Multiparametric quantitative magnetic resonance neurography of the lumbosacral plexus in chronic demyelinating polyneuropathies

Full Text

Introduction

Chronic polyneuropathies (PNPs) comprise a heterogeneous group of peripheral nervous system disorders with marked clinical and etiological variability. Their prevalence in the general population is estimated at approximately 1–2%, reaching 7% in older age groups. Moreover, recent data suggest that PNP prevalence is increasing over time, underscoring its significant contribution to the burden of neurological morbidity [1, 2].

Among chronic demyelinating PNPs, dysimmune and hereditary forms hold a prominent place. Chronic inflammatory demyelinating polyneuropathy (CIDP) and hereditary motor and sensory neuropathy type 1 (HMSN1, Charcot–Marie–Tooth disease type 1) are among the most common demyelinating peripheral nerve disorders. Despite substantial differences in pathogenesis — CIDP being an autoimmune condition mediated by T-cell dysfunction (particularly Th1 and CD8⁺ cells) and impaired regulatory T-cell function [3], while HMSN1 results from various genetic defects, such as duplication of the 17p11.2-p12 region with PMP22 overexpression in HMSN1A [4] or mutations in the MPZ gene encoding the major peripheral myelin protein in HMSN1B [4, 5] — the clinical presentation, electrophysiological features, and some neuroimaging findings may overlap, significantly complicating differential diagnosis and potentially leading to misdiagnosis.

Accurate diagnosis of chronic demyelinating PNPs remains a clinically significant challenge. An initial misdiagnosis was made in 68.3% of CIDP cases prior to referral to a specialized center. Among misdiagnoses, Guillain–Barré syndrome (23.3%) and polyneuropathy of unclear etiology (45%), including hereditary neuropathies (29.6%), diabetic (18.5%), and chronic idiopathic axonal PNP (14.8%), were the most common. The mean delay to correct diagnosis was 21.3 months (range 2–132 months), resulting in a mean immunotherapy delay of 22.4 months [6]. Conversely, other data, including studies from the Russian population, indicate the opposite issue, namely, overdiagnosis of CIDP, with misdiagnosis rates reaching approximately 67% [7]. This underscores the need for additional diagnostic tools and objective quantitative biomarkers.

Electrophysiological methods, particularly nerve conduction study (NCS), have traditionally been the mainstay of PNP diagnosis [8]. This technique confirms neural involvement, differentiates demyelinating from axonal damage, identifies conduction blocks, and serves as the primary tool for confirming CIDP according to the European Academy of Neurology/Peripheral Nerve Society (EAN/PNS 2021) criteria [9]. However, electrophysiological assessment has several limitations that increase the risk of diagnostic error: substantial operator dependence; difficulty in assessing proximal nerve structures (including the lumbosacral plexus) and in severe soft tissue edema; and the lack of direct quantitative markers for assessing the degree of myelin damage [7, 8, 10].

Ultrasound (US) of peripheral nerves is an important complement to NCS, allowing visualization of nerve thickening and changes in the intraneural echostructure in chronic demyelinating PNPs, which may be useful for diagnosis and monitoring treatment response [11–14]. However, US is limited by considerable measurement variability, operator dependence, difficulty in assessing deep neural structures, lack of quantitative markers for demyelination or axonal damage severity, and insufficient standardization of criteria for differential diagnosis of various demyelinating PNP forms [13, 14].

Modern magnetic resonance neurography (MRN) opens new possibilities for in vivo visualization of peripheral nerve pathology, revealing morphological changes inaccessible to other non-invasive methods [15, 16]. However, qualitative methods are often insufficient for accurate characterization of demyelination, axonal damage, and inflammatory changes. Moreover, morphological parameters alone (such as nerve cross-sectional area and MR signal intensity changes) do not allow differentiation of various mechanisms of nerve tissue injury [10]. Therefore, quantitative MR techniques are of particular interest for improving diagnostic accuracy and establishing objective biomarkers. These include the following methods.

Diffusion tensor imaging (DTI) is based on measuring the magnitude and direction of water diffusion in tissues. Key DTI parameters include fractional anisotropy (FA), which reflects the directionality of diffusion within a voxel; mean diffusivity (MD), representing the average diffusion magnitude; and, most importantly for microstructural tissue characterization, axial diffusivity (AD) and radial diffusivity (RD), which characterize diffusion along and perpendicular to the principal diffusion direction, respectively [17, 18]. In neuropathies, FA is generally reduced [20], reflecting disruption of organized nerve fiber architecture. Increased RD is thought to indicate myelin sheath damage and may serve as a marker of demyelination, while axonal loss is associated with decreased AD [19–21].

