Proteomic Analysis to Identify Prospective Biomarkers of Treatment Outcome After Microvascular Decompression for Trigeminal Neuralgia: A Preliminary Study

Abstract

Trigeminal neuralgia (TN) is a severe neuropathic facial pain disorder, often caused by vascular or neuronal compression of the trigeminal nerve. In such cases, microvascular decompression (MVD) surgery can be used to treat TN, but pain relief is not guaranteed. The molecular mechanisms that affect treatment response to MVD are not well understood. In this exploratory study, we performed label-free quantitative proteomic profiling of plasma and cerebrospinal fluid (CSF) samples from patients undergoing MVD for TN, then compared the proteomic profiles of patients graded as responders (n = 7) versus non-responders (n = 9). We quantified 1090 proteins in plasma and 1087 proteins in the CSF, of which 12 were differentially regulated in the same direction in both sample types. Functional analyses of differentially regulated proteins in protein-protein interaction networks suggested pathways of immune system, axon guidance, and cellular stress response to be associated with response to MVD. These findings suggest potential biomarkers of response to MVD, as well as possible mechanisms of variable treatment success in TN patients.

Introduction

Trigeminal neuralgia (TN) is a chronic pain disorder characterized by intense neuropathic pain in the distribution of the trigeminal nerve. In many cases, TN is caused by compression of the trigeminal nerve by an overlying structure, such as an artery, vein, or other nerve. Microvascular decompression (MVD) is a neurosurgical procedure in which a suboccipital craniotomy is performed, allowing access to the trigeminal ganglion near the brainstem, and a small piece of Teflon is inserted to lift the compressive structure off the trigeminal nerve.⁶ MVD is highly effective, with approximately 75% of patients experiencing long-term pain relief.²² However, pain relief is not guaranteed, and complications of MVD include infection, cerebrospinal fluid leakage, facial paralysis, hearing loss, and, in rare cases, death.⁵⁰ The pathophysiologic mechanisms contributing to poor pain relief following MVD are not well-characterized, and there are few objective indicators that reliably predict successful MVD. Factors reported to be associated with worsened pain relief outcomes after MVD include shorter duration of disease, atypical (dull, aching) pain rather than typical (sharp, stabbing) pain, lack of immediate postoperative relief, increased severity of neurovascular compression, and the presence of focal arachnoiditis.^6,37,41,48 Although the incidence of TN increases with older age and female sex, clinical variables such as age, sex, race, duration of TN, and preoperative medications are not consistently associated with MVD outcome, and many putative predictive factors, such as postoperative relief, severity of compression and arachnoiditis, cannot be fully assessed until surgery has already been performed. Currently, TN patients and surgeons must rely on patient symptoms and inferences from neuroimaging to guide whether a patient is likely to benefit from MVD. The discovery and validation of pain biomarkers would yield new insights regarding the mechanisms of analgesia following MVD and suggest objective markers to predict TN treatment success.

An emerging approach in biomedical research uses proteomic profiling to quantify variations in protein expression (i.e., “proteomes”) corresponding to different diagnoses or treatment outcomes. In contrast to traditional pain treatment approaches that rely on trial-and-error, clinical observations, and a priori assumptions regarding underlying pain mechanisms, a proteomic approach to pain medicine investigates the active molecular pathways that are altered in painful disease states and applies that knowledge to optimize diagnosis, prognosis, and treatment. Proteins are attractive biomarkers because they are key effectors of any disease process, and changes in their abundance, structure, or function can herald or influence pathology even before clinical manifestations are observed. Since most drugs act on proteins, proteomics may also identify potential pharmacological targets. Previous studies of genetic and molecular mechanisms in TN have suggested that ionotropic channels, oxidative stress, and/or neuroinflammation could play a role in TN pathophysiology;¹² proteomic studies would provide additional evidence for these prospective pathways or identify others. In this exploratory study, we analyzed the proteomic profiles of plasma and cerebrospinal fluid (CSF) in patients undergoing MVD for TN to identify markers associated with treatment response.

