Neuroinflammation in chronic pain: Myth or reality?

1 Introduction

There has been a rapid rise in the term “neuroinflammation” (NI) in recent years in medical literature from several disciplines in both clinical and basic science. This has spawned a new journal, the Journal of Neuroinflammation, with a rising citation index now above nine (j neuroinflammation.biomedcentral.com). The increased interest in “neuroinflammation” is currently popular in the field of pain. Junior colleagues often comment that they now know the cause of diagnoses such as fibromyalgia syndrome, depression, headache, complex regional pain syndrome, etc., and attribute NI as the cause of various functional syndromes. Many researchers and clinicians focus on NI without an understanding of the difference between cause, effect, and association in relation to exploratory findings in the research on biomarkers. There is also a lack of consensus on a definition. The concept is like the fable of the blind men and the elephant.

2 Methods

A PubMed search for citations using the search words “neuroinflammation” and “pain” from the years 1983, 1993, and 2023 was undertaken. In the first instance, this was to assess the rise in publications concerning “neuroinflammation” and, secondly, the rise in publications using both terms. A second PubMed search was to find the origin of the term “neuroinflammation” and the mechanisms originally studied to explain the process. A continued search was performed to retrieve articles explaining the evolution of the term to its present all-embracing use.

3 Results

In 1983, there were 101 citations in PubMed using the search word “neuroinflammation.” A review of the abstracts indicated that six were associated with neurodegenerative diseases, and the remainder were all focused on central nervous system (CNS) infection. In 1983, PubMed had no citation that included the search word combination “neuroinflammation” AND “pain.”

In 1993, there were 247 citations in PubMed using the search word “neuroinflammation.” Again, reviewing the first 100 abstracts, six were on chronic farigue syndrome (CFS), four neurodegenerative diseases and the remainder on CNS infection. There were no citations that included the search words “neuroinflammation” AND “pain.”

In contrast, in 2023, there were 2,198 citations in PubMed using the search word “neuroinflammation,” and the search words “neuroinflammation” AND “pain” found 563 citations. A review of the first 100 under “neuroinflammation” focused on degenerative disorders (50/100), CNS injuries (15/100), and infection (13/100). Psychiatric/neuropsychiatric disorders also appear (9/100).

The first 100 “neuroinflammation” and “pain” co-citations are remarkable. 44/100 discuss peripheral neuropathic pain, often using animal models (36/100) for the demonstration of possible mechanisms and propose new treatments focusing on “neuroinflammation.” Some specifically discuss “peripheral neuroinflammation” as opposed to CNS NI as a mechanism in their models. In other areas, arthritis, central neuropathic pain, cancer-related pain (often bone pain), CFS, psychiatric co-morbidities, COVID, and headache also appear.

4 Discussion

As to definitions of NI, there is a list of four criteria suggested as necessary (all four) for the diagnosis of NI [1,2], as it was originally defined relative to brain trauma, stroke, or infection before the recent expansion of the term. These are the following:

1. elevation in pro-inflammatory cytokines and chemokines,

2. activation of macrophages,

3. recruitment of lymphocytes,

4. local tissue damage.

Wikipedia defines NI as “inflammation of the nervous system due to infection, trauma, brain injury, toxic metabolites or autoimmunity,” a further expansion of the term. Perhaps this is an indication of the changing research environment that has now met the ultimate information resource for many.

Somewhat related is the definition by Galea et al.: “the process by which circulating leukocytes and plasma proteins are brought to sites of infection in tissues and are activated to destroy and eliminate the infection. Inflammation is also the main reaction to damaged or dead cells and to the accumulation of abnormal substances in cells and tissues” [3].

Back to the beginning. The Roman writer Celsus (25 BCE to 50 CE) is credited as the first to describe the signs of inflammation:

1. calor (warmth),

2. dolor (pain),

3. rubor (redness),

4. tumor (swelling).

This definition was based on clinical observation only. These signs have been in use in the definition of inflammation arising from injury, infection, exposure to toxins, etc. However, despite its continued use, Celsus’ definition is too simplistic in all instances where there is evidence of inflammation. Also, the “flame” of inflammation implies that all inflammation is harmful, which is not the case, especially after injury or surgery, where the inflammatory response is essential for healing to occur. We are still waiting for a universally acceptable alternative definition of inflammation [4].

There are many further expansions of the concept of NI. DiSabato et al. [5] state that “neuroinflammation is defined as an inflammatory response within the brain or spinal cord.” To further muddy the waters, many articles are now written on NI in the periphery associated with neuroreceptor function at all levels of the nervous system, including the dorsal root ganglion [6]. How should we define the recent concept of “peripheral neuroinflammation”?

The author predicts that the next expansion of the term will refer to NI in the gut as the extensive enteric nervous system is also called “the second brain” [7].

