Cutaneous nerve fiber and peripheral Nav1.7 assessment in a large cohort of patients with postherpetic neuralgia

Supplemental Digital Content is Available in the Text.

Bilateral skin biopsies from a large cohort of patients with postherpetic neuralgia (PHN) showed no correlation of intraepidermal nerve fiber density with baseline pain and confirmed that concurrent loss of contralateral nerve fibers is common. Nav1.7 immunolabeling revealed no increase in PHN-affected skin compared with that in the contralateral side.

Abstract

The mechanisms of pain in postherpetic neuralgia (PHN) are still unclear, with some studies showing loss of cutaneous sensory nerve fibers that seemed to correlate with pain level. We report results of skin biopsies and correlations with baseline pain scores, mechanical hyperalgesia, and the Neuropathic Pain Symptom Inventory (NPSI) in 294 patients who participated in a clinical trial of TV-45070, a topical semiselective sodium 1.7 channel (Nav1.7) blocker. Intraepidermal nerve fibers and subepidermal Nav1.7 immunolabeled fibers were quantified in skin punch biopsies from the area of maximal PHN pain, as well as from the contralateral, homologous (mirror image) region. Across the entire study population, a 20% reduction in nerve fibers on the PHN-affected side compared with that in the contralateral side was noted; however, the reduction was much higher in older individuals, approaching 40% in those aged 70 years or older. There was a decrease in contralateral fiber counts as well, also noted in prior biopsy studies, the mechanism of which is not fully clear. Nav1.7-positive immunolabeling was present in approximately one-third of subepidermal nerve fibers and did not differ on the PHN-affected vs contralateral sides. Using cluster analysis, 2 groups could be identified, with the first cluster showing higher baseline pain, higher NPSI scores for squeezing and cold-induced pain, higher nerve fiber density, and higher Nav1.7 expression. While Nav1.7 varies from patient to patient, it does not seem to be a key pathophysiological driver of PHN pain. Individual differences in Nav1.7 expression, however, may determine the intensity and sensory aspects of pain.

1. Introduction

Postherpetic neuralgia (PHN) is the most common complication after acute herpes zoster (shingles) virus reactivation. Prior reports of punch biopsies collected from PHN-affected skin revealed reductions in intraepidermal nerve fiber (IENF) density of up to 80% compared with the contralateral side.^28,30,31,39 Some studies also noted reductions in IENF contralateral to zoster.²⁷ Most small-caliber IENF are considered nociceptors, and loss of these IENF would seem counterintuitive for pain conditions. Of importance, across these other biopsy studies, there was marked variability in the percentage of fibers lost from study to study and from patient to patient within each study.^27,28,38,39 Patients differed based on the amount of cutaneous nociceptor degeneration; loss of nociceptors was reported to be inversely correlated with allodynia^5,16,39 and pain severity.²⁸ After acute herpes zoster, those patients who later developed PHN had lower numbers of remaining nerve fibers on the PHN side than pain-free patients,²⁸ but improvement in PHN pain after 6 months did not track with the regrowth of nerve fibers. Regenerative capacity of intraepidermal fibers is sparse,³¹ and late recovery after 7.7 years was highly variable.³⁴ Regardless of innervation status, however, these patients all developed PHN, demonstrating that this type of neuropathic pain is not correlated with IENF density alone and that additional mechanisms are involved.

Voltage-gated sodium channels (Nav) have been implicated in pain mechanisms, particularly the Nav1.7 subunit. Autosomal recessive loss-of-function mutations in the gene that codes for Nav1.7 (SCN9A) are seen with congenital insensitivity to pain.^11,19 Inherited pain disorders, such as erythromelalgia and paroxysmal episodic pain disorder, as well as idiopathic small fiber neuropathy, are associated with gain-of-function mutations.^13,14,15 Nav1.7 is abundantly expressed in rodent dorsal root ganglion (DRG) small somata whose axons are the presumptive small diameter C nociceptor and Aδ nociceptor IENFs and in peripheral nerve fiber endings in skin.⁹ Nav1.7 is involved in the amplification of subthreshold sensory stimuli and is regarded as the key determinant of nociceptor excitability.¹³ Nav, including Nav1.7, have also been identified on skin epidermal keratinocytes (the terminal cells of IENF), which modulate nociceptor signaling.^7,43,49

Although several drug development programs targeted Nav1.7, the efficacy of systemic Nav1.7 inhibitors has been modest.³ A preliminary phase 2 crossover trial in patients with PHN with a topical formulation of the semiselective Nav1.7 inhibitor TV-45070 showed good tolerability and suggestive signals of efficacy.³² Subsequent double-blind placebo-controlled phase 2 clinical trials were conducted with the same topical compound in PHN and knee osteoarthritis. In both trials, TV-45070 failed to demonstrate significant analgesic effect.^42,44

We analyzed bilateral skin biopsies collected in the phase 2 PHN trial with TV-45070 to assess cutaneous nerve fiber density and Nav1.7-positive nociceptive fiber density from zoster-affected and homologous contralateral sites and sought correlations with spontaneous pain and mechanical hyperalgesia, measured with quantitative sensory testing, as well as patient symptoms captured in the neuropathic pain symptom inventory (NPSI). Our hypothesis was that cutaneous Nav1.7 may be overexpressed in patients with PHN.