T2 relaxometry is a multi-echo imaging sequence that enables quantitative assessment of the T2 signal and calculation of tissue relaxation parameters, namely the apparent T2 relaxation time (T2app) and proton density (PD) [22]. Increased T2app is primarily associated with an increase in free water protons — more precisely, with endoneural edema, which corresponds to increased signal on standard T2-weighted sequences. PD appears to correlate with microstructural changes in the extracellular matrix caused by plasma protein extravasation through the physiological endovascular barrier and a pro-inflammatory environment, more characteristic of dysimmune demyelinating lesions [16].

Magnetization transfer imaging (MTI) is based on the presence of two main proton pools in biological tissues: free and macromolecule-bound. The relaxation time of bound protons is much shorter than that of free protons. Using a preparatory saturation radiofrequency pulse, the bound proton pool can be selectively suppressed. Due to energy transfer via dipole-dipole interaction to the free proton pool, partial suppression of their longitudinal magnetization occurs, detected as signal reduction — a phenomenon termed magnetization transfer [23]. In biological tissues, this effect is quantified by the magnetization transfer ratio (MTR), which reflects the integrity of the macromolecular pool. For example, damage to the macromolecular pool in peripheral nerves (disruption of myelin, collagen, and axonal proteins) results in reduced MTR, which serves as a marker of myelin loss or structural alteration and represents a promising tool for detecting and quantifying demyelination, particularly in chronic nerve injury [24].

To date, comparative studies simultaneously evaluating multiple quantitative MRI parameters in patients with CIDP, HMSN1, and healthy volunteers remain scarce [10, 25]. However, such comparative analyses with subsequent selection of relevant MRI markers are of high clinical relevance and represent a promising direction in MRI diagnostics for peripheral nervous system disorders. Such approaches could improve early diagnosis and differential diagnosis of chronic demyelinating PNPs, stratify patients by the degree of abnormality, enable earlier initiation of appropriate therapy, and predict treatment response.

The aim of this study was to evaluate the diagnostic potential of multiparametric quantitative MR neurography of the lumbosacral plexus by comparing quantitative MRI parameters — diffusion tensor imaging (FA, MD, RD, AD), magnetization transfer imaging (MTR), and T2 relaxometry (T2app, PD) — between patients with HMSN1, CIDP, and healthy volunteers, and to identify parameters with the greatest potential for differential diagnosis of these conditions.

Materials and methods

A prospective, cross-sectional, single-center study was conducted at the Research Center of Neurology and Neurosciences (RCNN) in 2024–2025. The study protocol was approved by the RCNN Local Ethics Committee (Protocol No. 10-8/23 dated December 20, 2023), and all participants provided written informed consent.

Patients were enrolled sequentially from the RCNN database using ICD-10 codes (G60.0, G62.8 for HMSN1; G61.8, G62.8 for CIDP) from January 1, 2020, to October 15, 2025, followed by review of outpatient records for eligibility. Eligible patients were invited to participate by phone or email; upon agreement, a study date was scheduled. The control group consisted of healthy volunteers with no history of neurological disease, recruited through convenience sampling from center staff and their social contacts. Participant enrollment for all groups took place from February 2024 to November 2025. The patient and healthy volunteer selection process is shown in the flowchart (Figure 1).

 

Fig. 1. Study flow diagram.

 

Following all screening stages, the following patient groups were included in the study:

• 30 patients with genetically confirmed HMSN1, including 28 with HMSN1A and 2 with HMSN1B;
• 31 patients with CIDP fully meeting the EAN/PNS 2021 electrodiagnostic criteria [9], with disease duration of no more than 5 years and no spontaneous remission;
• 27 healthy volunteers over 18 years of age with no specific complaints or clinical signs of peripheral nerve involvement at inclusion or in their medical history. NCS verification of subclinical polyneuropathy was not performed due to the low prevalence in the population of individuals without risk factors or neurological complaints.

The exclusion criteria were decompensated medical conditions, comorbid conditions that could independently affect peripheral nerve structures (diabetes mellitus of any type, alcohol abuse, use of neurotoxic drugs, etc.), and contraindications to MRI.

