A Rat Model Mimicking Treatment-Induced Painful Diabetic Neuropathy Following Experimental Rapid Weight Loss

Abstract

Purpose

This study investigated the feasibility of constructing a painful diabetic neuropathy (PDN) rat model mimicking treatment-induced neuropathy (TIND) through rapid weight reduction via caloric restriction combined with metformin.

Methods

Type 2 diabetes was induced using a high-fat diet combined with low-dose streptozotocin (STZ, 30 mg/kg). At week 11, rats were classified into spontaneous PDN and non-painful diabetic neuropathy (NPDN) groups based on mechanical withdrawal threshold (MWT) decline (≥50% from baseline). Rapid weight loss was subsequently induced over 10 weeks (weeks 12–21) via 40% food restriction by weight combined with metformin gavage (200 mg/kg·d). Mechanical withdrawal threshold (MWT), motor and sensory nerve conduction velocities (MNCV and SNCV), sciatic nerve histopathology, and intraepidermal nerve fiber density (IENFD) were assessed.

Results

The spontaneous PDN rate at week 11 was 54.8% (17/31). Following the intervention, MWT decreased by 91.47% in the NPDN-WL group (P = 0.001) and by 91.08% in the PDN-WL group (P < 0.01). MNCV was significantly lower in both diabetic weight-loss groups than in their respective diabetic control groups. SNCV was numerically lower in the weight-loss groups, but the corresponding pairwise differences were not statistically significant after Bonferroni adjustment. Histological examination showed myelin disorganization, vacuolar degeneration, and axonal swelling, whereas IENFD showed a non-significant decreasing trend. Notably, all NPDN-WL rats (7/7) achieved complete phenotypic conversion (100%), whereas no conversion occurred in the NPDN-Con group (0/3, 0%). However, the control groups were small and unbalanced (NPDN-Con, n = 3; PDN-Con, n = 4), and these findings should be interpreted with caution.

Conclusion

These preliminary findings suggest that rapid weight reduction via caloric restriction combined with metformin may offer a feasible approach for establishing a TIND-like PDN rat model, and that rapid weight loss under diabetic conditions could be an important contributing factor in the conversion toward a painful neuropathy phenotype. Further studies with larger sample sizes are warranted to substantiate these observations.

Introduction

Painful diabetic peripheral neuropathy (PDN) is one of the most challenging chronic complications of diabetes mellitus, with a global incidence ranging from 17.9% to 65.3%.¹ PDN typically manifests as symmetrical distal neuropathic pain and can substantially impair patients’ quality of life.^2,3 However, due to the complex pathophysiological mechanisms, current clinical interventions combining glycemic control and symptomatic pain management often yield unsatisfactory outcomes.⁴ Accordingly, establishing animal models with high clinical relevance and stability remains essential for elucidating the pathogenic mechanisms of PDN and identifying therapeutic targets.

Currently, most PDN animal models are established using high-fat diet combined with low-dose streptozotocin (STZ) in rodents. This approach involves inducing a diabetes model first, followed by screening rats that exhibit pain-related behavioral manifestations such as hyperalgesia. However, such models have limitations including prolonged screening time for pain phenotypes, substantial variability in success rates, and difficulty in simulating the metabolic dysregulation state observed during the disease progression in PDN patients.⁵

Notably, clinical observations have revealed that some diabetic patients develop new-onset or exacerbated neuropathy during intensive glycemic control and weight management, despite improved blood glucose levels.⁶ This suggests that rapid weight loss may act as an important trigger.

Building upon traditional diabetic rat models, the present study incorporated caloric restriction combined with metformin administration to simulate clinical short-term, high-intensity metabolic adjustment. Using a multidimensional verification strategy encompassing behavioral, electrophysiological, and histopathological assessments, we investigated whether rapid weight reduction is associated with PDN phenotype conversion and exacerbates pain sensitivity under diabetic conditions.

This study has dual significance: First, from a methodological perspective, it provides a novel strategy for establishing PDN animal models that better reflect clinical situations such as treatment-induced neuropathy (TIND). Second, from a translational perspective, the model may facilitate investigation of neuropathic changes occurring during intensive metabolic and weight-loss interventions and may provide a basis for future studies of safer weight-management strategies in diabetes.