2. Methods

This report is based on a retrospective analysis of a phase 2, randomized, double-blind, placebo-controlled, parallel-group study to evaluate the safety and efficacy of topically applied TV-45070 (4% and 8% wt/wt ointment) in patients with PHN. Male and female individuals aged 18 to 80 years, less than 10 years after the appearance of herpes zoster rash, with a clinical diagnosis of persistent PHN pain for >6 months, and an average daily pain score of ≥4 on the 11-point numerical rating scale (NRS), who also had intact skin over the treatment area were eligible for inclusion. Postherpetic neuralgia affecting trigeminal dermatomes was exclusionary. Verification of the initial acute herpetic infection or if the medical history could not be obtained, evidence of treatment with antivirals for an acute painful rash was required. Patients were genotyped by Q² Solutions (a division of Quest Labs; Valencia, CA) for the rs6746030 allele, a gain of function, single-nucleotide polymorphism of the SCN9A gene, which codes for Nav1.7, that has been linked to slower Nav1.7 channel inactivation and lower pain thresholds.³⁵ The polymorphism A allele contains a substitution of tryptophan for arginine at position 1150 of Nav1.7 (R1150W); G denotes the common allele. The phase 2 study (TV-45070-CNS-20013) was sponsored by Teva Pharmaceuticals, conducted in collaboration with Xenon Pharmaceuticals, and registered on clinicaltrials.gov (NCT02365636). The study was approved by a central ethics committee, performed exclusively in the United States in full accordance with ICH, GCP, and all applicable national and local laws and regulations, and subjects provided informed consent.

2.1. Pain assessments

2.1.1. Baseline pain

Patients entered their daily average and worst pain scores into an electronic diary; those with an average daily NRS score <4 over the 7-day baseline period before randomization or a score of 10 on any day during the baseline period were excluded. Also excluded were patients with another explanation for their pain, such as diabetic neuropathy or fibromyalgia, or those who had used topical capsaicin within 6 months before screening.

At the randomization visit, after recording baseline pain measurement for 7 days in e-diaries and before receiving any treatment, patients were assessed for allodynia and mechanical hyperalgesia. Patients also completed the NPSI, a self-reported questionnaire to assess symptoms of neuropathic pain, with 10 descriptive questions (rated 0-10, highest possible score = 100) and 2 temporal items.⁴

2.1.2. Punctate hyperalgesia and allodynia

At screening, patients identified the location of most severe hyperalgesia. To assess the intensity of punctate or mechanical hyperalgesia, a Medipin (US Neurologicals, LLC/Medipin Ltd, Bushey, Hertfordshire, United Kingdom) was applied in 3 successive perpendicular applications to the most painful area of skin identified at screening. The patient then assessed the pain intensity using an 11-point NRS to answer the question, “Please rate the intensity of pain caused by application of the Medipin, where 0 = not painful at all and 10 = worst pain possible.” The highest pain rating of the 3 applications was recorded.

Brush-evoked allodynia was assessed at screening by mapping the area of allodynia using a standardized 1″ foam brush and then assessing the intensity of allodynia by performing 3 brush strokes within the mapped area of allodynia. The patient then assessed pain intensity using an 11-point NRS to answer the question, “Please rate the intensity of pain caused by brushing the area of skin where 0 = not painful at all and 10 = worst pain possible.” The pain was rated for intensity (using the highest pain rating reported) both at prebrush testing and during application of the brush and recorded; in addition, the difference between the 2 scores was calculated and recorded.

2.2. Skin biopsy analyses

Patients underwent one 3-mm skin punch biopsy from the area of maximal PHN pain and the other from a contralateral homologous (ie, identical dermatome) site. The biopsies were sectioned as described further for quantification of validated Nav1.7 immunolabeling as coexpressed with PGP9.5 in the epidermal and adjacent subepidermal nerve fiber (SENF) density.

2.2.1. Skin biopsy fixation and sectioning

Limitations of onsite logistics for preparing on-demand freshly made fixative, patient scheduling and rescheduling, and off-hour personnel availability were a challenge in this large multisite clinical trial. Standardization of biopsy preparation can affect immunolabeling for a wide variety antibodies such as anti-Nav1.7. To assess these potential variables, a study was conducted in collaboration with MD Biosciences, Neurology R&D Division (Nes-Ziona 74140, Israel) in which 40 closely spaced biopsies of pig skin were assessed, comparing 3 different anti-Nav1.7 antibodies, cold fixation with PLAP or 4% paraformaldehyde (PFA), different durations from 4 to 12 hours fixation, and with fixative that was made fresh or that was stored in the cold without air contact for a shelf life of 1, 2, 3, or 4 weeks. Nav1.7 immunolabeling of nerve fibers was best with PFA, and was comparable in 4 to 12 hours, and with fixative no more than 1 week old, stored under refrigeration.³⁷

This protocol was used for patient skin biopsies, which were then rinsed in 3 changes of fresh PBS and transferred to fresh PBS for storage at 4°C. Exact times for each biopsy specimen into and out of fixative were recorded. Biopsy specimens were shipped on cold packs (unfrozen) by overnight courier to Integrated Tissue Dynamics (INTiDYN; Rensselaer, NY) for ChemoMorphometric Analysis (CMA) processing. Fixatives, buffers, biopsy tools, and shipping materials were supplied to each clinical site by a third-party provider (Therapath, New York, NY) following specifications provided by INTiDYN, including that all specimens remain refrigerated (ie, unfrozen). PFA fixative was to be used within 1 week of receipt at the clinical site, according to CMA protocol. Integrated Tissue Dynamics was fully blinded to location of each biopsy because biopsies were designated only as right side or left side at the clinical sites before shipment.