Methods

Participants

The current study utilized a subsample from a larger, prospective, observational study of predictive factors for MVD outcome. Following Institutional Review Board approval, participants scheduled to undergo MVD surgery at Johns Hopkins Hospital were recruited and enrolled after written informed consent. Inclusion criteria were patients at least 18 years of age, able to provide informed consent, meeting International Classification of Headache Disorders (ICHD-3) diagnostic criteria for TN, who failed to achieve adequate pain relief from oral analgesics, and were considered by the surgeon (M.L.) to be appropriate surgical candidates for MVD. Exclusion criteria included patients unable to communicate in English, which was a requirement for the larger study.

Procedures

Prior to MVD surgery, enrolled participants answered demographic questions, including age, sex, race, characteristics of the TN pain, coexisting pain conditions, and medications (Table 1). During the surgical procedure, immediately after opening of the dura, CSF was aspirated from directly around the trigeminal nerve, and whole blood was drawn from a peripheral vein into an EDTA tube. Specimens were placed on ice and centrifuged within 30 minutes of collection, then CSF and plasma were stored in 500-ul aliquots at −80°C until assayed. Participants were then followed at 3 months postoperatively to assess response to MVD using the Barrow Neurological Institute (BNI) Pain Intensity Score,¹⁰ a widely used ordinal rating scale of MVD outcomes. Patients with BNI I (no pain, no medications), BNI II (occasional pain, no medications), and III (some pain, adequately controlled with medications) were considered as MVD responders, while BNI IV (some pain, not adequately controlled with medications) and BNI V (severe pain or no pain relief) were considered as MVD non-responders. A convenience sample of the lowest-scoring (i.e., most successful, BN I-II) responders and the highest-scoring (i.e., least successful, BN IV-V) non-responders for whom there was complete outcome data was used for proteomic analysis.

Table 1.

Demographics of Patient Cohort

	Responders (n = 7)a	Non-Responders (n = 9)b	p-valuec
Barrow Neurological Institute (BNI) Pain Intensity Score, mean ± SDd	1.4 ± 0.5	4.4 ± 0.5	<0.0001
Age (years), mean ± SD	52.1 ± 8.5	47.7 ± 19.2	0.58
Sex, n (%)
Female	5 (71%)	8 (89%)	0.55
Male	2 (29%)	1 (11%)
Race, n (%)
Black	1 (14%)	2 (22%)	0.59
White	6 (86%)	5 (56%)
Other/Decline to Answer	0	2 (22%)
Duration of Symptoms (years), mean ± SD	4.9 ± 3.7	6.3 ± 5.7	0.58
Average Pain Score (0–10), mean ± SD	8.1 ± 1.6	7.7 ± 2.5	0.72
Pre-Operative Medications, n (%) e
Over-the-Counter/Anti-Inflammatory	0	3 (33%)	0.21
Anticonvulsants	6 (86%)	5 (56%)	0.31
Antidepressants	1 (14%)	4 (44%)	0.31
Muscle Relaxants	0	4 (44%)	0.088
Opioids	1 (14%)	3 (33%)	0.59

^aBNI Pain Intensity Score at 3 months = I, II, or III (no patients in this sample had score of III)

^bBNI Pain Intensity Score at 3 months = IV or V

^cBased on t-test for continuous outcomes and Fisher’s exact test for categorical outcomes.