Materials and Methods

Experimental Animals and Ethics

Forty-seven specific-pathogen-free (SPF) healthy male Sprague-Dawley (SD) rats, aged 8 weeks, with body weight (200 ± 20) g, were provided by the Animal Experimental Center of Guangzhou University of Chinese Medicine, Daxuecheng Campus [Animal Production License Number: SCXK (GD) 2023–0068]. The experimental animals were housed in an SPF-level standard barrier system at the Animal Experimental Center of the Sanyuanli Campus of Guangzhou University of Chinese Medicine [Animal Use License Number: SYXK (GD) 2025–0347]. Environmental parameters were controlled as follows: temperature (22 ± 2)°C, relative humidity (60 ± 10)%, 12-hour light-dark cycle (light period 06:00∼18:00), with free access to food and water. This study protocol was reviewed and approved by the Ethics Committee for Experimental Animals of Guangzhou University of Chinese Medicine (Ethics Approval Number: 20240402009). All animal experimental operations were conducted in facilities complying with the Regulations for the Administration of Affairs Concerning Experimental Animals (State Council of the People’s Republic of China).

For surgical exposure of the sciatic nerve to measure nerve conduction velocity, rats were anesthetized with isoflurane (Beijing Obo Pharmaceutical Technology Co., Ltd., China) delivered via an anesthesia induction chamber at an induction concentration of 3% and maintained at 1.5% throughout the procedure. At the conclusion of the experiment, all rats were euthanized by exposure to an overdose of isoflurane followed by cervical dislocation to ensure death. This euthanasia method is consistent with the recommendations of the American Veterinary Medical Association (AVMA) Guidelines for the Euthanasia of Animals.

Main Reagents and Equipment

Main Reagents

Diets

High-fat diet: Formulated by weight percentage as follows: basic diet 67%, lard 10%, sucrose 20%, cholesterol 2.5%, sodium cholate 0.5%. Provided by Guangdong Provincial Medical Experimental Animal Center (Batch number: 202405073).

Standard diet: Formulated according to the Chinese National Standard GB/T 14924.10–2001 Formulation and Preparation of Experimental Animal Feed. The main components of the diet include corn, wheat bran, fish meal, soybean meal, and trace amounts of added vitamin and mineral mixtures.

Sterilized by Cobalt-60 irradiation and used for the maintenance and breeding of standard SPF-level experimental rats. Provided by the Animal Center of Guangzhou University of Chinese Medicine.

Chemical Reagents and Kits

Streptozotocin (Beijing Solarbio Science and Technology Co., Ltd., China BS185); 4% Paraformaldehyde (Lanjing Technology Co., Ltd., China BL539); Isoflurane (Beijing Obo Pharmaceutical Technology Co., Ltd., China 153717015); Metformin hydrochloride tablets (Merck Pharmaceuticals (Jiangsu) Co., Ltd., China ACS3016); Citrate-citric acid buffer solution (Fuzhou Feijing Biotechnology Co., Ltd., China PH1716); Anhydrous ethanol (China National Pharmaceutical Group Chemical Reagent Co., Ltd., China 100092683); Xylene (China National Pharmaceutical Group Chemical Reagent Co., Ltd., China 10023418); HE staining kit (Wuhan Pausswald Biotechnology Co., Ltd., China BH0001); Hematoxylin differentiation solution (Wuhan Pausswald Biotechnology Co., Ltd., China B0020); Hematoxylin bluing solution (Wuhan Pausswald Biotechnology Co., Ltd., China B0021); Hematoxylin staining solution (Wuhan Pausswald Biotechnology Co., Ltd., China B0002); DAB colorimetric reagent kit (Wuhan Pausswald Biotechnology Co., Ltd., China B0053); Goat serum blocking solution (Wuhan Bos De Biotechnology Co., Ltd., China AR1009); HRP-conjugated goat anti-rabbit secondary antibody (Wuhan Pausswald Biotechnology Co., Ltd., China B0047); HRP-conjugated goat anti-mouse secondary antibody (Wuhan Pausswald Biotechnology Co., Ltd., China B0048); HRP-conjugated rabbit anti-goat secondary antibody (Wuhan Pausswald Biotechnology Co., Ltd., China B0064); HRP-conjugated goat anti-rat secondary antibody (Wuhan Pausswald Biotechnology Co., Ltd., China B0056); Citrate antigen retrieval solution (Wuhan Pausswald Biotechnology Co., Ltd., China B0034); EDTA antigen retrieval solution (Wuhan Pausswald Biotechnology Co., Ltd., China B0035); Citrate-EDTA antigen retrieval solution (Wuhan Pausswald Biotechnology Co., Ltd., China B0063); Hydrogen peroxide (China National Pharmaceutical Group Chemical Reagent Co., Ltd., China 10011218); PBS (Wuhan Pausswald Biotechnology Co., Ltd., China B0040); BSA (Beijing Solarbio Science and Technology Co., Ltd., China A8010).