On arrival at INTiDYN, biopsies were designated with identifiers that further blinded the study site sources before sectioning and analysis. The biopsies were cryoprotected by overnight immersion in 30% sucrose in PBS. Biopsies were then imbedded in Optimum Cutting Temperature (OCT) gel, snap frozen, and mounted to a cryostat cutting platform chuck. Biopsies were then cryostat serial sectioned at 14 µm thicknesses in a plane perpendicular to the skin surface. Sections were mounted with consecutive sections rotated across 10 slides to allow for multiple immunolabel combinations, as per CMA protocol. Sections from the left and right biopsies from the same patient were mounted on separate rows on the same 25 × 75-mm slide to assure immunolabeling consistency between both biopsies. Biopsies were excluded from the study if they visibly lacked the epidermis or dermis (ie, biopsy tissue not fully intact) or the fixation time exceeded CMA protocol limits.

2.2.2. Immunolabeling

Based on prior evaluations of various commercial and proprietary antibodies directed at human and homologous rat Nav1.7 epitopes,³⁶ 2 slides of sections from the first 50 patients were processed by the indirect immunofluorescence method. A first slide was processed with a cocktail of rabbit polyclonal antirat Nav1.7 (Alomone Labs, Jerusalem, Israel, ASC-008, 446–460 aa sequence; 1:100) and mouse monoclonal Human PGP (Protein Gene Product) 9.5 (CedarLane, Burlington, Canada, 31A3, [source UltraClone Ltd, Isle of Wight, United Kingdom]; 1:200). A second slide was processed with a cocktail of rabbit polyclonal antihuman Nav1.7 (Abcam, Waltham, MA, ab85167, human 1000–1100 aa sequence; 1:500) and the mouse monoclonal human PGP9.5. Secondary antibodies were donkey antirabbit Cy3 (Jackson Immuno Research, West Grove, PA, #711-165-152; 1:500) to visualize Nav1.7 labeling in red fluorescence and donkey anti mouse Alexa488 (Life Technologies, Carlsbad, CA, A11015; 1:250) to visualize PGP labeling in green fluorescence. 4′,6-diamidino-2-phenylindole (DAPI) was included in the secondary antibody incubation step to stain cell nuclei for blue fluorescence. The best Nav1.7 immunolabelling was with the Abcam antibody and anti-PGP combination, which was used for all biopsy analyses in the study.

To assure an even and consistent distribution of antibodies over all the sections from both biopsies on the same slide, an open rectangular frame of parafilm (thicker than the section thickness) was placed around the perimeter of all the sections, which were then flooded with an aliquot of primary antibody cocktail. This was then overlaid with a 60 × 24-mm coverslip to create a closed chamber having the space for the antibody to spread evenly over all the sections without drying out or running off the slides. Incubation under the primary antibody cocktail was horizontal for 2 days under refrigeration. The coverslips were floated off while vertical in Coplin jars to rinse the slides with copious PBS. The incubation process was repeated for the cocktail of secondary antibodies for 4 hours at room temperature. After a last PBS rinse, coverslips were mounted under 90% glycerin/10% PBS and stored at −20°C. Slides were warmed to room temperature only for the period of microscopic imaging and then returned to −20°C for long-term storage and any needed subsequent imaging. The unused 18 or 19 slides of alternating sections for each pair of biopsies were also covered with glycerin/PBS coverslips and archived at −20°C in the INTiDYN BioBank for future additional biomarker assessments as desired.

2.2.3. Epifluorescence imaging and mapping

Five sections evenly spaced through each biopsy were imaged for red/green/blue epifluorescence. Full-section montage images were collected using a 20X objective on an Olympus BX51-WI microscope equipped with conventional fluorescence filters (Cy3:528-553 nm excitation, 590–650 nm emission; Cy2/Alexa 488: 460-500 nm excitation, 510-560 nm emission), a Hamamatsu ER, DVC high-speed camera, linear focus encoder, and a 3-axis motorized stage system interfaced with Neurolucida software (MBF Bioscience, Essex, VT). Identical standardized camera sensitivity and exposure times were used for all sections across all biopsies.

The depth of field for the 20X objective enabled fully focused imaging through 10 µm of the 14 µm in 1 focal plane, with the remainder only slightly out of focus. NeuroLucida software provided for automated frame-by-frame capture, making automated focus adjustments and a subsequent seamless montage of the entire section (Figs. 1 and 2). NeuroLucida routines were used to zoom in and out to plot and measure the entire length of the epidermis in each section and plot the location of each immunolabeled profile using different markers to note their morphology, plane of sectioning completeness, and whether they were immunolabeled only for PGP9.5 or also double-labeled for Nav1.7.

Figure 1.