The withdrawal criteria included patient/healthy volunteer refusal to continue participation at any stage of the study, including inability to stay in the scanner for the required duration; MRI data with significant artifacts (e.g., due to motion); decompensation of medical conditions; and identification of other etiologies of polyneuropathy during the study.

Imaging was performed on a Magnetom Prisma 3T MRI scanner (Siemens Healthineers) using an 18-channel body coil. Patients were examined in the supine position with neutral lower limb positioning; the lumbosacral plexus was centered at the L5–S1 intervertebral space based on sagittal images. MRI sequence parameters used in the scanning protocol are provided in Table 1.

 

Table 1. MRI sequence parameters used in the study

Parameter	3D T2 SPACE	3D T1 VIBE Dixon	3D STIR SPACE	T2 dual-echo	T1 GRE MT-ON/OFF	DW-EPI (DTI)**
Plane	Sagittal	Coronal	Coronal	Axial	Axial	Axial
TR, ms	1500	5.52	3000	5900	33	4900
TE, ms	139	2.46	250	9/63	7	52
Voxel size, mm	0.94 × 0.94 × 0.94	0.94 × 0.94 × 0.94	0.4 × 0.4 × 0.8	0.34 × 0.34 × 3.0	0.34 × 0.34 × 3.0	1.25 × 1.25 × 2.5
FOV, mm	324	380	350	220	220	305
Number of slices	72	144	144	27–45	27–45	55
Scan time	6 min 14 s	5 min 58 s	8 min 27 s	11 min 20 s	1–4 min*	7 min 47 s

Note: Scan duration depended on the number of slices and the presence of the saturation pulse.** — b = 0/1000 s/mm²; 20 directions; AP/PA.

 

Quantitative measurements were performed at the level of the L5, S1, and S2 anterior spinal nerve roots, 5–10 mm distal to the corresponding spinal ganglia, using anatomical landmarks (Figure 2). For each nerve trunk, regions of interest (ROIs) were manually delineated on two adjacent slices, and values were averaged to reduce random measurement variability.

 

Fig. 2. Visualization of lumbosacral plexus elements using 3D STIR SPACE in the coronal plane in a patient with HMSN1A (own data).

Yellow markers indicate ROI placement at the L5 anterior spinal nerve roots; orange — at S1; red — at S2.

 

T2 relaxometry and magnetization transfer imaging data, after conversion from DICOM to NIfTI format, were analyzed using ITK-SNAP¹ software for manual ROI contouring on the corresponding anatomical slices of each sequence separately. ROI localization across different sequences was matched using anatomical landmarks without formal image coregistration. Quantitative parameters (T2app, PD, MTR) were calculated using standard formulas [26].

DTI data were processed using DSI Studio software [27] with standard preprocessing, including distortion and motion correction, followed by calculation of FA, MD, AD, and RD at manually defined ROIs. All measurements were performed by a single investigator, eliminating inter-observer variability but precluding assessment of inter-rater reproducibility.

The statistical analysis was performed using the IBM SPSS Statistics, version 27.0 (IBM).

The distribution of continuous variables was assessed using frequency histograms. Additionally, the normality of the distribution in the pooled sample was assessed using the Kolmogorov–Smirnov test to guide data description. To verify ANOVA assumptions, normality within each group was assessed using the Shapiro–Wilk test. The vast majority of MRI parameters, age, and body mass index (BMI) were normally distributed. The three study groups were compared by sex using Pearson's chi-square test, and by age and BMI using analysis of variance (ANOVA). Comparison of quantitative MRI parameters across the three groups (CIDP, HMSN1, and healthy volunteers) was performed using one-way analysis of variance (ANOVA).

Levene's test was used to assess homogeneity of variances. When statistically significant intergroup differences were identified, post-hoc pairwise comparisons were performed with Bonferroni correction for multiple comparisons, or with Tamhane's T3 test in cases of unequal variances. The null hypothesis was rejected at p < 0.05 (or at p_adj < 0.05 after correction for multiple comparisons).

Results

Sex distribution differed significantly between groups overall (p = 0.033); however, post-hoc pairwise comparisons with Bonferroni correction showed no significant differences (p_adj = 0.054–1.000). Age and BMI did not differ significantly between groups (p = 0.084 and p = 0.183, respectively). Therefore, subsequent analyses were performed without adjustment for baseline characteristics, which are summarized in Table 2. Data are presented as mean ± standard deviation (M ± SD) values.