Main Equipment

Blood glucose meter (Bayer Healthcare Co., Ltd., Germany); Von Frey filaments (Sigma Corporation, USA NC12775); Electronic analytical balance (Sartorius Corporation, Germany BP121S); Electromyography data acquisition system (Edith Instruments International Trade (Shanghai) Co., Ltd., China PL15T02); Anesthesia machine (Shenzhen Ruiwode Life Science and Technology Co., Ltd., China R580S); Nerve conduction test instrument (ADInstruments (Sydney) Corporation., Australia PL15T02); Dehydrator (Wuhan Junjie Electronics Co., Ltd., China JT-12J); Pathological microtome (Shanghai Leica Instruments Co., Ltd., China LEICA RM2016); Light microscope (Nikon Corporation, Japan NIKON ECLIPSE CI); Vortex mixer (Scilogex Corporation, USA MX-S); Embedding machine (Wuhan Junjie Electronics Co., Ltd., China JB-L5); Tissue spreader (Wuhan Junjie Electronics Co., Ltd., China JK-6); Oven (Shanghai Forma Experimental Instruments Co., Ltd., China DGX-9003B); Cryostat (ThermoFisher Scientific, USA HM525); Destaining shaker (Jiangsu Xinkang Medical Equipment Co., Ltd., China XK-8); Vortex mixer (Scilogex Corporation, USA MX-F).

Experimental Protocol and Grouping Criteria

The entire experimental process was divided into three stages. As illustrated in Figure 1, the timeline and sample flow for each stage are detailed.

Figure 1.

Experimental protocol and animal flow chart. The flowchart shows the three experimental stages (Stage 1, diabetes background modeling, weeks 0–4; Stage 2, spontaneous PDN phenotype observation and identification, weeks 5–11; Stage 3, rapid weight-loss intervention, weeks 12–21) and the formation of the six subgroups. The “withdrawal (n = 1)” box indicates the withdrawal that occurred during Stage 2; a complete record of all attrition events is provided in Table S1.

Stage 1: Diabetes Background Modeling (0–4 Weeks)

After one week of adaptive housing, 47 male SD rats were randomly divided into control group (n = 12) and diabetes modeling group (n = 35) using a random number table method. The modeling group was administered high-fat diet combined with low-dose streptozotocin (STZ) via intraperitoneal injection (30 mg/kg) to simulate diabetic pathological background, while the control group received regular diet. This stage aimed to establish a stable hyperglycemic metabolic background.

Stage 2: Spontaneous PDN Phenotype Observation and Identification (Weeks 5–11)

The original diet regimen was maintained while observing the persistent effects of hyperglycemia on peripheral nerves. At the end of week 11, mechanical withdrawal threshold (MWT) was measured using Von Frey filaments to identify PDN phenotypes. Based on the prespecified behavioral criterion, rats with a decrease in MWT of ≥50% from baseline were classified as having the spontaneous PDN phenotype, whereas rats with a decrease of <50% were classified as having the non-painful diabetic neuropathy (NPDN) phenotype. This procedure was used to identify rats that spontaneously developed a painful neuropathy phenotype.^7,8

Stage 3: Rapid Weight Loss Intervention (Weeks 12–21)

To simulate the clinical occurrence of treatment-induced neuropathy (TIND), rats in each phenotype group were further randomly divided into control group (Con) and rapid weight loss group (WL). The WL group received caloric restriction combined with metformin intragastric administration for a continuous 10-week intervention. Finally, six subgroups were formed:

1. Normal control group (NC): Normal rats receiving standard diet with free access to food and daily intragastric administration of distilled water;

2. Healthy weight loss group (NWL): Normal rats receiving caloric restriction combined with metformin intervention;

3. Non-painful diabetic neuropathy control group (NPDN-Con): Rats with baseline NPDN status receiving standard diet with free access to food and daily intragastric administration of distilled water;

4. Non-painful diabetic neuropathy weight loss group (NPDN-WL): Rats with baseline NPDN status receiving caloric restriction combined with metformin intervention;

5. Painful diabetic neuropathy control group (PDN-Con): Rats with baseline PDN status receiving standard diet with free access to food and daily intragastric administration of distilled water;

6. Painful diabetic neuropathy weight loss group (PDN-WL): Rats with baseline PDN status receiving caloric restriction combined with metformin intervention.