Skin biopsy taken from homologous area, contralateral to PHN side. Example quantification of PGP9.5-positive and Nav1.7-positive cutaneous innervation in a nonpainful (contralateral) skin biopsy section; PGP9.5 is visualized in green, Nav1.7 in red, and DAPI cell nuclei in blue. (A) Contralateral biopsy triple-color (PGP9.5, Nav 1.7, and DAPI) epifluorescence montage captured with a 20X objective. The area indicated by the white rectangle is enlarged for only PGP9.5 immunolabeling in B and for Nav1.7 and PGP9.5 in (D). (B) Enlarged area image of only PGP9.5 immunolabeling. The dotted line indicates the location of the epidermal basement membrane, and the broken line indicates the lower boundary of the compact upper (papillary) dermis. Narrow straight green arrows indicate examples of intraepidermal nerve fibers (IENF) and green arrowheads indicate examples where IENF can be seen at the point of entry into the epidermis. Broad green arrows indicate examples of subepidermal nerve fibers (SENF) located immediately proximal to the epidermal basement membrane, and curved green arrows indicate examples of upper dermal nerve fibers (UDNF). (C) The locations of all PGP9.5-positive IENF, IENF entry points, SENF, and UDNF were mapped to denote various morphological and plane of sectioning features. The solid white line indicates the mapping of the entire length of the epidermis, which was used to calculate the total densities of IENF and SENF. (D) Enlarged image of Nav1.7 and PGP9.5 immunolabeling. The symbols showing examples of labeled innervation are the same as in (B). The locations indicated by the small white rectangles labeled (a-e) are shown enlarged. The left image of each pair shows the double immunolabeling with PGP9.5 and Nav1.7, indicated by yellow symbols, and the right image depicts the Nav1.7 alone, indicated by red symbols. (a-d) are examples of a Nav1.7-labeled UDNF, IENF, entry point IENF, and SENF, respectively. Note that the image in (Dc) is designated both as an IENF (arrow) and an IENF entry point (arrowhead). (De) is an example of a Nav1.7-negative SENF. (E) NeuroLucida mapping of anti-Nav1.7–labeled IENF, IENF entry points, SENF, and UDNF as described in (C). Nav, voltage-gated sodium channel; PGP, protein gene product; PHN, postherpetic neuralgia.

Figure 2.

Skin biopsy taken from painful area on PHN side. Example quantification of PGP9.5-positive and Nav1.7-positive cutaneous innervation in the PHN-affected skin biopsy section from the same patient as in Figure 1; PGP9.5 is visualized in green, Nav1.7 in red, and cell nuclei in blue. (A–E) Images are as described for the same in Figure 1. Note that Dc shows a Nav1.7-negative (green arrow) and Nav1.7-positive (yellow and red arrow) UDNF fiber in the same image. All the densities of IENF, IENF entry points, SENF, and UDNF were much lower than that of the contralateral biopsy from the same patient. IENF, intraepidermal nerve fibers; Nav, voltage-gated sodium channel; PGP, protein gene product; PHN, postherpetic neuralgia; SENF, subepidermal nerve fibers; UDNF, upper dermal nerve fibers.

2.2.4. Assessment of intraepidermal and subepidermal nerve fiber densities

The procedure to quantify PGP9.5-labeled innervation density was designed for 14-µm thick sections that are amenable to double labeling.^30,31,34 Small sensory nerve fiber terminals in the epidermis (IENF), along with neuronal profiles immediately subjacent to the epidermis, SENF, were quantified. The SENF profiles represent sensory fibers that are the source of the IENF that penetrate the basement membrane to terminate in the epidermis (Fig. 1A). The IENF and SENF densities for each biopsy were calculated by summating the number of NeuroLucida-mapped IENF and SENF markers mapped in each of 5 sections and dividing by the NeuroLucida-mapped length of the epidermis. Whereas the original protocol assessed only 1 section per biopsy,³¹ in this study, 5 sections were averaged to express the IENF and SENF density for each biopsy as # fibers/mm length of epidermis.

2.2.5. Quantification of Nav1.7 immunolabeling on nerve fibers and epidermal keratinocytes

In the epidermis, Nav1.7-positive IENF labeling is frequently masked by keratinocyte Nav1.7-positive labeling. Analysis of PGP9.5 immunolabeling showed a very close correlation of SENF densities with IENF densities. Therefore, assessments of the density and proportions of Nav1.7-positive innervation were quantified based on SENF immunolabeling. To perform these assessments, the SENF profiles were mapped first for PGP9.5 by using the NeuroLucida routines labeling only the green fluorescent channel. Then, the sections were mapped for colabeling with Nav1.7 by gradually phasing in the red channel to decide whether, and when, a coexpressed red signal could be detected.

Following procedures previously published,^2,37 quantification of Nav1.7 among keratinocytes was assessed from the standardized Nav1.7 epifluorescence images of the full width and thickness of the epidermis captured from each of the 5 sections from each biopsy. Standard image analysis tools (Photoshop, Adobe Systems, San Jose, CA) to quantify the average pixel intensity (API) from specific-sized sampling masks and identical standardized camera sensitivity and exposure times were used for all sections. To best determine Nav1.7 keratinocyte API across the entire live, deep to superficial keratinocyte layers (stratum basalis, spinosum, and granulosum), 10 equally spaced locations over the entire length of the epidermis from each of the 5 sections per biopsy (50 measures total) were assessed. In addition, potential stratification of immunolabeling was assessed for approximate lower (basalis), middle (spinosum), and upper (granulosum) layers of epidermal keratinocytes.

2.3. Subgrouping of patients based on neuropathic pain symptom inventory items and histological parameters

A hierarchical cluster analysis was performed using Ward hierarchical clustering method⁴⁶ with Euclidian distance measure to identify subgroups using the combined information of patient-reported pain characteristics, as captured by the NPSI, related to small fiber function in PHN (various pain qualities, pain evoked by brush, pressure, or cold), as well as IENF and SENF innervation density. Because Nav1.7 labeling of keratinocytes compromised detection of IENF Nav1.7, the proportion of subepidermal Nav1.7-positive fibers to subepidermal PGP-positive fibers was used as a measure of relative Nav1.7 expression. Any hierarchical cluster analysis will form a tree-like structure of the data, where every participant forms a leaf, which are combined along branches. Two branches are combined in a knot, forming a new branch above the knot.