^dBNI I/1 = no pain, no medications, II/2 = occasional pain, no medications, III/3 = some pain, adequately controlled with medications, IV/4 = some pain, not adequately controlled with medications, V/5 = severe pain or no pain relief

^eMedications classified as over-the-counter/anti-inflammatory (acetaminophen, ibuprofen), anticonvulsants (carbamazepine, oxcarbazepine, gabapentin, pregabalin), antidepressants (duloxetine, nortriptyline), muscle relaxants (baclofen, cyclobenzaprine), opioids (hydrocodone, oxycodone)

Proteomic Analysis

Plasma and CSF were subjected to immunodepletion of abundant proteins using the Proteome Purify 12 protein immunodepletion resin according to the manufacturer’s instructions (R&D Systems, Minneapolis, MN). Immunodepleted samples were washed, reduced, alkylated, and trypsinized on filter as previously described^16,47 and implemented by this group,^11,23,26 allowing for increased sample recovery and quantification of secreted, cytosolic, and membrane proteins. Proteomic profiling was conducted using liquid chromatography-tandem mass spectrometry where tryptic peptides were separated on a BEH130 C18, 1.7 μm, 75 μm × 200 mm column (Waters Corporation, Milford, MA) over a 165-minute linear acetonitrile gradient from 3 – 40% with 0.1 % formic acid using a nanoACQUITY UPLC (Waters Corporation, Milford, MA) coupled to an Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Scientific, San Jose, CA) as previously described.^11,23,26,46 Full scans were acquired at a resolution of 240,000 and precursors were selected for fragmentation by collision induced dissociation (normalized collision energy at 35 %) for a maximum 3-second cycle. Tandem mass spectra were searched against a UniProt Homo sapiens reference proteome using the Sequest HT algorithm¹⁵ and MS Amanda algorithm¹³ algorithm with a maximum precursor mass error tolerance of 10 ppm. Carbamidomethylation of cysteine and deamidation of asparagine and glutamine were treated as static and dynamic modifications, respectively. Resulting hits were validated at a maximum false discovery rate of 0.01 using a semi-supervised machine learning algorithm, Percolator.⁷ Label-free quantification was performed using Minora, an aligned accurate mass and retention time (AMRT) cluster quantification algorithm (Thermo Scientific, 2017). Protein abundance ratios between two groups were measured by comparing MS1 peak areas of peptide ions, with identity confirmed by MS2 sequencing.

Criteria used to select significant proteins were greater than twofold change in protein expression, with a Benjamini-Hochberg corrected false discovery rate (FDR) cutoff of 0.05. Functional analyses were performed in STRING⁴³ and MCODE,⁵ two widely used open-access tools to assess protein-protein interaction networks. Since one intent of this study was to identify potential non-invasive (plasma) biomarkers that would be reflective of the trigeminal nerve milieu, we specifically focused on proteins that were differentially expressed in the same direction in the plasma and CSF.

Results

Plasma and CSF samples from 7 MVD responders and 9 MVD non-responders were used for analysis. Basic patient demographics are provided for descriptive purposes in Table 1.

A total of 1087 proteins were quantified in the CSF samples, and 1090 proteins were quantified in the plasma samples. Of these, 62 CSF proteins were significantly different between responders compared to non-responders (i.e., twofold difference in gene expression at FDR-adjusted p < 0.05), while 136 plasma proteins were significantly different.

Twelve proteins were differentially regulated in the same direction between responders and non-responders in both CSF and plasma; ten proteins corresponded to known (characterized) proteins, while two proteins are uncharacterized. These proteins and their abundance ratios in non-responders vs. responders are shown in Table 2.

Table 2.