Modeling and Intervention Methods

Establishment of Diabetes Model

At the end of week 4, rats in the modeling group were fasted for 12 hours, then received a single intraperitoneal injection of 2% STZ solution (dissolved in 0.1 mol/L citrate buffer, pH 4.5) at a dose of 30 mg/kg; the control group received an equal volume of buffer solution. Seventy-two hours after injection, blood glucose was measured from tail vein blood samples, with blood glucose level ≥ 16.7 mmol/L defined as successful diabetes modeling.⁸

Observation and Identification of Spontaneous PDN Phenotype

Following successful diabetes modeling, the original diet regimen was maintained. At the end of week 11, mechanical withdrawal threshold (MWT) in rats in each group was measured using the Von Frey filament method. Painful and non-painful neuropathy phenotypes were classified according to the percentage decrease in MWT from baseline.

Rapid Weight Reduction Intervention Protocol

The rapid weight reduction intervention consisted of caloric restriction combined with metformin intragastric administration. This protocol was designed to simulate the clinical characteristics of severe metabolic fluctuations observed in PDN patients undergoing intensive glycemic control and weight management. The intervention commenced at week 12 and continued for 10 weeks.

Weight Reduction Intervention Group

Dietary restriction: Rats were switched from a high-fat diet to a standard diet with caloric restriction implemented. Daily food intake was restricted to 60% of the mean daily food intake (measured in grams) recorded during the week preceding intervention — that is, food quantity was reduced by 40% by weight (ie, 40% caloric restriction).

Pharmacological intervention: Metformin (200 mg/kg·d) was administered via intragastric gavage. The dose was calculated based on body surface area (BSA) conversion between humans and rats, simulating the routine clinical therapeutic dose of metformin. The dose conversion was performed using the following formula:⁹

where K_m denotes the body surface area correction factor (rat: 6; human: 37). Medication was administered daily at 10:00 AM, with gavage volume controlled at no more than 1 mL per 100 g body weight.

Control Group

Rats received a standard diet with ad libitum access to food and were administered an equal volume of distilled water via intragastric gavage at the same time points daily. The detailed modeling protocol is shown in Figure 1.

Observation Indices and Detection Methods

All rats were subjected to general condition observation, blood glucose monitoring, MWT detection, nerve conduction velocity detection, and other indicators.

General Condition Observation

Daily observation of rat mental status, activity level, hair luster, diet and water intake, and other general conditions were recorded, with abnormal manifestations documented. An electronic analytical balance was used to measure body weight daily at 9:00 AM in the fasted state, and body weight changes were recorded.

Blood Glucose Monitoring

Random blood glucose was measured from the tail vein at a fixed time each week (10:00–11:00 AM). The procedure was as follows:

1. The root of the rat’s tail was gently massaged to promote blood circulation;

2. The tail tip was disinfected with 75% ethanol;

3. A disposable blood collection needle was rapidly inserted 1–2 cm from the tail tip;

4. The first drop of blood was discarded, and the second drop was collected on a glucose test strip via capillary action;

5. The reading was taken after 5 seconds.

Mechanical Withdrawal Threshold (MWT) Detection

The Von Frey filament “Up–Down method” was used to detect mechanical withdrawal threshold (MWT). Detection time was fixed at the same time each week (14:00–16:00).

Procedure

Rats were placed in a transparent acrylic test box with a metal mesh floor. After 15 minutes of adaptation, testing commenced. All behavioral assessments were performed by two researchers who were blinded to group assignments. Response judgments were made independently by both researchers; in cases of disagreement, a third researcher was consulted to reach a final consensus decision.

1. Von Frey filaments were applied vertically from below the mesh to stimulate the plantar surface of both hind paws.

2. The filaments were bent into a “C” or “S” shape and maintained for 6–8 seconds.

3. Observations were made for paw withdrawal, licking, or shaking responses. A positive response (withdrawal, licking, or shaking) was recorded as “X”; no response was recorded as “O”.

4. The next test filament was selected based on the response: a positive response led to selection of a smaller force value filament, while a negative response led to selection of a larger force value filament.

5. The interval between two stimulations was ≥10 seconds.

Calculation formula: The 50% mechanical withdrawal threshold (MWT) was calculated using the Up-Down method.¹⁰

X_f = the filament number of the last tested filament; δ = logarithmic mean difference of each filament (≈ 0.224); k = coefficient value obtained from the test sequence table.