All items were transformed to a z scale to correct for outliers within the SENF and IENF density measures and Nav1.7 subepidermal ratio after log transformation. A secondary cluster analysis was performed using the UPGMA (Unweighted Pair Group Method with Arithmetic Mean)⁴⁰ method instead of Ward as a validation method; grouping in both resulting trees was compared using Cohen kappa for classification in the final groups. All statistical analyses were completed using SAS 9.4 and R version 4.1.0.

3. Results

3.1. Baseline parameters

The mean age of patients was 58.0 (SD, 16.0) years (females 57%), and the mean duration of time from herpes zoster to study entry was 2.3 (SD, 1.9) years (Table 1). The average baseline pain score for the full study cohort (n = 296) was 5.5 (SD, 1.1); 82% experienced punctate hyperalgesia, with a mean NRS intensity of 2.7 (SD, 2.4). Of the 296 randomized patients, 231 were genotyped as GG (homozygous for the common allele) and 65 were either AG (heterozygous) or AA (homozygous) for the R1150W polymorphism. No significant differences were seen in baseline pain scores (Table 1) or change in pain in response to treatment in these subgroups (Table 3). The NPSI mean score at baseline was 43.4 (SD, 16.1). Quality control review of allodynia data collected revealed that numerous investigators, despite specific training, did not perform the brush stroke procedure in the protocol-specified manner, and therefore, the allodynia results could not be used in the evaluation.

Table 1.

Baseline demographic and pain data.

Parameter	Genotype		Total(n = 296)
	GG(n = 231)	AG and AA(n = 65)
Mean age [y] (SD)	58.2 (15.3)	57.5 (18.4)	58.0 (16.0)
Female sex (%)	57	58	57
Ethnicity (%)
White	83.5	84.6	83.8
Black/African American	11.3	13.8	11.8
Other	5.2	1.5	4.4
Average pain intensity at baseline [NRS, range 0-10] (SD, n)	5.5 (1.1, 231)	5.5 (1.1, 65)	5.5 (1.1, 296)
Duration of PHN [y] (time from diagnosis of herpes zoster, SD, n)*	2.4 (2.1, 208)	1.9 (1.3, 56)	2.3 (1.9, 264)

NRS, numeric rating scale; PHN, postherpetic neuralgia; SD, standard deviation, genotype refers to single-nucleotide polymorphism rs6746030 of SCN9A (A allele is R1150W polymorphism, G is the common allele).

*The Pearson correlation coefficient between average pain intensity at baseline and duration of PHN is 0.044 with p-value of 0.485.

Table 3.

Correlations of pain, sensory symptoms, and age with skin biopsy data.

	IENF [/mm]				SENF Nav1.7-positive [%]
	PHN		Contralateral		PHN		Contralateral
	r	P	r	P	r	P	R	P
Baseline weekly average pain (NRS) [0-10]	−0.10	0.099	−0.09	0.123	0.08	0.208	0.06	0.304
Change from baseline in weekly average daily pain intensity at week 4	−0.05	0.436	0.08	0.212	−0.05	0.445	−0.02	0.726
Mechanical hyperalgesia (NRS) [0-10]	−0.12	0.078	0.08	0.22	0.24	0.000	0.24	0.000
NPSI (total score)	0.00	0.996	0.09	0.123	0.18	0.003	0.17	0.004
Age [y]	−0.39	0.000	−0.290	0.000	0.040	0.519	0.020	0.731

IENF, intraepidermal nerve fibers; Nav, voltage-gated sodium channel; NRS, numerical rating scale; PGP, protein gene product; PHN, postherpetic neuralgia; r, Pearson correlation coefficient; SENF, subepidermal nerve fibers.

3.2. PGP9.5-positive intraepidermal nerve fiber density in zoster affected and contralateral biopsies

Analysis of cutaneous innervation was performed using computer-assisted microscopy and image analysis, as demonstrated in Figure 1 Contralateral and Figure 2 PHN side. Routine image mapping procedures were used to determine the presence of IENF and SENF that were PGP9.5 positive and those that were also Nav1.7 positive. Skin biopsies from the site of PHN pain and the contralateral homologous region were available for 271 patients.

Across the total cohort, average PGP-positive IENF were observed at 10.3 fibers/mm epidermal length on the contralateral side and 8.3 fibers/mm epidermis on the PHN-affected side (Table 2). There was a reduction in nerve fiber density with increasing patient age on both the PHN and contralateral sides; this effect was significantly more prominent on the PHN side (Table 3). Although this calculated to an overall 20% average reduction in PHN side IENF density, in older individuals, the percentage of PHN side fiber loss was much higher, reaching 40% in those patients older than 70 years (Fig. 3).

Table 2.

Skin biopsy data.

Parameter	Contralateral side (SD)	PHN side (SD)
Nerve fiber density (epidermis, PGP9.5 positive) [IENF/mm]	10.32 (5.98)	8.25 (6.14)
Nerve fiber density (subepidermis, Nav1.7 positive) [SENF/mm]	1.19 (1.84)	0.98 (1.78)
Nerve fiber density (subepidermis, PGP9.5 positive) [SENF/mm]	3.47 (2.69)	2.85 (2.75)
Relative nerve fiber density (subepidermis, Nav1.7 positive) [%]	33.83	34.70
Upper subepidermal keratinocyte Nav1.7 (average pixel intensity)	102.20 (33.01)	102.45 (32.66)

IENF, intraepidermal nerve fibers; Nav, voltage-gated sodium channel; PGP, protein gene product; SD, standard deviation; SENF, subepidermal nerve fibers.