Proteins Differentially Regulated in the Same Direction in CSF and Plasma

Gene Name	Description	Plasma Abundance Ratio (NR vs. R)	Plasma Adjusted P-Valuea	CSF Abundanc e Ratio (NR vs. R)	CSF Adjusted P-Valuea
Protein More Abundant in Non-Responders Compared to Responders
C4A	Complement C4-A	7.945	7.70745E-17	8.118	0.001674683
TUBB3	Tubulin beta-3 chain	1000	7.69745E-17	1000	2.12697E-16
IGKV1–8	Immunoglobulin kappa variable 1–8	3.733	4.82893E-08	1000	2.12697E-16
DPYSL3	Dihydropyrimidinase-related protein 3	1000	7.69745E-17	1000	2.12697E-16
CLIC1	Chloride intracellular channel protein 1	1000	7.69745E-17	1000	2.12697E-16
WDR1	WD repeatcontaining protein 1	1000	7.69745E-17	1000	2.12697E-16
(None) Accession F5H423	Uncharacterized protein	1000	7.69745E-17	5.46	0.018410269
CAP1	Adenylyl cyclase-associated protein 1	1000	7.69745E-17	1000	2.12697E-16
(None) Accession A0A0G2JRQ 6	Uncharacterized protein (Fragment)	1000	7.69745E-17	1000	2.12697E-16
Protein Less Abundant in Non-Responders Compared to Responders
SHISA7	Protein shisa-7	0.001	7.69745E-17	0.001	2.12697E-16
NEU1	Sialidase-1	0.001	7.69745E-17	0.001	2.12697E-16
SORCS1	VPS10 domain-containing receptor SorCS1	0.001	7.69745E-17	0.001	2.12697E-16

^aAdjusted by Benjamini-Hochberg false discovery rate (FDR) correction

Nine proteins were more abundant in non-responders compared to responders, corresponding to the following genes: C4A, TUBB3, IGKV1–8, DPSYL3, CLIC1, WDR1, CAP1, and two uncharacterized proteins corresponding to gene accession numbers FSH423 and A0A0G2JRQ6. Three proteins were more abundant in responders compared to non-responders, corresponding to the following genes: SHISA7, NEU1, and SORCS1. Representative box-and-whisker plots for the normalized protein abundance of selected differentially regulated proteins in responders compared to non-responders are shown in Figures 1–3, respectively.

Figure 1.

Complement C4A relative abundance, plasma (left) and CSF (right).

Figure 3.

SHISA7 relative abundance, plasma (left) and CSF (right).

Functional analyses of all differentially regulated proteins in CSF and plasma were used to identify potential protein-protein interaction pathways that may be involved in response to MVD. A summary of the top pathways identified using STRING and MCODE algorithms for CSF and plasma is provided in Table 3, along with FDR-adjusted p-values. In both STRING and MCODE analyses for CSF samples, the most significant pathway was for proteins related to axon guidance. In plasma samples, both STRING and MCODE analyses found that differentially regulated proteins in responders compared to non-responders were most significantly associated with neutrophil-related and immune-related pathways. STRING analysis of plasma samples also identified axon guidance, while MCODE analysis of plasma found that differentially regulated proteins were associated with tissue morphogenesis and regulated exocytosis pathways.

Table 3.

Summary of most significant pathways identified in protein-protein interaction networks.

Functional Analysis	Sample Type	Pathways	Adjusted P-Valuea
STRING	CSF	Axon guidance	0.00013
		Neutrophil degranulation	1.27E-07
	Plasma	Immune system	2.08E-06
		Innate immune system	4.58E-06
		Axon guidance	0.0063
MCODE	CSF	Axon guidance	5.01E-10 7.94E-10
		HSP90 chaperone cycle for steroid hormone receptors
		Tissue morphogenesis	3.16E-09
	Plasma	Neutrophil-mediated immunity	1.58E-08
		Regulated exocytosis	3.16E-08

^aAdjusted by Benjamini-Hochberg false discovery rate (FDR) correction

Discussion

MVD responders and non-responders had significant differences in protein abundance in the CSF surrounding the trigeminal ganglion, associated with similar differences in plasma. A previous proteomic study of only lumbar CSF samples found differential expression of proteins involved in high-density lipoprotein and the complement cascade between TN patients compared with controls.¹ An earlier study from the same group found increased expression of inflammation-related proteins in the lumbar CSF of TN patients compared to healthy controls, and these levels decreased after MVD surgery.¹⁷ The current study similarly obtained proteomic profiles in TN patients but used both plasma and CSF directly surrounding the trigeminal ganglion of TN patients and further stratified based on treatment outcome. Our findings suggest that pathways involving axon guidance, neuroinflammation and neurotransmission may be associated with surgical treatment outcomes in TN.