Interestingly, the majority of the publications reviewed for this article do not define NI. There is a definition distinct from that of Di Sabato et al.; however, from Kennedy and Silver [8], which is as follows: the infiltration into brain parenchyma of peripheral immune cells such as T cells, dendritic cells, and peripheral mast cells. The problem with this definition is that it does not pertain to the spinal cord or periphery, where NI is often imputed, and the investigations are not common in the exploration of NI in various pain syndromes. It also does not include any reference to glial cell activation or the presence of cytokines that are considered by many as the evidence for NI.

Is there a more conservative approach than in the current research environment where biomarkers from brain imaging and analysis of serum, plasma, and spinal fluid are the basis for the diagnosis? We should take a step further and look at the accepted but often forgotten hallmarks of NI as clearly set out by Estes et al., based on a discussion by Denes et al., as listed above. This has support from basic scientists who are concerned about the possible imputations of the loosely used term NI [3,9,10].

It is clear from the definition from Wikipedia that the hallmarks of inflammation from Celsus and the hallmarks of NI from Estes, Denes et al. all primarily refer to an acute inflammation response as a reaction to infection, stroke, or trauma. In the case of chronic neurodegenerative disorders such as Alzheimer’s disease or Parkinson’s disease, these apply since there is ample evidence for ongoing tissue (neuron) damage.

But how does the putative evidence for NI relate to what we see in chronic pain and the ubiquitous, associated biopsychosocial comorbidities? A critique of the methods used in the diagnosis identifies major problems [9]. Although the article by VanElzakker et al. highlights research in CFS, the critique holds for all the studies in pain since it criticizes the methods used to identify possible biomarkers in pain states as well.

PET scanning is used to identify glial activation in the CNS as a surrogate for NI as per the above definition. This assumption may be suspect but the critique of the technique needs addressing. VanElzakker identifies the following problems:

1. Standard imaging techniques were not designed for brainstem study

2. The first-generation radioligand PK11195 has a high non-specific binding and low signal-to-background ratio

3. PET signal calculated with an anatomical reference brain region relies on equal radioligand uptake in that region across cases and controls

4. Radioligand access to the brain is modified by general metabolism, which can differ from case and controls

5. Activated peripheral immune cells bind radioligand and can differ in quantity across cases and controls

6. A single nucleotide polymorphism in the TSPO gene causes differential radioligand binding

7. Use of healthy controls harms discriminant validity

Magnetic resonance spectroscopy is used to identify specific brain metabolites indicative of glial activation and glutamate as an indication of activity in various brain regions involved with pain processing. The assumption is that glial activation means the NI is present. How these relate to the inflammatory mediators in the brain is difficult to identify since much of the background data, animal and human, comes from an intravenous injection of pro-inflammatory cytokines or bacterial polysaccharides in normal animals or humans to induce the “sickness syndrome.” These are models for acute systemic illness and do not necessarily relate to chronic pain syndromes.

Cytokine measurement is a very common approach to identifying NI. The remainder of this article will focus mostly on this aspect.

Cytokines include inflammation-related molecules investigated, such as interferon (INF), tumor necrosis factor, chemokines, lymphokines, and interleukins [8]. More and more molecules are being added to the list to be tested. The sample techniques include bioassay and ELISA, but more commonly, multiplex array assays (MAA) that simultaneously measure up to 100 different biomarkers. Sampling is mostly of whole blood, serum, plasma, and cerebrospinal fluid (CSF), but peripheral blood mononuclear cell isolates are also used. These supposedly represent the state of the brain, but this is questionable. When comparing whole blood and blood component sampling to CSF in human and animal models, the data are mixed, with some showing a correlation, but many show no correlation. CSF samples also differ between those taken by lumbar puncture and those taken by cisternal puncture. The fact that cytokines primarily act locally after an injury may be a part of the explanation. Various other technical factors in the handling of samples can influence cytokine levels. For instance, time from sampling to testing and storing temperature (frozen or not) can also affect the results. Comparing different studies using different body fluids at different times after sampling with different storage techniques only confuses the picture. Comorbidities [11] and even age and body mass index [12] are also variables that can influence results.

MAA results are complicated. Various manufacturers have different kits that give different results for the same cytokine on the same sample. Different labs, different personnel, and different protocols give different results, sometimes with no detection on some and positive on others, using the same kit from different lots in the same labs [13,14].

Thus, there is variability causing difficulty in comparing studies. The larger question is if the fluids tested are an indication of the state of the brain. Cytokines do not readily cross the blood/brain barrier unless there is brain trauma, and there should be differences in the levels intra- and extradurally. Cytokines tend to be locally active, and higher levels are likely to be found near the source. Suppose the source is CNS microglia since these are the theoretical source for those with various pain states associated with NI. In that case, it is unlikely that blood samples will reflect what is happening centrally if the CNS microglia are the source. There may be some leakage into the CSF from central sources and the cisternal versus lumbar sample difference could reflect this.

