Brigitte Brouwer

General discussion and future perspectives

Outcome measures and clinical relevance
As mentioned in the introduction, in order to perform scientifically well-based research it is paramount to use reliable, valid and responsive outcome measures that fulfill modern clinimetric requirements. Only then we can interpret trials in the most proper way. In SFN, we do not know completely which outcome measures are optimal, since no longitudinal data are available examining the responsiveness of the various known metrics.

Most outcome measures are ordinal scales (based on the classical test theory (CTT)). CTT has several disadvantages, and scores do not represent a true numerical value, hampering statistical analyses [32, 33]. Both Rasch and Item Response Theory (IRT) models have the possibility of transforming ordinal scores to interval data. The IRT approaches include additional model parameters to reflect the patterns observed in the data [34, 35]. According to the Rasch approach, the data should fit the Rasch model, before any reliable claim about the presence of a trait can be made. Therefore misfitting responses require further examination to explore the reason for the misfit, and may be excluded from the data set if one can explain substantively why they do not address the latent trait [35, 36]. Using the Rasch method is challenging, since it demands an educational bridging to physicians less known how to handle data in an appropriate way. Based on knowledge over the last two decades, we noticed that clinicians are generally unfamiliar with handling the transforming methods like IRT and Rasch, and may encounter problems interpreting the logistical scales (logits). However, transforming the data to a convenient centile metric, will help physicians in digesting the obtained scores much better. If we really want to implement this method we should aim to change the mindset of clinicians and researchers in this field and we should compare these psychometric approaches through different familiar Patient-Reported Outcome instruments.

The SFN-RODS© is the first interval outcome measure for SFN on the activity and participation level. However, some questions may have become outdated such as ‘are you able to use dustpan and brush together’ and ‘are you able to fill in a form’. However, the questions can also be used as capabilities: ‘would you be able to ….’, and therefore the actuality of items may seem less relevant. An instruction manual is currently being constructed facilitating in the interpretation and scoring of each item of this metric.

Furthermore, the capability of the SFN-SIQ© and the SFN-RODS© to capture changes over time (responsiveness) should be determined. However, we recommend to use the SFN-RODS© as a measurement for activity and participation in future clinical trials, because this is the first and only outcome measure at this level. A similar concept has been executed in patients with inflammatory neuropathies using the Inflammatory-RODS (I-RODS), an outcome measure designed using the same principles of modern clinimetric as adopted for the construction of SFN-RODS [37-39]. Longitudinal data of patients being treated were used to determine its responsiveness, hereby introducing the concept of minimum clinically important difference (MCID), taking into account the varying standard error across the metric constructed through Rasch methodology. The SFN-RODS is currently being used in trial designs in SFN to determine its responsiveness [40].

Cross-cultural adaption of a questionnaire is essential when using a questionnaire in other countries and cultures; only a translation is not enough. Guillemin et al wrote in 1993 a guideline to reserve equivalence in cross-cultural adaption of health-related quality of life measures [41]. For the SFN-RODS we did not check the cross-cultural validity yet, which will also be part of future trial designs [38, 39].

Currently no activity and participation item bank for SFN exists. An item bank consists of questions (items) that are unidimensional (in other words measuring the same concept). In near future it should be possible to create unidimensional item banks based on the conceptual model of the ICF (see figure 1). New questions/items can be calibrated and outdated items can be removed.

Furthermore it is important to define responders by MCID taking into account the varying standard errors at the individual level (MCID-SE) [37]. In clinical research, it is common practice to compare one group with the other. However responsiveness at the individual level may be even more important in daily life; a meaningful change for the group may be not as important/relevant for the individuals in the group. Besides this, a meaningful change may be different between individual patients, e.g. a 85-year old patient may define relevant clinical improvement different from a 25-year old athlete. We therefore may want to use more personalized items, relevant for the patient. By using an item bank these relevant items can be arranged for every level of difficulty. With this statistical model it is for example possible to use computer adaptive testing (CAT), by which the software selects questions based on earlier answers. A next step would be to let the patients themselves tell us what they would see as a sufficient amount of change on their health status for a treatment to be qualified as successful.