Neuro-Immune and Neuroinflammatory Proteins

Several differentially regulated proteins identified play a role in the immune system. The complement system is a group of related proteins that activate the innate immune response to pathogens and tissue injury.⁴⁵ Complement dysregulation has a putative role in neuroinflammation and nociceptor sensitization in some chronic pain states with neuropathic features, including complex regional pain syndrome (CRPS), postsurgical pain, peripheral nerve injury, and chemotherapy-induced peripheral neuropathy, with most studies focusing on complement factors 3 and 5.⁴⁵ C4A is the acidic component of complement factor 4, which has not been as well studied in the context of pain, but has been associated with fibromyalgia²⁵ and myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS), which often features pain symptoms similar to fibromyalgia, particularly after exertion.^32,42 One early study of CFS patients found that exercise exacerbated CFS symptoms and was associated with elevated blood levels of C4A; no other complement proteins were significantly different from healthy controls, leading the authors to suggest that C4A could be a biomarker for CFS.⁴² However, a subsequent study found that post-exercise C4A was not significantly different in CFS patients versus healthy controls, but in CFS patients, the post-exercise C4A levels were positively correlated with postexertional malaise (including bodily pain).³² A small study found that C4A is elevated in the CSF of fibromyalgia patients but not rheumatoid arthritis patients, leading the authors to suggest that C4A may be associated with symptoms of fatigue commonly seen in fibromyalgia.²⁵ It has been proposed that complement activation may differentiate chronic widespread pain, including fibromyalgia, from healthy controls.¹⁸ Our study found significantly increased expression of C4A in MVD non-responders, suggesting that general inflammation (or perhaps concomitant nociplastic pain) may be associated with poorer surgical outcomes.

Immunoglobulins are the polypeptide antibodies produced by B cells as part of the adaptive immune system. They are famously comprised of two identical light (kappa or lambda) chains and two identical heavy chains, all containing constant and variable domains. The immunoglobulin kappa variable domain IGKV1–8 does not have a prominent presence in the pain literature, and given the nature of the adaptive immune system, it is unlikely to play a unique role in TN treatment response. Notably, intravenous immunoglobulin has been proposed as a treatment option for neuropathic pain, including trigeminal neuralgia, although the evidence is not robust and large-scale trials are lacking.^9,19

Sialidase 1 (also known as neuraminidase 1, NEU1) is a lysosomal enzyme whose deficiency is associated with sialidosis, a genetic disorder that variably manifests as visual deficits, myoclonus, ataxia, seizures, and/or developmental abnormalities.⁴⁰ Although there is no clear association between Neu1 and pain, there is some evidence that neuraminidases play a role in immune activation, leukocyte recruitment, and neuroinflammation. Recent in vitro studies have found that microglia activated by endotoxin/lipopolysaccharide have increased cell surface translocation and release of Neu1.^2,3 Increased cell surface expression of Neu1 enhances the release of pro-inflammatory cytokines and other immunomodulatory molecules that promote inflammatory activation of microglia.³ Release of Neu1 facilitates microglia phagocytosis and neuronal sensitization to glutamate, which can precipitate neuronal death.² Decreased Neu1 activity is therefore associated with decreased inflammation, suggesting that some neuroinflammatory response may be necessary for successful response to MVD. This contrasts with our findings of decreased innate immune response with C4 levels in MVD responders versus non-responders, with the implication of a more nuanced relationship between various aspects of the immune system and MVD response.

