanistic understanding of the neural basis of PLP is thus needed to treat it successfully.

Why do people experience PLP? Early observations showing that PLP can be evoked by applying pressure to the stump led to the theory that it may relate to the peripheral nerves. This was elegantly demonstrated using intraneural recordings from the residual limb of people who had undergone an amputation. Even though the receptors of the peripheral nerve are missing, the residual axons still transmit signals. Both spontaneous and evoked PLP are reflected in the electrical activity of the residual nerve.

This finding inspired a simple mechanistic explanation for PLP: Because peripheral nerves normally provide information about touch and pain originating from the extremity, inputs provided by these nerves will be interpreted by the central nervous system as arising from the missing limb. Although clinical attempts to use local anesthesia to block this activity have proved difficult to implement, this approach has produced rapid and reversible attenuation—and often complete elimination—of PLP. This provides a powerful demonstration that PLP originates in the periphery.

And yet this simple mechanism has been largely marginalized in comparison to more ambiguous explanations based on psychopathology or mechanisms in the brain.

Brain Reorganization and Phantom Limb Pain

Many theories dominating the early 20th century assumed that PLP was neurotic in nature, manifested by “denial” or even “hysteria.” For example, R.D. Langdale Kelham, a pioneer in post-amputation rehabilitation, concluded that the typical patient with phantom limb sensation was, more often than not, someone with an “unsatisfactory personality.” He argued: “It may be he is an anxious, introspective, dissatisfied, ineffective [sic] who, becoming obsessed by his symptoms, and brooding upon them and his disability, tends to dramatise their degree, using undoubted exaggerations in his description of his sufferings.”

Such theories have been conclusively debunked. Other investigators have considered the anatomical origins of PLP to lie in the sensorimotor central nervous system. This possibility paved the way for a flurry of surgical interventions in the latter half of the 20th century, with relatively poor clinical outcomes.

Since the end of the 20th century, the prevailing theory for the development of PLP has been that of maladaptive brain plasticity. This idea is based on an observation in monkeys where loss of input to the brain’s hand area leads to redistribution of brain resources. Under this theory of brain plasticity, or reorganization, the cortical resources of the (now missing) hand become freed up and get taken over by a new body part. You might expect that the brain’s ability to reassign resources across body parts based on altered demand should be helpful, and perhaps even allow people who have lost a limb to better adapt to their disability. An example of adaptive plasticity would be early-blind individuals, where the visual cortex becomes involved in non-visual processing for perception and language.

However, according to the maladaptive plasticity theory of PLP, reorganization in the adult brain can be harmful. This idea is rooted in an observation that a relatively crude measure of brain reorganization in amputees correlates with PLP. A theory evolved that phantom limb pain arises from misrouted neural inputs, generating an “error” signal that the brain interprets as pain signals from the missing limb. This theory provides clear predictions on how to treat PLP: If pain is caused by maladaptive reorganization, then we need to reverse the reorganization to alleviate PLP.

Some of today’s most widely used treatments for PLP aim to achieve this reversal by reinstating the representation of the missing limb to its original territory. Mirror-box therapy operates on this principle. Related approaches, using implicit and explicit motor imagery, aim to gradually “reawaken” the motor representation of the missing limb, while virtual reality-based treatments aim to achieve a similar effect via improved phantom motor execution. All these techniques target neuroplasticity mechanisms as a means to reverse harmful rewiring and reinstate “normal” sensory and motor perceptions.

However, a closer examination of the theory of maladaptive neuroplasticity reveals a number of unsupported assumptions. First, the notion that amputation “erases” the representation of the missing limb, leading to cortical reorganization, has been negated. Multiple lines of evidence demonstrate that the brain retains the representation of the missing limb despite its being physically absent, and advanced neuroimaging techniques fail to find foreign signals entering the brain territory that corresponds to the missing limb. In other words, there is no need to “reinstate” the representation of the limb, because it persists after amputation.

Second, the idea that a foreign input causes pain by triggering an error signal in the brain has also been refuted. For example, researchers have injected very small electrical currents directly into the brain to artificially stimulate the somatosensory limb territory in patients with severe nerve injuries. According to the theory of maladaptive reorganization, this should result in an error signal, potentially giving rise to pain. But instead, this procedure triggers non-painful sensations on the insensate limb. These findings are consistent with results from studies involving amputees, where stimulating the part of the motor cortex corresponding the missing limb does not cause pain. Instead, pain sensations are more closely linked to brain areas and neural networks distinct from the sensorimotor network.

Where’s the Clinical Proof?

Perhaps the most compelling evidence against the theory of maladaptive reorganization theory is its poor clinical outcomes. This theory has become assertively dominant over the past three decades, with the original paper reporting a correlation between brain plasticity and PLP being cited almost 2,000 times in the scientific literature. Consequently, its therapeutic derivatives have dominated clinical practice. According to a recent international survey, four of the six most widely recommended PLP treatments (including both pharmacological and non-pharmacological options) are based on reversing maladaptive plasticity. Yet PLP is still a common condition, and the overwhelming consensus across clinical trials, systematic reviews, and meta-analyses is that there is no strong evidence that these clinical approaches provide consistent and long-lasting PLP relief, beyond a placebo control.

Then why are we continuing to use these unsuccessful therapies? Their limited efficacy is exacerbated by the fact that much of the first-level evidence supporting these treatments is compromised. To begin with, PLP and its relief are ultimately measured by subjective report, which is fundamentally susceptible to suggestion and biases. Without a direct comparison to a double-blind placebo-controlled study arm, any observed changes in PLP reports should be treated with skepticism. Yet this gold standard is rarely adopted in PLP research.

A further challenge is that PLP phenomenology is diverse, and therefore studies aimed at tracking PLP tend to use multiple pain scales. This becomes a problem when researchers cherry-pick a particular outcome measure post hoc, without accounting for the multiple potential comparisons that have been performed.

A third problem relates to the mechanisms of pain alleviation. Some of the newest virtual/augmented reality treatments use principles from gamification to make the therapy more engaging, but attentional distraction is known to have pain-relieving benefits. Let’s consider an ongoing clinical trial, where phantom movements are used to control a video game in a virtual/augmented environment. This intervention is compared to a control condition where participants are asked to imagine moving the phantom but not engage in or control the game. Any benefits incurred by the main treatment might be attributable to distraction arising from the game, rather than resulting from the therapy itself.

Since no effective PLP treatment is currently available, one might argue that there is no harm in providing patients with suboptimal treatments. Indeed, the placebo effect is extremely powerful and could be harnessed to ease the suffering of individuals struggling with intractable pain. But we should also consider the consequences of deliberately developing and using suboptimal treatments. From an ethical standpoint, if we know the treatment is not more effective than a placebo, we should make this explicitly clear to the patient and the clinical team. This is especially true when the treatment might be expensive or time-consuming for the patient.

From a policy perspective, the development and assessment of treatments based on the theory of maladaptive plasticity has consumed an enormous share of the resources available to the small community that’s pursuing targeted PLP treatments. This misallocation leaves fewer research and innovation opportunities available to identify potentially more successful treatments that aren’t rooted in the maladaptive reorganization theory. Rather than rehashing unsuccessful treatments, we should instead work toward practical methods to suppress the PLP generators that have been identified.

Unfortunately, we still don’t have a consensus understanding of the neural drivers of PLP. We don’t even know if this condition is mechanistically any different from other more common, and arguably less romanticized, chronic pain conditions, such as joint pain. But considering how futile our focus on maladaptive brain plasticity has been so far, it is time for us to shed our romantic notions around this pain condition and start thinking outside the (mirror) box.