PART-QPO©
We have developed an overall PDN Anxiety Rasch-Transformed 30-item questionnaire (PART-QPO©). For this study, patients with a DNS score ≥ 1 and neuropathic pain in the feet (and DN4 ≥ 4) for at least 3 months and being clinically stable were included. In this chapter, we only mentioned the DNS questionnaire. However, we also used the Douleur Neuropathique 4 Questions (DN4-interview). The DNS has been validated for diabetic polyneuropathy, but not for painful diabetic polyneuropathy [42]. Neuropathic pain was assessed using the interview section of the Douleur Neuropathique 4 Questions (DN4-interview). This interview consists of seven items relating to the pain description (Burning, painful, cold, electric shocks) and to its abnormal sensations (tingling, pins and needles, numbness and itching) and has a high diagnostic accuracy for painful diabetic polyneuropathy (sensitivity 74%, specificity 79%) [43, 44]. Furthermore, we included patients with diabetes only by letting them fill in the Diabetic Neuropathy Symptom (DNS) Score (self-report diagnosis). No physical examination was performed to diagnose a painful diabetic neuropathy, and it has been questioned whether these patients truly have diabetic neuropathy. However, based on a recent Dutch study, symptoms alone may be helpful in the diagnosis of polyneuropathy (Hanewinckel et al, submitted).

Measuring pain
The consistency of pain scores over 4 weeks with little if any influence of circadian rhythm contradicts with literature findings, making its clinical relevance difficult to interpret [45-47]. In our design, we only asked for mean pain during night and day. Others have asked for pain three times a day, and did not ask for mean pain, which may or may not explain the bigger differences in chronobiology in these studies [46, 48]. However, the latter studies were performed in patients with neuropathic diabetic and post herpetic pain and not specifically in patients with SFN. Our study findings may be fine-tuned by considering other ways to collect data such as random measurements (with identical questions). The advantages of repeated measurements are capturing symptom fluctuations, possibility to analyze triggers and the increase of power of measurement. The use of the experience sampling method, a digital questioning method for assessing symptoms and influencing factors during consecutive days, can decrease recall bias, ecological bias and gives a higher compliance rate [49-52]. This method, or other e-health modules, may also be used to personalize treatment management specifically designed to the patient’s life style, providing direct feedback on specific symptoms and patterns and can be applied as well as a self-management tool. In future, we can use this method to study whether pain in SFN has a circadian rhythm, or an infradian rhythm (such as seasonal rhythms), these data can be associated with data about activity, temperature, mental states (such as anxiety and depression), and biological measurements such as cortisol, melatonin and catecholamines.

Linking stress with chronic pain
Research shows a strong association between conditions such as anxiety and neuropathic pain [53, 54]. The corticolimbic system in the brain is considered to be the bridge in linking stress with chronic pain [55]. But is there another link? All sensory nerves from the peripheral nerve system are derived from neural crest cells. Interestingly paraganglia, also called chromaffin cells, also derive from the neural crest. The cells of paraganglia lie often in association with sympathetic ganglia and are also located in the medullary cells of the adrenal glands. These chromaffin cells express Nav1.7 channels [56-58]. Neuropathic pain seems to be a stronger inducer of basal corticosterone levels than long-term induced stress in mice but both result in higher levels of corticosterone production [54]. In female patients with depression and anxiety increased levels of norepinephrine was found in 24-hour urine [59]. Future studies could therefore focus on chromaffin cells as a possible treatment target in SFN patients with or without a proven Nav1.7 gain-of-function variant. Furthermore, gene-editing may be an alternative treatment possibility, such as silencing these specific voltage-gated ion channels, as was described with adeno-associated virus carrying a short hairpin RNA [60].