Another differentially regulated protein, CLIC1 or chloride intracellular channel 1, has been associated with neuroinflammation and neurodegeneration. An in vitro study found that β-amyloid protein, which forms the hallmark pathological plaques of Alzheimer’s disease, promotes translocation of cytosolic CLIC1 to the plasma membrane, enhancing transmembrane chloride ion conductance and generating damaging reactive oxygen species.²⁹ A more recent study found that peripheral blood mononuclear cells (PBMCs) isolated from Alzheimer’s disease patients showed overexpression of CLIC1 mRNA and increased transmembrane CLIC1 compared to healthy controls, leading the authors to suggest CLIC1 as a marker of chronic inflammation in the central nervous system (CNS).⁸ Another study found that increased blood RNA levels of CLIC1 were correlated with degree of cognitive decline in Alzheimer’s disease.²⁸ Thus, although it has not been widely reported in the pain literature, CLIC1 may be a biomarker for CNS inflammation and neurodegeneration, and the elevated levels observed in MVD non-responders could be indicative of ongoing inflammatory or neurodegenerative processes.

Proteins Related to Axonal Growth and Function

Other proteins identified in our analysis are involved in axonal growth and function. TUBB3 is the only β-tubulin isoform (βIII-tubulin) present in all neurons, making it a classical neuronal marker with a presumed role in neuronal structure and function via microtubule formation.²⁷ Although absence of TUBB3 does not correspond to any neurobehavioral or neuropathological deficits in a knockout mouse model, it does appear to be necessary for normal nerve fiber growth and peripheral axon regeneration that cannot be replaced by any other β-tubulins.²⁷ Interestingly, a recent study has found that capsaicin-induced depolymerization of axonal microtubules, detected by levels of unpolymerized free tubulin in dissociated sensory neurons, is associated with analgesia in trigeminal neuropathic pain.⁴

Dihydropyrimidinase-related proteins (DPYSLs, also known as collapsin response mediator proteins (CRMPs), Unc-33-like phosphoproteins (Ulips), or dihydropyrimidineamidohydrolase-related proteins (DRPs), are phosphoproteins that have been increasingly recognized as important contributors to neuronal development and function, including axon guidance and inflammation.³¹ DYPSL3, more commonly known as CRMP4, has been shown to facilitate the growth of hippocampal axons and dendrites through interaction with actin and tubulin.²⁴ By contrast, a knockout mouse study found that deletion of CRMP4 had a neuroprotective effect in spinal cord injury that was associated with decreased inflammatory response and decreased scar formation.³⁰ The authors characterized CRMP4 as an axonal growth inhibitor and enhancer of scar formation, suggesting that CRMP4 deletion could provide an environment conducive to axonal growth and recovery after spinal cord injury. In our clinical cohort, increased abundance of TUBB3 and DPYSL3 were associated with MVD non-responders, suggesting that appropriate axonal degeneration and regeneration contribute to MVD success.

Two proteins involved in actin dynamics were more highly expressed in MVD non-responders. WD (tryptophan-aspartic acid) repeat protein 1 (WDR1), also known as actin-interacting protein 1 (AIP1), enables actin filament binding and is an important regulator of processes dependent on actin dynamics, such as neutrophil chemotaxis and neurotransmission.^33,36 In a knockout mouse model, WDR1 deficiency in the hippocampus was associated with alterations in associative learning and synaptic plasticity.⁴⁴ CAP1 (cyclase-associated actin cytoskeleton regulatory protein 1 or adenylyl cyclase-associated protein 1) is a ubiquitous protein involved in the cyclic AMP pathway and interactions with actin. Like WDR1, CAP1 appears to play a role in actin organization and dynamics, and its deficiency is associated compromised neuronal connectivity, defects in hippocampal neuron differentiation, and dysfunctional hippocampal growth and motility.³⁹ However, the roles of WDR1 and CAP1 in pain and pain treatment remain unclear.