Figure 3.

Nerve fiber density measurements on PHN and contralateral side, according to age decile. IENF measurements by age group and skin biopsy location. IENF, intraepidermal nerve fibers; PGP, protein gene product; PHN, postherpetic neuralgia.

As seen in prior PHN biopsy evaluations, this study cohort showed variability in biopsy IENF findings from patient to patient and in the relative numbers on PHN side. Comparing individual IENF values on the PHN and contralateral sides demonstrated that although 50% of the patients had the anticipated ≥10% lower IENF density on the PHN side, 25% had ≥10% higher IENF density on the PHN side (Fig. 4).

Figure 4.

Subgrouping of patients based on ratio of nerve fiber density on PHN to contralateral side. Total study cohort subgroups based on innervation ratios (R) of PHN to contralateral IENF values. (A) Most of the study patients had a lower innervation density in the PHN biopsy, while others showed no difference or had a higher PHN biopsy density. (B) Scatter dot box plots of the total cohort innervation (R) subgroups and age, demonstrating that the 3 subgroups have relatively similar age distributions. IENF, intraepidermal nerve fibers; PHN, postherpetic neuralgia.

The overall 20% reduction in the average number of fibers on the PHN-affected side is likely partly due to the lower mean age of 58 years for the patients in this clinical trial, compared with approximately 70 years of age in the prior reports (Table 4). Intraepidermal nerve fiber counts showed statistically significant declines with age, but the decline with age was more prominent on the PHN side, suggesting that advanced age predisposes to PHN-induced loss of IENF. For patients older than 70 years, the decline on the PHN side was approximately 40%. Intraepidermal nerve fiber density for the PHN side in prior studies (adjusting for differences in section thickness in how IENF densities were calculated and expressed) ranged from 3.6 to 7.4 fibers/mm epidermal length (Table 4).

Table 4.

Prior reports in the literature of IENF assessments in PHN.

Author/year [Reference]	No. of patients	Mean age (y)	PHN pain score	PHN site*	Contralateral site*	Zoster without PHN*	Reduction at PHN site [%]
Rowbotham/199639	18	77	Mean worst daily pain [VAS, 0-100]: 40	6.9	23.4		71
Oaklander/199827	18	71	Mean pain [NRS, 0-10]: 6.9	4.7	14.9	23.2	68
Petersen/200029	17	76.9	Mean average daily pain [NRS, 0-10]: 6.9	3.6	16.0		78
Oaklander/200128 19 PHN 19 non-PHN	19	Not stated	Not stated	4.7		21.8	79
Petersen/201031 57 patients with Herpes zoster biopsied at 6 mo after rash	42 did not develop PHN	68	Average daily pain = 0 [VAS, 0-100]:		14.8†	10.8†	27
	15 developed PHN	69	Average daily pain ≥20	7.4†	16.5†		55

^*Units have been normalized to fibers/mm for comparison purposes.

^†Values taken from Reference 31, Figure 3 and normalized to fibers/mm.

IENF = intra-epidermal nerve fibers; NRS, numeric rating scale; PHN, postherpetic neuralgia; VAS, visual analogue scale.

3.3. Nav1.7-positive subepidermal nerve fiber density in zoster affected and contralateral side

Nav1.7 SENF immunolabeling also varied considerably between patients. Some patients had nondetectable to faint immunolabeling, while others showed very robust immunolabeling. However, for each patient's individual Nav1.7 immunolabeling, the level at which a red Nav1.7 signal could be detected by phasing in the red channel onto the green channel images of the PGP9.5 profiles was similar on both the PHN side and contralateral side biopsies for each patient. This indicates that there was no detectable difference between the intensity of Nav1.7 expression on the PHN and contralateral sides.

On average, Nav1.7-positive SENF were found at 0.98 fibers/mm epidermal length on the PHN side and 1.19 fibers/mm epidermal length on the contralateral side (Table 2). Using SENF, total PGP9.5-positive counts with Nav1.7-positive double-immunolabeled counts, the relative proportion of Nav1.7-positive SENF was not different between the PHN and contralateral sides (35% and 34%, respectively; Table 2); hence, neither an upregulation nor a downregulation of Nav1.7 was detected among the cutaneous innervation from PHN-affected skin compared with that from the contralateral skin.

3.4. Nav1.7 immunolabeling on keratinocytes

A cohort wide analysis of keratinocyte Nav1.7 immunofluorescent average pixel intensities demonstrated no significant differences from PHN-affected skin compared with those from contralateral skin (Table 2).

3.5. Association of nerve fiber counts and baseline parameters

There was no significant correlation between the average PGP9.5-positive IENF and duration of PHN, baseline pain score, NPSI total score, or punctate mechanical hyperalgesia (Table 3). There was also no significant correlation between the baseline pain score and duration of PHN (Table 1), nor between the baseline pain score and the proportion of Nav1.7-positive fibers (Table 3). No difference was noted between Nav 1.7-positive fiber counts on the PHN and contralateral sides. Nav 1.7-positive fiber counts correlated significantly with more punctate mechanical hyperalgesia, and a higher total NPSI score (Table 3).