Bodily distress and autonomic symptoms
The symptoms of the SFN-SIQ© are also symptoms that can be seen in patients with (other) bodily distress syndrome, such as transpiration, diarrhea, dry mouth, palpitations [61]. Although the peripheral autonomic nerves are affected in SFN, one could question whether SFN is more than a small fiber disease alone and autonomic symptoms could be caused by other factors. Bodily distress is seen in multiple functional disorders in which there is an altered autonomic balance, a dysfunction of the stress system (sympathetic and hypothalamic-pituitary adrenal (HPA) axis) and altered inflammatory response and sensitization of the nervous system [62-66]. The question remains whether these different labeled conditions are a spectrum of multifactorial biopsychosocial etiologies [67, 68].

Point of no return?
Despite the evidence for the involvement of sodium channels in neuropathic pain in SFN patients, their exact pathogenic role in SFN is not entirely elucidated. Current research focuses on resolving the genetic architecture of painful neuropathy in order to 1. achieve a stratification of persons with a high risk for neuropathic pain by novel biomarkers and 2. to enhance our understanding of underlying mechanisms, circuitries and target druggable sites and 3. to identify new molecules on potentially responder patients.

The question also rises whether there is a point of no return in chronic pain. In other words, can we see the transformation of “no pain” to “chronification of neuropathic pain” as the analogy of the sublimation from water to ice, which is reversible, or do we have to look for mechanisms such as denaturation (a point of no return)? And if so, could the chronification phase switch or activate various parts of the (central) nervous pain system, orchestrating a “perpetual” mobile pattern? The question is whether the use of very specific drugs is enough to break through the complex mechanisms (sensitization) of chronic neuropathic pain. It is plausible that we need to target more than one specific part of the “pain system” because of the many changes within the pain-circuit. The WHO and American Pain Society (APS) recommend multimodal analgesia in the treatment of pain. The theory is that targeting multiple sites will act in an additive or synergistic way, and may lead to better efficacy and maybe a better action [69].

As mentioned earlier, CRMP2 could be another possible target. CRMP2 is important in neurite outgrowth, neuronal polarization, but it also regulates the surface expression of voltage gated calcium channels (VGCC) and voltage-gated sodium channels (VGSC) [70, 71]. The antiepileptic drug Lacosamide, for example, binds on CRMP2 on five binding pockets; this binding may alter the action of the VGSC [72]. Other possible candidate drugs for multimodal targeting may be selective sodium channel blockers [73], tropomyosin receptor kinase A inhibitors [74, 75], anti-nerve growth factor antibodies [76-78], immunoglobulins [40], or TRP receptor antagonists [79]. Special caution is necessary for the use of opioids in chronic neuropathic pain [80].

Searching for analogies
What can we learn from the pathologies of other diseases? When we unravel the genetic code, the fundamental units are the five nucleobases Adenine (A), Cytosine (C), Guanine (G), Thymine (T) and Uracil (U). Using this metaphor, one may wonder what the similarities are between, at first sight, different pathologies?

For example, can we compare chronic pain with a status epilepticus; in other words, is chronic pain a “status doloris”? There are some similarities for example in the key players GABAA-R and NMDA-R in both conditions. The first step in treating an epileptic seizure is influencing gamma-aminobutyric acid A (GABA-A) receptors by treating the patients with benzodiazepines. Benzodiazepines bind to GABA-A receptors at a binding site that is different than GABA [81]. This binding increases affinity for GABA and facilitates the opening of GABA-activated chloride channels. The possible analgesic effects in neuropathic pain is also believed to work by antagonizing hyperexcitability, through the enhancement of inhibitory GABAergic pathways, which has an effect in acute pain, such as labor pain and in acute neuropathic pain (animal) models [82-85]. However it does not work for chronic neuropathic pain. During a status epilepticus the GABA-A receptor becomes internalized, causing decreased sensitivity for benzodiazepines after long-lasting seizures. This may implicate that comparable changes in GABA-A receptors occur in chronic pain as in status epilepticus.