Nerve Conduction Velocity Detection

After 10 weeks of weight reduction intervention, nerve conduction function was detected using the PowerLab neurophysiological system. Rats were anesthetized with isoflurane delivered via an anesthesia induction chamber at an induction concentration of 3%, then maintained at 1.5% throughout the procedure. Following anesthesia, the right sciatic nerve was exposed after hair removal, with room temperature maintained at approximately 25°C.

Motor Nerve Conduction Velocity (MNCV): Stimulating electrodes were placed at the tail root end of the sciatic nerve, recording electrodes were placed at the gastrocnemius muscle belly on the same side, and reference electrodes were placed between the recording and stimulating electrodes near the recording point. Square wave stimulation was applied (pulse width 0.05 ms, intensity 10–15 V) to record compound muscle action potentials.

Sensory Nerve Conduction Velocity (SNCV): Stimulating electrodes were placed on the tibial nerve at the medial ankle, and recording electrodes were placed at the tail root end of the sciatic nerve. Stimulation parameters were identical to those used in MNCV detection.

Three electrical stimulations were applied to each rat, and the average conduction time was used to calculate nerve conduction velocity (NCV).

Upon completion of nerve conduction measurements, rats were euthanized by exposure to an overdose of isoflurane via induction chamber, followed by cervical dislocation to confirm death.

Histopathological Examination

Hematoxylin and Eosin (HE) Staining of Sciatic Nerve: Following completion of electrophysiological testing, the left sciatic nerve was isolated and a segment approximately 1 cm in length was immediately placed in 4% paraformaldehyde fixative for 24 hours. After dehydration using a dehydrator and paraffin embedding, consecutive transverse cross-sections with a thickness of 5 µm were prepared. Following HE staining, pathological changes including nerve fiber morphology, myelin sheath integrity, and axonal structure were observed under light microscopy, and photographs were recorded.

Intraepidermal Nerve Fiber Density (IENFD) Detection: Rats were euthanized and skin tissue from the plantar surface of the hind foot (3 mm diameter) was collected. Tissues were fixed overnight in 4% paraformaldehyde at 4°C. Following paraffin embedding, consecutive sections with a thickness of 4 µm were prepared.

Immunohistochemical Staining Procedure:

1. Sections were dewaxed to water and washed with PBS three times (5 min each).

2. Endogenous peroxidase was blocked with 3% H₂O₂ at room temperature for 25 minutes, and tissues were blocked with 3% BSA for 30 minutes.

3. Rabbit anti-PGP9.5 polyclonal antibody (1:200 dilution, Abcam) was added and incubated overnight at 4°C.

4. After PBS washing, biotinylated secondary antibody was added and incubated at 37°C for 20 minutes, followed by incubation with streptavidin-HRP working solution at 37°C for 20 minutes.

5. Sections were developed with DAB to brown-yellow, counterstained with hematoxylin for nuclear visualization, and processed through routine dehydration and mounting.

IENFD Quantification Method: By light microscopy (400×), PGP9.5-positive nerve fibers that crossed the basement membrane into the epidermis were counted. Counting criteria: Complete crossing of the basement membrane was counted as 1 fiber; fibers that bifurcated after crossing were still counted as 1 fiber; incomplete crossing was not counted. IENFD = number of nerve fibers/epidermal length (fibers/mm). Five randomly selected non-overlapping fields were counted on each slide, and the average value was calculated using double-blind counting.

Statistical Methods

Statistical analysis was performed using SPSS version 26.0. All quantitative data were first subjected to Shapiro–Wilk normality testing. Data conforming to normal distribution were presented as mean ± standard deviation (), while non-normally distributed data were presented as median (interquartile range) [M(P25, P75)]. Comparisons between two groups were analyzed using independent samples t-test or Mann–Whitney U-test. For cross-sectional comparisons among multiple groups (eg, nerve conduction parameters and IENFD at the endpoint), one-way analysis of variance (ANOVA) or Kruskal–Wallis H-test was used. For longitudinal data measured repeatedly over time (eg, body weight, blood glucose, and MWT), a two-way repeated-measures ANOVA was employed, with Group as the between-subjects factor and Time as the within-subjects factor. When the sphericity assumption was violated (Mauchly’s test, P < 0.05), the Greenhouse–Geisser correction was applied. For data with missing values due to animal attrition, a linear mixed-effects model was used, which is robust to missing data under the missing-at-random assumption. Post-hoc multiple comparisons were performed using Bonferroni or Dunn methods with appropriate adjustments for the specific analysis.