Proteins Involved in Neurotransmission

The last category of differentially regulated proteins identified in our analysis are associated with neurotransmission. Shisa7 is a transmembrane protein that has only recently been characterized. It interacts with GABA-A receptors, facilitating GABAergic transmission, and has been demonstrated to enhance the action of benzodiazepines in the brain.²¹ Furthermore, Shisa7 appears to be involved in hippocampal plasticity and inhibitory long-term potentiation, with a rodent model of dysfunctional Shisa7 exhibiting hyperactivity and impaired sleep homeostasis.⁴⁹ SORCS1 belongs to a family of neuropeptide receptors that represent some of the largest known human genes and are strongly expressed in the brain.²⁰ The protein SorCS1, or sortilin-related CNS expressed 1, is a sorting receptor that regulates trafficking of important neuronal surface proteins, including the synaptic adhesion molecule neurexin and the AMPA glutamate receptor.³⁸ Absence of SorCS1 has been found to alter the balance of axonal and dendritic neurexin in the same neuron, leading to impaired presynaptic differentiation and function.³⁵ In addition, decreased SORCS1 expression has been associated with increased amyloid precursor protein and β-amyloid in Alzheimer’s disease.³⁴ Thus, decreased expression of Shisa7 and SorCS1 in MVD non-responders could suggest synaptopathy or impaired neurotransmission contributing to treatment response.

General Themes

Protein-protein interaction analyses were consistent with the above themes of neuro-immune, axonal, and neurotransmission pathways as potential contributors to MVD response. Importantly, proteins involving axon guidance were significantly differentially expressed in both CSF and plasma, suggesting that this pathway may be of particular interest.Major limitations of this study include the small sample size and relatively short time frame of post-surgical follow-up (3 months). In addition, patient factors such as co-morbid mental health conditions and chronic stress may have affected both treatment outcome and proteomic profiles, but these were not assessed in detail in this exploratory study. Larger cohorts and consideration of other potential covariates may reveal additional protein expression profiles and pathways involved in treatment outcome. Moreover, although most patients’ MVD outcomes are clear within days to weeks after surgery and tend to persist for months to years afterward, it must be acknowledged that some patients with apparent treatment success in the early postoperative period may eventually progress to treatment failure, or vice versa. However, because there is a correspondence between short-term and long-term success, the intermediate term of 3 months was felt to be a reasonable surrogate for longer-term outcomes.

In the pathway to successful biomarker implementation in TN treatment, this study represents the initial steps in target identification (discovery) and target confirmation (verification) ¹⁴. Although our findings suggest some proteins that are differentially expressed in TN responders versus non-responders, larger studies are necessary to determine prospectively whether these proposed biomarkers can accurately predict TN responder status, either individually or in combination as a composite biomarker. It is also noteworthy that many of the proteins identified are typically cytosolic or membrane-bound, rather than secreted, suggesting that the study of intact cells through single-cell proteomics (e.g., peripheral blood mononuclear cells) could provide additional perspective on pathways relevant to TN and its treatment.

Conclusion

This preliminary study identified protein markers associated with treatment outcome in both the CSF and plasma of TN patients undergoing MVD. Given previous reports that proteins involving myelination and inflammation are differentially expressed in the CSF of TN patients versus healthy controls,^1,17 the proteomic differences found in our study suggest pathways that may be associated with both presence and severity of TN pathology. Thus, processes such as neuroinflammation, axon guidance, and neurotransmission may be promising targets, both to identify which patients are most likely to benefit from surgical treatment in TN, and to understand the underlying mechanisms that lead to successful outcomes.

Figure 2.

TUBB3 relative abundance, plasma (left) and CSF (right).

Highlights.

• Need biomarkers to predict surgical response in trigeminal neuralgia

• Proteomic studies of plasma and cerebrospinal fluid may yield potential biomarkers

• Detectable proteomic differences between surgery responders and non-responders

• Proteins involved in neuroinflammation, axonal function, and neurotransmission

• Insight on possible mechanisms of trigeminal neuralgia and treatment response

Perspective.

This exploratory study evaluates proteomic profiles in plasma and cerebrospinal fluid of patients undergoing microvascular decompression surgery for trigeminal neuralgia. Differential expression of proteins between surgery responders versus non-responders may serve as biomarkers to predict surgical success and provide insight into surgical mechanisms of pain relief in trigeminal neuralgia.