In February of 1880, Frederick William Pavy, a physician and researcher at Guy’s Hospital in London, met with a patient whom he had first seen nine years earlier. The woman, 48, had been diabetic for nearly two decades when she began to feel pain emanating from her legs. By April, she was uncoordinated in her gait.
This patient was not alone. In an address to the British Medical Association in 1885, Pavy would reveal that many of his patients with diabetes experienced persistent pain—and grew numb—in their limbs, ultimately becoming disabled. And yet, their pain did not come in just one form.
“Darting or ‘lightning’ pains are often complained of,” Pavy told his audience. Or, “it might be that the patient is unable to bear the contact of the seam of a dress against the skin on account of the suffering it causes,” he said. “Not unfrequently there is deep-seated pain, located, as the patient describes it, in the marrow of the bones…”
Pavy was not the first to identify a link between diabetes and changes in the nervous system; the earliest mention of this phenomenon dates back to the end of the eighteenth century. Still, he captured the complexity of what doctors now call painful diabetic neuropathy (PDN), a condition in which diabetes not only leads to nerve damage but also to chronic pain.
Today, diabetes is a global epidemic. According to the International Diabetes Federation (IDF), there were 415 million adults worldwide living with the disease in 2015. By 2040, that number is predicted to reach 640 million. And while estimates vary, about a third of these people will suffer from pain.
“It’s a very interesting question as to why some patients with diabetes develop neuropathic pain and some don’t,” says David Bennett, a pain researcher at the University of Oxford, UK.
“One explanation is quite simple,” he says. “It seems to us that patients with severe neuropathic pain also have more severe neuropathy—the more damage done to your sensory nerve fibers, the more likely you are to get neuropathic pain.”
But over the years, researchers have recognized that even patients with similar levels of nerve damage are not all alike. It’s now clear that they can have different “sensory profiles”—constellations of sensory symptoms, including different kinds of pain. And, as animal studies suggest, each type of pain results from changes scattered throughout the peripheral (outside of the brain and spinal cord) and central nervous systems. In the future, a better understanding of these alterations may even lead to more personalized treatment for PDN.
In 2009, Ralf Baron at Christian-Albrechts-Universität Kiel, Germany, systematically confirmed what Pavy’s observations had only hinted at. He and his colleagues provided approximately 500 PDN patients with a survey asking them about the nature of their symptoms. The researchers then analyzed if and how different types of pain and other abnormal sensations clustered together. In doing so, they could disentangle five distinct groups.
One of the largest consisted of patients who were numb in one or more parts of their body—what’s known as a “sensory loss.” Paradoxically, they also reported a tingling sensation, in addition to burning pain—each a “sensory gain.” Others instead said they had sudden pain attacks.
Another group, albeit a smaller one, said that their threshold for pain had dropped, such that even harmless touch now hurt. Baron and his colleagues concluded that PDN patients could be stratified on the basis of their self-reported symptoms.
That doesn’t mean, however, that patients will necessarily fall into one of these five groups. “How many groups you pull out depends a bit on how you define your groups,” says Bennett.
Indeed, by testing the sensitivity of PDN patients across a range of sensations, Bennett and his colleagues recently parsed out two main clusters. In a study published last year, most of the patients the researchers examined had both spontaneously occurring pain and a general sensory loss, while a minority felt pain when their skin was lightly brushed.
Patients can even find themselves in more than one group. Using a new sorting algorithm, Baron and his colleagues have now shown that people with PDN can have symptoms that span multiple sensory profiles.
These issues aside, the differences between PDN patients have shifted how researchers think about pain in diabetes. The question of ‘What causes pain?’ has now become ‘What causes pain in each patient?’ There is no one-size-fits-all explanation.
From stratification to better treatment?
At the moment, the damage done to nerves in diabetes cannot be reversed. While there is some regeneration, “it’s not usually enough to make the neuropathy go away,” says Brian Callaghan, a neurologist at the University of Michigan, Ann Arbor, US.
So, when it comes to treatment, patients can do two things, he says. They can prevent further nerve damage by controlling their blood sugar levels, and they can also try to manage their pain using medications such as antidepressants and anticonvulsants (anti-seizure drugs).
“They all seem to work relatively similarly—not nearly as well as we would like, but they work,” says Callaghan.
Still, they do not always work. In 2010, researchers at Aarhus University, Denmark, compiled findings from over 100 clinical trials on how likely these drugs were to reduce pain from nerve damage, including from diabetic neuropathy.
They found that for antidepressants, the number of patients who needed to be treated for one patient to experience decent pain relief ranged from about 2 to 7; this “number-needed-to-treat” (NNT) is a common metric used in clinical studies to assess the impact of a drug. For anticonvulsants, the NNT was somewhere between 2 and 6. In other words, many PDN patients may not benefit from these medications.
That may not come as too much of a surprise, since these drugs were approved for PDN without testing which sensory profiles would respond best to which particular treatment. But in the future, that strategy could change.
Andrew Rice, a pain researcher at Imperial College London, UK, proposes, for example, that researchers could first figure out whether patients have become more or less sensitive to touch before giving them a drug. Then, once patients are stratified in this way, they could ask: “Does that sensory profile predict the underlying pain mechanisms, and therefore, does it predict the response to treatment?”
There’s reason to think so. In a 2014 study, 83 patients with nerve damage from a variety of causes were stratified into two groups. The first was made up of 31 patients with heightened sensitivity to pain but not to touch or temperature, while the second included 52 others without that sensory profile. They instead experienced numbness or no sensory symptoms at all (at least, those that were measured). As it turned out, oxcarbazepine, an anticonvulsant, was more likely to relieve pain in the former group than in the latter.