3.6. Patient subgroups with low vs high proportion of Nav1.7-positive fibers

A cluster analysis indicated 2 groups of patients based on biggest height difference between knots. When cut into 2 groups, cluster assignment was identical in 87% of cases, and similarity between solutions indicated decent stability (Cohen kappa: 0.70, [95% CI: 0.61-0.79]).

Patients sorted to cluster 1 were on average younger than patients in cluster 2 (age 55.5 years, SD 15.7 vs 62.2 years, SD 16.0, P < 0.001) with no differences in sex distribution. The average baseline pain was NRS 5.6 (SD 1.1, range 3.9-9.3) in cluster 1 and 5.3 (SD 1.0, range 3.9-7.9) in cluster 2 (P = 0.021).

Cluster 1 is characterized by a relatively high level of immunolabeling for Nav1.7-positive fibers on the PHN side of 40%, whereas in cluster 2, only 25% were Nav1.7 positive. Cluster 1 showed a trend to higher values of the NPSI, which was particularly prominent for squeezing and cold-induced pain (Fig. 5). The differences in Nav1.7-positive expression were nearly identical on the contralateral side in both groups (39% vs 23%, respectively for cluster 1 vs 2).

Figure 5.

Cluster analysis. Cluster analysis revealed 2 distinct subsets of patients with significantly different Nav1.7 expression levels by using a pattern analysis. The high-level cluster group (40% Nav1.7-positive fibers) showed higher scores for neuropathic symptoms (using the Neuropathic Pain Symptom Inventory). Nav, voltage-gated sodium channel; NPSI, neuropathic pain symptom inventory.

4. Discussion

Bilateral skin biopsies from nearly 300 patients were collected, creating a large repository of human skin samples from patients with PHN. In this study, PGP9.5-positive innervation to the epidermis (ie, IENF) and the SENF were analyzed, and the proportion of the latter fibers that also demonstrated positive immunolabeling for Nav1.7 was determined. Spontaneous and evoked types of pain were assessed by quantitative sensory bedside testing for mechanical punctate hyperalgesia and the NPSI. This combined set of measures provided the opportunity to correlate patient-specific (ie, personalized medicine) anatomical data from cutaneous fiber counts with functional measures addressing the perception of pain and sensory symptoms. These are strengths of this study, but it is important to note the limitations, such as the lack of contemporaneous biopsy material from a distant site or from healthy controls.

4.1. Degeneration of fibers in the postherpetic neuralgia-affected skin and in contralateral skin

The overall average IENF density on the PHN side was similar, but at the high end of the range previously reported, at 8.3, again, probably due to the lower mean age of patients. Although comparisons are imperfect, particularly given how long ago some of the prior studies were conducted, the average fiber counts on the contralateral side in this study (10.3 fibers/mm epidermal length) were lower than those reported in the 3 previous studies that included contralateral comparisons (14.9-23.4 fibers/mm epidermal length). Thus, the relative reductions on the PHN side in those 3 studies were higher, at 68% to 78% (Table 4).^27,29 The fact that this recent cohort of patients probably received better antiviral treatment during the acute phase of zoster may also explain the lower relative fiber loss on the PHN side. Of note, approximately 25% of the patients in this study had lower fiber counts on the contralateral side compared with those on the PHN side. While no contemporaneous biopsies from distant skin or control subjects were included in this study, the average contralateral fibers counts in this study (10.25/mm) is lower than the average in thoracic skin of 9 control subjects from a prior study (18/mm), using identical analytic techniques.¹ This suggests that bilateral, if somewhat variable, fiber loss occurs in many patients with PHN and demonstrates that not all pain from PHN is caused by a loss of IENF, further supporting a role for other mechanisms in PHN.

The loss of contralateral primary neurons had been noted occasionally in prior studies comparing PHN with contralateral skin or to distant body sites, ³⁹ and was focused upon in another study, ²⁷ but in this study, the loss was much more prominent. A peripheral reduction of the C nociceptor axon reflex was described in homologous skin contralateral to PHN⁵ compared with adjacent dermatomes. A subclinical viral infection of the contralateral dorsal root ganglion during the acute zoster stage or localized meningoencephalitis affecting the nearby spinal cord may explain these contralateral anatomical changes.^38,47 In zoster sine herpete, a varicella reactivation occurs without dermatomal rash or other clinical phenomena.¹⁸ Thus, it is possible that herpes zoster is frequently a segmental bilateral infection, yet in most cases, only 1 side manifests the full clinical picture. Additional physiological mechanisms that are poorly understood may also contribute to the reduced contralateral innervation, as Oaklander found greater than 50% loss of epidermal innervation in contralateral homologous hind paws of rats after unilateral tibial nerve ligation and section.²⁶

The observed heterogeneity of fiber counts among this large cohort of patients with PHN could also reflect different sensory phenotypes of patients with PHN.³⁹ Similar to the differences in the anatomical integrity of skin nociceptors, quantitative sensory testing revealed patients with impaired and others with preserved small fiber function.^6,16 To date, it is unclear why some patients have different clinical phenotypes.