Another similarity is the moving of N-Methyl-D-aspartic acid receptors (NMDAR) from the interior of the cell to the synaptic and extrasynaptic wall, which causes more excitability. This is the case in both chronic pain and in status epilepticus. In refractory status epilepticus (RSE) ketamine, a non-competitive NMDA receptor antagonist is used for blocking NMDA receptor mediated glutamatergic neurotransmission [86, 87]. Ketamine, a NMDA receptor antagonist for pain is used for analgesic and antihyperalgesic effects in acute (e.g. postoperative) as well as chronic pain management (e.g. complex regional pain syndrome) [88]. In future oral NMDA antagonists should be developed, and maybe the oral NMDA antagonist felbamate is a potential drug for animal experimental research.

When talking about status epilepticus we speak of hours (among refractory status epilepticus (RSE) the mortality rate is as high as 23-61%), talking about chronic pain we speak about 3 months at least, and changes of the receptors already occurred, possibly explaining the disappointing effect of pain medication.

Prevention
One of the questions in future research should not only be how to treat chronic pain, but also how to prevent chronification of pain by searching for predictors like genes and other factors. Furthermore, chronification of pain itself is probably associated with changes in the expression of multiple genes and channels. At current stage, we don’t have any effective strategy for prevention, early diagnosis and tailored treatments for pain in SFN. This would likely require a wide integration based on individual biological variations, such as gender, aging, genetics, epigenetics and pharmacogenomics, as well as a variety of personal, clinical, societal, and occupational determinants that may influence onset and evolution of chronic pain and response to analgesics.

Changing perception towards pain
How can we change the perception towards pain and the concept of health? The traditional WHO definition of health was: a state of complete physical, mental and social well-being and not merely the absence of disease or infirmity [89]. The new perception of health is the ability to adapt and to self-manage, in the face of social, physical and emotional challenges [90]. This new concept opens new opportunities for patients and caregivers. It would be interesting to approach the SFN patient through the six dimensions of health (bodily functions, mental functions and perception, spiritual existential dimension, quality of life, social and societal participation and daily functioning), from the perspective of the patient. This would require a change in our health organization. We need patient information that is accessible worldwide; we need to speak “one universal language” to standardize not only our diagnostics, but also our explanation towards the patients.

Multidisciplinary pain teams should organize the patient care within a network with the patient in the center. We should look for decision tools that patients can use prior to the admission of the outpatient clinic in order to prepare a custom made team for this specific patient, such as a combination of a neurologist, psychologist and rehabilitation doctor or the combination of anesthesiologist and psychiatrist. It is remarkable that we consider pain as a combination of somatic and emotional problems and yet we still diagnose these separately.

E-health and decision support systems
Neuropathic pain treatment is often difficult, probably because of multiple factors such as failure of medication verification country wide. Most patients visiting the outpatient clinic are not aware of what kind of neuropathic pain medication they are using or have used in the past. Especially, they are not aware of the dosages they took and why it was stopped. A nation-wide digital decision support system could enable clinicians and patients to: 1) Make tailored decisions per individual within the framework of e.g. individual contraindications and wishes; 2) Record medication use, in order to have an adequate overview of medications used, effect of this medication, and potential side-effects, which will contribute also to reduction of costs due to less re-prescriptions; 3) Understand the best treatment for the individual; and last but not least 4) have the opportunity to collect large amounts of data in order to find information that may help us in prevention and developing new treatment facilities.

For discovering new therapies we need to be inventive and we need to think out of the box! Who would have thought in 1870 that submarines were real (Jules Verne), or in the 60s that a video-conference could be for real (USS Enterprise).

Who knows, maybe in future we can erase painful memories like in Eternal Sunshine of the Spotless Mind.