 

Table 2. Baseline characteristics of the study groups

Parameter	Normal (n = 27)	HMSN1 (n = 30)	CIDP (n = 31)
Sex, n (%):
female	18 (66.7%)	23 (76.7%)	14 (45.2%)
male	9 (33.3%)	7 (23.3%)	17 (54.8%)
Age, years (M ± SD)	43 ± 13	43 ± 11	49 ± 13
BMI, kg/m2 (M ± SD)	26 ± 4	25 ± 5	27 ± 4

 

Statistical analysis of MRI data showed that the distribution of quantitative parameters was normal, allowing the use of parametric methods. Detailed results are provided in Tables 3 and 4.

 

Table 3. Comparison of quantitative MRI parameters between study groups, M ± SD (95% CI)

Parameter			HMSN1 (n = 30)	CIDP (n = 31)	Normal (n = 27)	χ2	padj*
L5	FA	R	0.32 ± 0.06 (0.30–0.35)	0.34 ± 0.06 (0.32–0.36)	0.43 ± 0.04 (0.41–0.44)	0.390	< 0.0001
		L	0.34 ± 0.07 (0.31–0.36)	0.35 ± 0.06 (0.33–0.37)	0.43 ± 0.05 (0.40–0.45)	0.285	< 0.0001
	MD, 10–3 mm2/s	R	1.40 ± 0.17 (1.33–1.46)	1.33 ± 0.21 (1.25–1.41)	1.21 ± 0.12 (1.17–1.26)	0.153	0.002
		L	1.37 ± 0.18 (1.30–1.43)	1.32 ± 0.20 (1.25–1.40)	1.22 ± 0.11 (1.17–1.26)	0.122	0.007
	AD, 10–3 mm2/s	R	1.90 ± 0.16 (1.84–1.96)	1.83 ± 0.22 (1.75–1.91)	1.83 ± 0.17 (1.77–1.90)	0.029	0.305
		L	1.89 ± 0.19 (1.82–1.96)	1.85 ± 0.23 (1.77–1.94)	1.84 ± 0.15 (1.78–1.90)	0.014	0.588
	RD, 10–3 mm2/s	R	1.14 ± 0.19 (1.07–1.21)	1.08 ± 0.22 (0.99–1.16)	0.91 ± 0.10 (0.86–0.95)	0.231	< 0.0001
		L	1.11 ± 0.19 (1.03–1.18)	1.06 ± 0.20 (0.98–1.13)	0.91 ± 0.11 (0.86–0.95)	0.188	0.0003
S1	FA	R	0.30 ± 0.07 (0.27–0.32)	0.33 ± 0.06 (0.31–0.35)	0.41 ± 0.06 (0.39–0.44)	0.388	< 0.0001
		L	0.30 ± 0.07 (0.28–0.33)	0.34 ± 0.07 (0.32–0.37)	0.42 ± 0.05 (0.40–0.44)	0.370	< 0.0001
	MD, 10–3 mm2/s	R	1.52 ± 0.17 (1.46–1.58)	1.47 ± 0.18 (1.40–1.53)	1.32 ± 0.13 (1.26–1.37)	0.218	0.0001
		L	1.53 ± 0.16 (1.47–1.60)	1.49 ± 0.23 (1.40–1.57)	1.32 ± 0.14 (1.26–1.37)	0.201	0.0002
	AD, 10–3 mm2/s	R	2.02 ± 0.13 (1.97–2.07)	2.03 ± 0.18 (1.96–2.09)	1.96 ± 0.16 (1.90–2.02)	0.036	0.235
		L	2.06 ± 0.15 (2.01–2.12)	2.06 ± 0.21 (1.99–2.14)	1.98 ± 0.19 (1.91–2.06)	0.040	0.205
	RD, 10–3 mm2/s	R	1.27 ± 0.20 (1.19–1.34)	1.19 ± 0.19 (1.12–1.26)	0.99 ± 0.14 (0.94–1.05)	0.287	< 0.0001
		L	1.27 ± 0.19 (1.20–1.34)	1.20 ± 0.25 (1.10–1.29)	0.99 ± 0.14 (0.93–1.04)	0.260	< 0.0001
S2	FA	R	0.29 ± 0.05 (0.27–0.30)	0.33 ± 0.06 (0.31–0.36)	0.39 ± 0.04 (0.37–0.40)	0.400	< 0.0001
		L	0.29 ± 0.06 (0.27–0.32)	0.33 ± 0.07 (0.30–0.35)	0.40 ± 0.05 (0.38–0.42)	0.363	< 0.0001