Perhaps this discussion could be considered as just splitting hairs. However, it is difficult to interpret the data when studies differ, and experts cannot agree. If elevated levels of cytokines are detected, irrespective of the differences in analysis technique, are we really dealing with inflammation? Not by the classical definition. There is another definition that one might consider.

“Inflammation is a physiological, tightly regulated, protective response to an underlying infectious or non-infectious process or condition that involves cells of the innate immune system, adaptive immune system, and inflammatory mediators; it is not a pathological response” [4].

This definition was used to describe inflammation in the acute instance with trauma or infection but fits with what is happening in chronic disease. The concept of it not being pathological would also apply to the slight increase in biomarkers for glial activation even seen in stressful life events where infection and/or tissue damage are not present. The point here is that this definition indicates that the process may be normal, even in acute inflammation. It should be viewed as a non-pathological reaction to both “bio” and “psychosocial” threats to the individual. The use of “inflammation,” whether peripheral or central, implies that the “bio” in “biopsychosocial” is the most important since the biopsychosocial model places “bio” first as the driving force [15]. This relates to the outdated concept of body/mind dualism.

This brings up the point of significance of the elevated levels of different cytokines. The levels are less than ten times baseline, but in classical inflammation, the cytokine levels are a hundred to a thousand times over baseline. Here, the definition should give some clues in that the situation is non-pathological. To address this, it is necessary to consider the function of the innate and adaptive immune systems. They are a part of the homeostasis process in the body and have both inflammatory and anti-inflammatory components that are active in the case of infection or trauma. The inflammatory components are necessary to signal tissue damage by activating pain signaling pathways but, more importantly, to begin to stimulate a defense against invading organisms, to stimulate salvage of dead cells and to promote healing. There is no healing without inflammation.

The morphology of glial cells that are the source of cytokines is quite varied and fluctuates constantly. It is not an “on–off” system. In what might be called the resting state, the cells are constantly testing the environment. Glial cells have dendritic-like extensions that grow out and retract depending on the environment. With immune system activation from trauma, infection, or disease, glial cells assume an amoeboid structure with large-scale cytokine production [16]. This is not the case in many of the models that propose to investigate NI.

A possible downside of the focus on NI is that many will assume that it is the cause of various disorders. “Correlation does not imply causation” – David Hume, A Treatise on Human Nature. The cause cannot be imputed where an effect is hypothesized but not adequately assessed. The findings of putative NI in association with a host of biopsychosocial diagnoses (economic deprivation [17], bullying [18], stress [19], mental health disorders [20], CFS [8], including pain [21,22], are only associations. There are also attempts to describe central sensitization as a type of NI [23,24]. For assessment of causality, we need to follow a strict protocol as proposed by Hill [25] but seldom considered in the NI literature. These include: Strength (effect size), Consistency (reproducibility), Specificity, Temporality, Dose–response relationship, Plausible mechanism, Coherence (with laboratory findings), Experimental evidence (if available), and Analogy (with other causal processes).

In discussions with clinical colleagues, it seems many assume that the cause of the above clinical diagnoses plus neurodegenerative diseases is NI. In these instances, some also consider it as a mechanism, and some also consider it a diagnosis. Attempts to treat NI with anti-inflammatory medications and biologicals used in autoimmune diseases have been proposed and sometimes tried with little success [26,27]. This misuse of terms and pharmacology is likely to continue if the putative presence of MI continues to be the explanation/cause for so many biopsychosocial maladies, including pain. This is a waste of time, effort, money and is potentially associated with severe side effects in the patients treated.

5 Conclusions

The expansion of the theory that glial activation is a sort of pathological NI is misguided. Without glial activation in the fetus and in childhood, normal CNS function and even normal social function would be impaired [28]. The same would apply to acute inflammation, where inflammation is necessary for healing after accidental trauma, surgery, and infection. This is not a pathological process.

Any threat to the homeostasis of the body can activate microglia, not only in the CNS but also in the periphery. This low level of activation can be viewed as priming the microglia to proceed to the mode seen with classical inflammation, central or otherwise. The homeostasis protection system is alert, observing and waiting to go into action against a major threat to the organism. “Microglial activation” is a better term for this state, as seen in many contexts, including various pain diagnoses [29]. Equally important is to view this as an association only since neither cause nor effect can be assessed in sampling done at a single time point in the clinical picture.

In the final analysis, we have several different definitions of inflammation and NI, all proposed by experts based on basic and clinical research and modified to fit many hypotheses. However, we must also consider Wilbert “Bill” Fordyce’s view that “Expert opinion is often wrong!” We are no longer using leeches to cure cancer.