Paired data were compared using paired t-test or Wilcoxon signed-rank test. Correlation analysis was performed using Pearson correlation or Spearman rank correlation analysis. Count data were analyzed using χ²-test or Fisher’s exact test. For post-hoc pairwise comparisons following the six-group one-way ANOVA, Cohen’s d was calculated as the pairwise mean difference divided by the pooled within-group residual standard deviation derived from the ANOVA model.

Results

General Conditions and Metabolic Index Changes in Experimental Animals

General Condition of Experimental Animals

During the diabetes background modeling stage, rats in the modeling group exhibited typical diabetic symptoms, characterized by a significantly increased food intake, water intake, and urine volume compared to the control group, accompanied by slow body weight gain and dull, lackluster coat appearance. Following the implementation of the weight reduction intervention, rats in the weight loss groups exhibited the aforementioned diabetic symptoms along with weight loss, and additionally developed neurological dysfunction-related symptoms including lethargy, reduced activity, and abnormal gait. Among these, the PDN-WL group exhibited the most pronounced symptoms.

Blood Glucose Level Changes

Diabetes background modeling stage: Prior to STZ injection, there was no significant difference in baseline blood glucose between the modeling group and the control group (P > 0.05). Seventy-two hours after STZ injection, blood glucose levels in all rats in the modeling group were ≥ 16.7 mmol/L and significantly elevated compared to the control group [(18.13 ± 1.21) mmol/L vs (5.34 ± 0.55) mmol/L, P < 0.001], indicating successful establishment of the diabetes model. Rapid weight loss intervention stage: With prolonged duration of the intervention, blood glucose in all weight reduction groups showed varying degrees of declining trends (Figure 2A). At the end of intervention (week 10), random blood glucose in NPDN-WL and PDN-WL groups was significantly reduced compared to their baseline before intervention (NPDN-WL: P < 0.05, PDN-WL: P < 0.001). Moreover, compared to corresponding NPDN-Con and PDN-Con groups, blood glucose levels in weight reduction groups were also significantly lower (all P < 0.05, Figure 2B).

Figure 2.

Effects of Rapid Weight Loss Intervention on Body Mass and Random Blood Glucose in Rats of Different Groups. (A) Random blood glucose change trends of rats in each group during the weight loss intervention; (B) Comparison of random blood glucose before and after intervention; (C) Body mass change trends of rats in each group during the weight loss intervention; (D) Comparison of body mass before and after intervention; In the box plots, the boxes represent the interquartile range (25–75%), the horizontal line inside the box indicates the median, the whiskers extend to the maximum and minimum values, and individual points represent outliers. Compared with before intervention (week 0) within the group, *P < 0.05, **P < 0.01, ***P < 0.001.

Body Weight Changes

Diabetes background modeling and spontaneous PDN phenotype observation and identification stage: From week 0 to the end of week 4, rats in the modeling group showed significantly faster body weight gain compared to the control group due to high-fat diet consumption (P < 0.05). Following STZ injection, the rate of body weight gain in the modeling group notably slowed. During the PDN identification period from weeks 5–11, body weight in the control group continued steady growth, while growth in the diabetic model group lagged behind. By the end of week 11 (prior to initiation of weight reduction intervention), baseline body weight among all diabetic subgroups was balanced with no statistically significant differences (P > 0.05).

Rapid weight loss intervention stage: Weight reduction effects progressively appeared with advancement of the intervention (Figure 2C). Early intervention phase: Beginning from week 2 of the weight reduction intervention, body weight in PDN-WL and NPDN-WL groups showed significant decrease compared to their respective control groups (P < 0.05). Mid-intervention phase: By week 6 of intervention, body weight in the NWL group was also significantly reduced compared to the NC group (P < 0.05). Endpoint of intervention: By the end of intervention (week 10, ie, week 21 of the experiment), body weight in all weight reduction groups (NWL, NPDN-WL, PDN-WL) was significantly lower than in their corresponding control groups (all P < 0.001) and showed substantial decline from baseline levels (Figure 2D).

Changes in Mechanical Withdrawal Threshold (MWT)

Changes in Mechanical Withdrawal Threshold During Spontaneous PDN Phenotype Identification (Weeks 5–11)

Following successful diabetes model establishment (beginning of week 5 of the experiment), the baseline MWT in diabetic rats was comparable to that in healthy control rats, showing no statistically significant difference [diabetic: (11.07 ± 3.38) g vs healthy control: (11.52 ± 3.13) g, P = 0.689; Table 1]. During the subsequent observation period, MWT in the diabetic group exhibited a progressive declining trend. By the end of week 7, MWT in diabetic rats had decreased to (8.30 ± 3.33) g, which was significantly lower than that in healthy controls [(11.60 ± 3.88) g, P = 0.008]. This declining trend continued through week 9 [diabetic: (6.81 ± 4.01) g vs healthy control: (10.03 ± 3.37) g, P = 0.018; Table 1].