	MD, 10–3 mm2/s	R	1.51 ± 0.17 (1.45–1.57)	1.45 ± 0.22 (1.37–1.53)	1.38 ± 0.12 (1.33–1.43)	0.080	0.040
		L	1.50 ± 0.16 (1.44–1.56)	1.48 ± 0.24 (1.39–1.56)	1.39 ± 0.15 (1.33–1.45)	0.056	0.112
	AD, 10–3 mm2/s	R	2.00 ± 0.20 (1.93–2.07)	1.99 ± 0.24 (1.90–2.08)	2.00 ± 0.18 (1.93–2.07)	0.001	0.966
		L	2.01 ± 0.20 (1.94–2.09)	2.03 ± 0.25 (1.93–2.12)	2.06 ± 0.19 (1.99–2.14)	0.009	0.690
	RD, 10–3 mm2/s	R	1.27 ± 0.17 (1.20–1.33)	1.18 ± 0.22 (1.10–1.26)	1.07 ± 0.11 (1.03–1.11)	0.171	0.001
		L	1.25 ± 0.17 (1.18–1.31)	1.21 ± 0.25 (1.12–1.30)	1.05 ± 0.14 (1.00–1.11)	0.152	0.002
L5	T2app, ms	R	103.9 ± 22.4 (95.6–112.3)	121.6 ± 35.5 (108.6–134.7)	96.4 ± 14.5 (90.7–102.2)	0.146	0.002
		L	103.9 ± 23.8 (95.0–112.7)	116.4 ± 35.8 (103.3–129.6)	97.9 ± 15.4 (91.8–104.0)	0.079	0.040
	PD	R	640.0 ± 81.8 (609.4–670.5)	553.4 ± 77.8 (524.8–581.9)	510.2 ± 80.3 (478.4–541.9)	0.317	< 0.0001
		L	636.3 ± 79.2 (606.7–665.9)	552.7 ± 78.4 (523.9–581.4)	534.6 ± 79.8 (503.1–566.2)	0.245	< 0.0001
S1	T2app, ms	R	122.3 ± 29.5 (111.3–133.3)	130.9 ± 38.3 (116.8–145.0)	107.6 ± 24.8 (97.8–117.4)	0.085	0.033
		L	128.8 ± 36.7 (115.1–142.5)	127.8 ± 37.5 (114.0–141.5)	111.6 ± 28.4 (100.4–122.8)	0.049	0.145
	PD	R	656.1 ± 87.5 (623.4–688.7)	570.4 ± 76.7 (542.3–598.5)	522.9 ± 77.6 (492.2–553.5)	0.321	< 0.0001
		L	691.3 ± 67.9 (666.0–716.6)	581.1 ± 81.1 (551.4–610.9)	560.8 ± 86.2 (526.7–594.9)	0.356	< 0.0001
S2	T2app, ms	R	125.2 ± 39.5 (110.5–140.0)	127.9 ± 37.8 (114.1–141.8)	109.8 ± 21.3 (101.4–118.2)	0.051	0.131
		L	133.8 ± 41.4 (118.4–149.3)	140.0 ± 55.4 (119.7–160.3)	119.5 ± 23.5 (110.2–128.8)	0.039	0.211
	PD	R	637.8 ± 84.1 (606.4–669.2)	518.2 ± 82.9 (487.8–548.6)	499.5 ± 69.6 (471.9–527.0)	0.381	< 0.0001
		L	651.5 ± 78.0 (622.4–680.7)	527.4 ± 77.4 (499.0–555.8)	501.8 ± 75.5 (471.9–531.6)	0.427	< 0.0001
L5	MTR	R	27.4 ± 4.2 (25.9–29.0)	22.2 ± 4.8 (20.5–24.0)	24.6 ± 3.9 (23.1–26.1)	0.208	0.0001
		L	27.0 ± 4.8 (25.2–28.8)	24.0 ± 4.2 (22.4–25.5)	24.2 ± 4.6 (22.4–26.0)	0.087	0.031
S1	MTR	R	27.2 ± 4.2 (25.6–28.7)	23.9 ± 4.8 (22.2–25.7)	23.6 ± 4.8 (21.8–25.5)	0.111	0.012
		L	27.1 ± 4.3 (25.6–28.7)	23.5 ± 4.6 (21.8–25.2)	24.9 ± 5.0 (22.9–26.8)	0.102	0.017
S2	MTR	R	26.4 ± 4.6 (24.6–28.1)	23.5 ± 4.7 (21.7–25.2)	24.5 ± 4.4 (22.7–26.2)	0.069	0.064
		L	26.5 ± 6.2 (24.2–28.8)	22.6 ± 4.8 (20.8–24.3)	24.0 ± 4.7 (22.1–25.9)	0.091	0.026