This has not been the only hint that patient stratification could lead to better treatment. Last year, a small clinical trial reported that botulinum toxin (a neurotoxin commonly known by its trade name, Botox) blunted pain to a greater extent in patients with a set of symptoms almost identical to those most helped in the oxcarbazepine study.
According to research from Bennett and Rice, this sensory profile—a normal response to harmless touch and temperature but exaggerated pain—is one that rarely shows up in those with PDN (it’s more common in other forms of nerve damage). Nonetheless, such studies suggest that grouping PDN patients according to their symptoms could perhaps unveil the benefits of drugs that would otherwise be underappreciated or hidden entirely when tested in clinical trials.
“It’s a very intriguing idea that needs to be further fleshed out before it really has an impact at the bedside,” says Callaghan.
Where pain begins
Despite the promise of this approach, researchers still know very little about how changes in the nervous system give rise to the different kinds of pain observed across PDN patients. What they have learned has been largely gleaned from animal models in which diabetes is experimentally triggered in various ways.
In the streptozotocin (STZ) model of PDN, a chemical is used to destroy the cells that make insulin, ultimately leading to diabetes and pain. In this model, rodents show striking changes in their pain-sensing neurons. For example, researchers led by Stephen Waxman, a neurologist at Yale University School of Medicine, New Haven, US and the Veterans Affairs Medical Center, West Haven, Connecticut, showed in 2002 that these cells contain greater amounts of proteins that help determine whether a neuron will send pain signals to other neurons in the nervous system. Collectively, these proteins are known as sodium channels.
One sodium channel, dubbed Nav1.7, has long been of interest to Waxman. “There are mutations [in the gene that makes] Nav1.7 that are pathogenic,” he says, “and we see them in patients with painful diabetic neuropathy.”
These genetic mutations can increase the activity of Nav1.7. Other genetic variants of Nav1.7, which have similar effects on the protein, are responsible for the chronic pain of erythromelalgia and paroxysmal extreme pain disorder, two rare disorders. Building on these findings, efforts are now underway to design new pain drugs that block Nav1.7 (see recent RELIEF related feature article).
Another sodium channel, Nav1.8, also seems to contribute to PDN. In 2012, a research team at University Hospital Heidelberg, Germany, reported that, compared to ten diabetic patients without pain, ten with pain had higher levels of a glucose metabolite called methylglyoxal. And, in the presence of methylglyoxal, pain-sensing neurons isolated from normal mice—and which contained Nav1.8—were more likely to produce electrical signals.
A follow-up study from another group of researchers, however, has cast doubt on the idea that methylglyoxal levels are tied to pain, as it found no relationship between the two in over 1,000 patients. While that leaves the role of methylglyoxal in PDN unclear, Nav1.8 is likely still involved. Its levels surge in pain-sensing neurons of STZ-treated rats, and its activity is influenced by that of Nav1.7.
According to Waxman, Nav1.7, Nav1.8, and other sodium channels act in concert in pain-sensing nerves. “They echo off each other,” he says.
Into the central nervous system
Deep within the spinal cord, there are other changes associated with PDN. Andrew Tan, a pain researcher at Yale University School of Medicine who has collaborated with Waxman, has studied the “spines” that sprout from neurons there. Specifically, they sprout from the cells’ branch-like structures called dendrites, which receive information from other neurons. Tan’s work shows that alterations in the spines are linked to pain in rats treated with STZ.
“The idea is that whenever you learn something, dendritic spines change their shape,” says Tan.
Tan and Waxman have also found that the brains of rats treated with STZ show important changes. In a study published in 2009, they recorded the activity of a small cluster of neurons nestled within the thalamus, which is a brain region that relays pain signals from the spinal cord up to the cerebral cortex. Over time, these neurons became more electrically excitable, producing stronger responses when an animal’s paw was gently brushed, pressed, or pinched.
How much can animal models of PDN really say?
The STZ model is far from perfect, however. As Rice points out, that model, and in fact all animal models of PDN, have yet to replicate the type of pain that most often arises in patients—that which is ongoing and spontaneous but strangely associated with numbness.
To better understand such pain, studying the reflexive responses of rodents—a withdrawal of the paw to touch or heat—won’t do, says Rice. “You have to look at complex behaviors. They might stop behaving as they normally do.”
For example, they might be less likely to engage in “naturalistic” behaviors such as burrowing into the ground, building a nest, or exploring their environment. Indeed, Rice has found that rats with nerve damage from an HIV-related protein or the anti-HIV drug stavudine rarely venture out into the open when placed into an unfamiliar area, possibly to avoid predators that could more easily catch injured prey.
Still, it’s difficult to know the extent to which rodents’ behaviors, naturalistic or otherwise, are representative of pain in people. Moreover, as clinical studies suggest, PDN is not monolithic, meaning that identifying an animal model for each and every kind of pain that patients experience could prove to be challenging.
Animal models are critical for developing new drugs, so it’s also not clear when more personalized treatment for PDN will arrive. For the time being, Bennett and his colleagues are delving deeper into the differences between patients, specifically their genetic dissimilarities.
They now want to know: If certain genetic mutations predispose patients to developing pain, “what is the sensory profile in those patients?” says Bennett.
Matthew Soleiman is a science writer currently residing in Nashville, Tennessee. Follow him on Twitter @MatthewSoleiman