Table 1.

Temporal Changes in Mechanical Withdrawal Threshold During the Spontaneous PDN Phenotype Observation Period (week 5–9) in All Surviving Animals

Group	n	Week 5 (Initial)	Week 7 (End)	Week 9 (End)
Diabetic rats	32	11.07±3.38	8.30±3.33	6.81±4.01
Healthy control rats	12	11.52±3.13	11.60±3.88	10.03±3.37
t-value		−0.40	−2.79	−2.47
P		0.689	0.008**	0.018*

At week 11, two rats died prior to the MWT assessment; therefore, the analysis at this time point was performed on the remaining surviving animals. Among the 31 surviving diabetic rats, the MWT was (7.54 ± 3.24) g, which remained significantly lower than that of the 11 surviving healthy controls [(12.80 ± 3.31) g, P < 0.001; Table 2]. These results indicate that under prolonged hyperglycemic conditions, diabetic rats exhibited marked spontaneous mechanical hyperalgesia progression.

Table 2.

Temporal Changes in Mechanical Withdrawal Threshold During the Spontaneous PDN Phenotype Observation Period (week 11) in All Surviving Animals

Group	n	Week 11 (End)
Diabetic rats	31	7.54±3.24
Healthy control rats	11	12.80±3.31
t-value		−4.60
P		< 0.001***

Note: *P < 0.05, **P < 0.01, ***P < 0.001.

Mechanical Withdrawal Threshold Changes During Rapid Weight Loss Intervention (Weeks 12–21)

With the implementation of rapid weight reduction intervention, MWT in all weight reduction groups of rats exhibited significant declining trends (Figure 3A). MWT in all six groups decreased after intervention (Table 3).

Figure 3.

Trends in MWT of rats in each group during rapid weight loss intervention. (A) MWT trends of rats in each group during the weight loss intervention; (B) Comparison of MWT before and after intervention. Significant reductions in MWT were observed in the NPDN-WL and PDN-WL groups (P < 0.01), and in the NWL group (P < 0.05). *P < 0.05, **P < 0.01, ***P < 0.001.

Table 3.

Changes in Mechanical Withdrawal Threshold (MWT) Before and After Intervention in Each Group

Group	n	MWT Before Intervention	MWT After Intervention	Decline Rate (%)	P
NPDN-WL	7	7.23±3.08	0.62±0.05	91.47%	0.001**
NPDN-Con	3	8.00±4.54	7.77±0.97	2.88%	0.934
PDN-WL	7	7.06±3.94	0.63±0.05	91.08%	0.005**
PDN-Con	4	8.34±3.24	7.17±0.72	14.03%	0.458
NWL	5	13.06±3.02	8.78±1.80	32.77%	0.031*
NC	4	13.45±3.10	11.99±4.85	10.86%	0.531

Note: *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 3B further illustrates the comparison of MWT before and after the 10-week intervention across all six groups. The 95% confidence intervals presented in Table are unadjusted confidence intervals for the individual group means. Significant reductions in MWT were observed in the NPDN-WL and PDN-WL groups (P < 0.01), and in the NWL group (P < 0.05). In contrast, MWT in the remaining groups (NPDN-Con, PDN-Con, and NC) showed no statistically significant changes following the intervention period (all P > 0.05).

Table 4.

Comparison of Nerve Conduction Velocity and Amplitude Among Different Groups

Group	n	MNCV (m/s)		Amplitude (μV)		SNCV (m/s)		Amplitude (μV)
		Mean ± SD	95% CI	Mean ± SD	95% CI	Mean ± SD	95% CI	Mean ± SD	95% CI
NPDN-WL	7	28.16±7.10cd	[21.59, 34.73]	7.66±4.88	[3.15, 12.17]	19.55±5.04c	[14.88, 24.21]	3.92±3.16	[0.91, 6.93]
NPDN-Con	3	40.55±1.68b	[36.39, 44.72]	6.39±4.21	[−4.05, 16.84]	27.24±4.86abc	[15.18, 39.31]	1.41±0.97	[−1.00, 3.83]
PDN-WL	7	23.07±3.63d	[19.71, 26.42]	7.33±4.87	[2.82, 11.83]	18.86±4.73c	[14.48, 23.24]	2.09±1.73	[0.49, 3.69]
PDN-Con	4	36.67±3.76bc	[30.68, 42.65]	6.01±2.62	[1.83, 10.18]	26.04±4.66bc	[18.62, 33.45]	3.01±3.16	[−2.01, 8.03]
NWL	5	44.32±4.87ab	[38.27, 50.37]	7.36±2.69	[4.01, 10.70]	31.39±7.23ab	[22.41, 40.38]	1.44±1.11	[0.06, 2.82]
NC	4	52.63±1.65a	[50.00, 55.25]	9.79±6.91	[−1.20, 20.79]	38.84±2.25a	[35.26, 42.41]	4.22±2.02	[1.00, 7.43]
F		27.935		0.319		11.171		1.341
P		<0.001***		0.896		<0.001***		0.281