Note: *p_adj — significance level for overall differences between diagnostic groups.

 

Table 4. Significance levels for post-hoc pairwise comparisons, p_adj(|d|)

Parameter			HMSN1 vs CIDP	HMSN1 vs Control	CIDP vs Control
L5	FA	R	0.711 (0.276)	< 0.001 (1.976)	< 0.001 (1.605)
		L	0.996 (0.239)	< 0.001 (1.435)	< 0.001 (1.292)
	MD, 10–3 mm2/s	R	0.395 (0.350)	< 0.001 (1.215)	0.049 (0.639)
		L	0.730 (0.238)	< 0.001 (1.013)	0.043 (0.642)
	AD, 10–3 mm2/s	R	Post-hoc comparisons not performed (no overall group differences)
		L	Post-hoc comparisons not performed (no overall group differences)
	RD, 10–3 mm2/s	R	0.493 (0.328)	< 0.001 (1.531)	0.001 (0.956)
		L	0.830 (0.250)	< 0.001 (1.254)	0.005 (0.900)
S1	FA	R	0.080 (0.558)	< 0.001 (1.854)	< 0.001 (1.427)
		L	0.052 (0.584)	< 0.001 (1.930)	< 0.001 (1.268)
	MD, 10–3 mm2/s	R	0.654 (0.298)	< 0.001 (1.336)	0.002 (0.942)
		L	0.883 (0.249)	< 0.001 (1.418)	0.003 (0.860)
	AD, 10–3 mm2/s	R	Post-hoc comparisons not performed (no overall group differences)
		L	Post-hoc comparisons not performed (no overall group differences)
	RD, 10–3 mm2/s	R	0.319 (0.384)	< 0.001 (1.575)	< 0.001 (1.160)
		L	0.471 (0.329)	< 0.001 (1.687)	< 0.001 (1.016)
S2	FA	R	< 0.001 (0.882)	< 0.001 (2.230)	< 0.001 (1.040)
		L	0.064 (0.572)	< 0.001 (1.917)	< 0.001 (1.226)
	MD, 10–3 mm2/s	R	0.534 (0.311)	0.005 (0.856)	0.388 (0.368)
		L	Post-hoc comparisons not performed (no overall group differences)
	AD, 10–3 mm2/s	R	Post-hoc comparisons not performed (no overall group differences)
		L	Post-hoc comparisons not performed (no overall group differences)
	RD, 10–3 mm2/s	R	0.247 (0.439)	< 0.001 (1.354)	0.060 (0.605)
		L	1.000 (0.195)	0.001 (1.206)	0.012 (0.736)
L5	T2app, ms	R	0.068 (0.595)	0.351 (0.393)	0.002 (0.906)
		L	0.296 (0.412)	0.591 (0.295)	0.036 (0.656)
	PD	R	< 0.001 (1.086)	< 0.001 (1.601)	0.129 (0.547)
		L	< 0.001 (1.062)	< 0.001 (1.279)	1.000 (0.228)
S1	T2app, ms	R	0.872 (0.252)	0.252 (0.536)	0.019 (0.712)
		L	Post-hoc comparisons not performed (no overall group differences)
	PD	R	< 0.001 (1.043)	< 0.001 (1.607)	0.084 (0.617)
		L	< 0.001 (1.471)	< 0.001 (1.694)	0.982 (0.244)
S2	T2app, ms	R	Post-hoc comparisons not performed (no overall group differences)
		L	Post-hoc comparisons not performed (no overall group differences)
	PD	R	< 0.001 (1.433)	< 0.001 (1.783)	1.000 (0.243)
		L	< 0.001 (1.597)	< 0.001 (1.949)	0.631 (0.335)
L5	MTR	R	< 0.001 (1.163)	0.046 (0.701)	0.115 (0.545)
		L	0.035 (0.666)	0.070 (0.590)	1.000 (0.049)
S1	MTR	R	0.023 (0.712)	0.015 (0.784)	1.000 (0.059)
		L	0.009 (0.820)	0.195 (0.493)	0.814 (0.281)
S2	MTR	R	Post-hoc comparisons not performed (no overall group differences)
		L	0.015 (0.710)	0.229 (0.453)	0.950 (0.296)