4.2. Association between nerve fiber density and baseline pain and sensory symptoms

This large cohort PHN study found no correlation of total PGP-positive IENF nor Nav1.7-positive nociceptors with duration of PHN or baseline pain score. In line with these results, sensory testing in PHN revealed no difference in nociceptor function between patients with and without pain.¹⁷ Previous trials in diabetic polyneuropathy, however, have demonstrated that patients experiencing pain showed a higher degree of IENF degeneration, functional sensory loss, as well as different keratinocyte biomarker patterns compared with nonpainful neuropathy patients.^2,33,45

These findings have to be interpreted in the light of different pathophysiological mechanisms causing neuropathic pain. In a subgroup of patients, nociceptive fibers in the skin are relatively preserved and show signs of peripheral sensitization (ie, irritable nociceptors), which is associated with pain.^6,39 In another subgroup of patients, severe degeneration of nociceptive fibers is likely associated with a pain generator located more centrally (ie, deafferentation pain).^6,16

Importantly though, the proportion of Nav1.7-positive fibers correlated with the total NPSI score and the intensity of punctate mechanical hyperalgesia. That NPSI scores robustly correlated with mechanical hyperalgesia and Nav1.7, but baseline pain scores did not, may be explained by the fact that NPSI captures a variety of neuropathic symptoms including evoked types of pain, whereas daily pain scores are mainly influenced by spontaneous ongoing pain symptoms, such as ongoing burning.

A similar correlation could also be found between these measures and the Nav1.7 density on the contralateral side, indicating that this finding is not associated with a herpes zoster specific mechanism.

In accordance with this finding, we could identify 2 distinct subsets of patients with significantly differing Nav1.7 expression levels by using a pattern analysis. The high-level cluster group (40% Nav1.7-positive fibers) showed higher scores for neuropathic NPSI symptoms, particularly for squeezing and cold-provoked pain (Fig. 5). However, because similar expression patterns were also observed on the contralateral side, these expression differences are not associated with the disease but likely represent biological variation in patients. People who possess more Nav1.7 on their nociceptors tend to experience more intense sensory symptoms and signs, if they are affected by PHN, than the group with low Nav1.7 expression.

4.3. The role of Nav1.7 in neuropathic pain due to postherpetic neuralgia

Several findings suggest that Nav1.7 poses an interesting target for analgesic therapies: Gain-of-function variants in the gene SCN9A encoding for the Nav1.7 sodium channel alpha subunit have been discovered in painful erythromelalgia and painful small fiber neuropathy.^15,41,48 In 111 patients with painful diabetic neuropathy, 10 rare Nav1.7 variants were identified, whereas none were present in painless diabetic neuropathy,¹⁰ implying a potential role in the pain phenotype. Furthermore, Nav1.7 is believed to be the most important sodium channel contributing to spontaneous and hyperactive discharges emanating from peripheral nociceptive neurons.¹² Animal and in vitro human tissue experiments demonstrate that Nav1.7 is expressed in peripheral axons with a presumably nociceptive function and seems to be upregulated in these small-sized neurons in neuropathic pain states.^20,21,23

However, the findings in this study using a large study cohort did not support Nav1.7 upregulation: (1) no association could be found between Nav1.7-positive nociceptors and the intensity of spontaneous pain; (2) although immunodetection was interindividually heterogeneous, only one-third of all nociceptors stained positively for Nav1.7; (3) no upregulation of Nav1.7 in PHN-affected skin was observed compared with that in the contralateral side (35% vs 34%, respectively). Thus, in the whole cohort analysis, it seems that Nav1.7 expression in cutaneous innervation does not seem to be critically involved in the pathophysiology of PHN pain.

4.4. Nav1.7 antagonists for the treatment of neuropathic pain

TV-45070 is a novel Nav1.7 antagonist being developed for the treatment of patients with various neuropathic pain indications. Unlike the nonselective Nav blocker lidocaine, TV-45070 was designed to semiselectively block activity in the Nav1.7 channel. There are several possible reasons why the present clinical trial, using a topical application of this compound, did not meet the primary endpoint: (1) the target, Nav1.7, was present only at low levels in the skin biopsies, and (2) in previous efficacious Nav1.7 rat studies, the target was not only prominently detected in presumptive nociceptor small peripheral DRG neurons, but also centrally, in the spinal dorsal horn.⁹ Because topically applied TV-45070 generates negligible plasma levels, it would not be expected to reach those more proximal or central sites.

Even systemic administration of selective Nav1.7 inhibitors, however, has not demonstrated robust analgesia activity in neuropathic pain.^8,24 It has been theorized that near-complete inhibition of Nav1.7 may be required or that congenital loss of Nav1.7 differs from pharmacologic inhibition because it is also associated with an increased endogenous opioid expression.^22,25

5. Conclusions

Within this large cohort of patients with PHN, on average, a 20% reduction in PGP-positive nerve fibers could be demonstrated on the PHN side compared with the contralateral side. This large cohort of patient biopsies confirms that bilateral nerve damage with PHN is common. Cutaneous innervation that specifically labeled for the Nav1.7 target was not seen in most of the PGP-positive fibers, and the Nav1.7 innervation biomarker was not able to differentiate between PHN-affected and contralateral skin biopsies. Two distinct patient clusters with high and low Nav1.7 immunolabeling demonstrated different clinical manifestations.

Conflict of interest statement

M. Fetell, T. Li, and L. Marinelli are current employees of Teva Pharmaceuticals and may hold stock in that company. R. Baron consults for Teva and has received honorariums for lectures.

Appendix A. Supplemental digital content

Supplemental digital content associated with this article can be found online at http://links.lww.com/PAIN/B851.

Supplementary Material

jop-164-2435-s001.docx ^{(53.5KB, docx)}