 

According to ANOVA, significant differences between groups were found for most L5–S2 levels in FA, MD, RD, PD, and MTR, whereas AD did not differ significantly (Table 3).

Compared to controls, patients with HMSN1 showed lower FA and higher MD, RD, and PD at almost all levels, indicating pronounced diffuse structural changes. In the CIDP group, decreased FA and increased MD and RD were also observed compared to controls, but PD changes were less pronounced and more variable across levels.

The HMSN1 and CIDP groups differed primarily in PD and MTR: in hereditary demyelinating PNP, these parameters were higher (Table 4).

Discussion

This study is the first in the Russian population to evaluate a comprehensive set of quantitative MR neurography parameters measured at the lumbosacral plexus level in healthy volunteers and patients with chronic demyelinating PNPs.

In this cross-sectional study, we demonstrated that multiparametric quantitative MR neurography of the lumbosacral plexus can identify distinct patterns of peripheral nerve involvement in HMSN1 and CIDP compared to healthy controls. Both pathological groups showed reduced FA and increased MD and RD relative to healthy volunteers, consistent with a demyelinating pattern of injury; however, the profiles of PD and MTR differed between HMSN1 and CIDP.

The observed reductions in FA and increases in MD/RD align with previous DTI studies of peripheral nerves, in which RD and FA are considered sensitive markers of myelin sheath integrity, while MD and AD serve as integrated measures of tissue structure and axonal integrity [28]. In CIDP, reduced FA and increased nerve size have been shown to correlate with clinical severity and electrophysiological signs of demyelination [10]. A systematic review and meta-analysis demonstrated that FA and nerve cross-sectional area are the most reliable quantitative markers for CIDP diagnosis, although threshold values vary between studies and by location, while qualitative assessment of MR signal intensity changes remains far less specific [29]. However, data on PD and MTR in this context remain limited, with only isolated reports of reduced MTR in CIDP patients correlating with demyelination [10]. Our findings confirm the key role of FA, MD, and RD as markers of demyelination in CIDP, and provide new insights into the significance of PD and MTR in differentiating hereditary from acquired chronic demyelinating PNPs.

The quantitative MRI changes observed in HMSN1 warrant separate discussion. Previous MRN studies in HMSN1A have demonstrated diffuse FA reduction, increased diffusion coefficients, and enlarged nerve diameters [25]. Nerve diameter and DTI parameters are currently considered promising MRI markers for assessing disease severity and progression in HMSN1A, supporting the relevance of quantitative MRN as a monitoring tool in hereditary neuropathies [30]. However, the role of quantitative parameters such as PD and MTR in HMSN1A, measured at the lumbosacral plexus level, has not been described to date. In this study, we found for the first time that PD and MTR values at the lumbosacral plexus were higher in HMSN1 patients compared to both CIDP and controls, while no significant differences in T2app were observed. Similar changes were reported by M. Vaeggemose et al., who found increased PD measured at the sciatic nerve level in HMSN1A patients compared to controls, with no differences in T2app [25]. Given previous quantitative MRI evidence that MTR reflects the macromolecular pool, including myelin, and that PD is described as a sensitive microstructural biomarker of nerve tissue integrity in the peripheral nervous system, the combined increase in PD and MTR in HMSN1A may be secondary to structural remodeling of nerve tissue in the context of hypermyelination [31].

The interpretation of quantitative MRI changes in the CIDP group was complicated by clinical heterogeneity. The study included patients with diverse disease profiles — from acute/subacute onset to remission or persistent minimal neurological deficit; some patients were examined before initiation of specific therapy, while others received long-term glucocorticoid or intravenous immunoglobulin therapy, and even monoclonal antibody therapy. Longitudinal studies by F. Preisner et al. showed that quantitative MRN parameters in CIDP change with immunotherapy and partially normalize, particularly in patients with good clinical response [10, 32]. Given these data, heterogeneity in clinical course and treatment within the CIDP group may have attenuated the PD and MTR differences, partially explaining the smaller gap between CIDP and controls compared to HMSN1 and controls.

This study has several limitations that should be considered when interpreting the data. First, the limited sample size reduces statistical power, particularly for analysis of individual levels and sides. Second, the cross-sectional design precludes assessment of longitudinal changes in quantitative MRI parameters over time, either in response to therapy or during the natural course of HMSN1 and CIDP. For example, in a longitudinal evaluation of HMSN1A, Y. Zhu et al. were able to detect changes in DTI parameters and proposed stratification of patients by disease severity and progression rate 30]. Third, PD and MTR depend on scanning parameters and post-processing methods; therefore, quantitative threshold values cannot be directly transferred to other centers without protocol standardization. In brain imaging, multicenter multiparametric mapping studies (R1, PD, MT, R2*) have shown high reproducibility with rigorous protocol standardization and normalization of PD in relative units, but emphasize the critical sensitivity of these parameters to scanner settings [33]. Similar protocol standardization is needed for peripheral nerve assessment.

Despite these limitations, our findings support the concept of multiparametric quantitative MRN of the lumbosacral plexus as a non-invasive tool for characterizing peripheral nerve injury mechanisms. FA, MD, and RD appear to be the most reliable markers of chronic demyelinating PNPs overall, while the combined use of PD and MTR shows potential for differentiating hereditary (HMSN1) from immune-mediated (CIDP) demyelinating PNPs and may form the basis for quantitative diagnostic criteria.

¹ URL: https://www.itksnap.org/.

About the authors

Dmitrii S. Tsaregorodtsev

Russian Center of Neurology and Neurosciences

Author for correspondence.
Email: tsaregorodtsev.dmitri@yandex.ru
ORCID iD: 0009-0009-4851-9649

Postgraduate Student, Radiologist, Radiology Department

Russian Federation, Moscow

Sofya N. Morozova

Russian Center of Neurology and Neurosciences

Email: tsaregorodtsev.dmitri@yandex.ru
ORCID iD: 0000-0002-9093-344X

Cand. Sci. (Med.), Researcher, Radiology Department

Russian Federation, Moscow

Darya A. Grishinа

Russian Center of Neurology and Neurosciences

Email: tsaregorodtsev.dmitri@yandex.ru
ORCID iD: 0000-0002-7924-3405

Dr. Sci. (Med.), Head, Center for Peripheral Nervous System Diseases, Institute of Clinical and Preventive Neurology

Russian Federation, Moscow

Aysylu F. Murtazina

N.P. Bochkov Research Centre of Medical Genetics

Email: tsaregorodtsev.dmitri@yandex.ru
ORCID iD: 0000-0001-7023-7378

Cand. Sci. (Med.), Researcher

Russian Federation, Moscow

Dmitrii M. Subbotin

N.P. Bochkov Research Centre of Medical Genetics

Email: tsaregorodtsev.dmitri@yandex.ru
ORCID iD: 0009-0007-9862-0273

Clinical Geneticist, Researcher

Russian Federation, Moscow

Olga A. Shchagina

N.P. Bochkov Research Centre of Medical Genetics

Email: tsaregorodtsev.dmitri@yandex.ru
ORCID iD: 0000-0003-4905-1303

Dr. Sci. (Med.), Leading Researcher, DNA Diagnostics Laboratory, Head, Laboratory of Molecular Genetic Diagnostics

Russian Federation, Moscow

Marina V. Krotenkova

Russian Center of Neurology and Neurosciences

Email: tsaregorodtsev.dmitri@yandex.ru
ORCID iD: 0000-0003-3820-4554

Dr. Sci. (Med.), Head, Radiology Department

Russian Federation, Moscow

Natalia A. Suponeva

Russian Center of Neurology and Neurosciences

Email: tsaregorodtsev.dmitri@yandex.ru
ORCID iD: 0000-0003-3956-6362

Dr. Sci. (Med.), Corr. Member of RAS, Director, Institute of Neurorehabilitation and Recovery Technologies

Russian Federation, Moscow

References

Supplementary files