Note: Data are expressed as mean ± standard deviation. Different letters (^{a, b, c, d}) in the same column indicate statistically significant differences between groups (P < 0.05). ***P < 0.001.

Abbreviations: MNCV, Motor Nerve Conduction Velocity; SNCV, Sensory Nerve Conduction Velocity.

MWT in the NPDN-WL group dramatically decreased from (7.23 ± 3.08) g before intervention to (0.62 ± 0.05) g, representing a decline of 91.47% (P = 0.001). This finding indicates that the NPDN-WL rats developed marked mechanical hypersensitivity during the combined food-restriction and metformin intervention. MWT in the PDN-WL group decreased further from (7.06 ± 3.94) g to (0.63 ± 0.05) g (P < 0.01), indicating increased mechanical hypersensitivity during the intervention period. MWT in the NWL group also decreased from (13.06 ± 3.02) g to (8.78 ± 1.80) g (P < 0.05). In contrast, MWT in the corresponding control groups (NPDN-Con, PDN-Con, and NC groups) remained stable throughout the 10-week observation period (all P > 0.05) (Table 3).

Dynamic inter-group comparison demonstrated that beginning from week 2 of intervention, MWT in the NPDN-WL group was significantly lower than in the NPDN-Con group (P < 0.05) and rapidly declined to extremely low levels with continued intervention. MWT in the PDN-WL group at weeks 2, 4, 6, and the end of week 10 of intervention was significantly lower than in the PDN-Con group (all P < 0.01; Figure 3A).

Correlation Analysis Between the Mechanical Withdrawal Threshold and Degree of Body Weight Reduction

Spearman correlation analysis was performed on all rats that completed stage three (rapid weight reduction intervention) (n = 30) to assess the relationship between MWT during the weight reduction period and the degree of body weight reduction. Results demonstrated a significant negative correlation between MWT at the end of 10 weeks of weight reduction intervention and the percentage of body weight loss during the intervention period (Spearman’s ρ = −0.713, P < 0.001; Figure 4A). These findings suggest that the greater the magnitude of body weight reduction in rats, the more severe the mechanical hyperalgesia. Additionally, MWT at the end of the intervention was positively correlated with endpoint body weight (Spearman’s ρ = 0.745, P < 0.001; Figure 4B).

Figure 4.

Correlation Analysis of MWT with Percentage Body Weight Loss and Endpoint Body Weight in Rats after 10 Weeks of Rapid Weight Reduction. (A) Scatter plot showing the negative correlation between MWT and percentage of body weight loss at the end of the 10-week intervention (Spearman’s ρ = −0.713, P < 0.001). (B) Scatter plot showing a positive correlation between MWT and body weight at the intervention endpoint (Spearman’s ρ = 0.745, P < 0.001).

Nerve Conduction Velocity Detection Results

Motor Nerve Conduction Velocity (MNCV)

Following the weight-loss intervention, MNCV differed significantly among the six groups (F = 27.935, P < 0.001; Table 4). Bonferroni-adjusted post-hoc comparisons showed that MNCV was significantly lower in the NPDN-WL group than in the NPDN-Con group [(28.16 ± 7.10) m/s vs (40.55 ± 1.68) m/s; mean difference = −12.39 m/s, Bonferroni-adjusted 95% CI: −22.98 to −1.79; adjusted P = 0.013]. Similarly, MNCV was significantly lower in the PDN-WL group than in the PDN-Con group [(23.07 ± 3.63) m/s vs (36.67 ± 3.76) m/s; mean difference = −13.60 m/s, Bonferroni-adjusted 95% CI: −23.22 to −3.97; adjusted P = 0.002]. The difference between PDN-WL and NPDN-WL was not statistically significant after Bonferroni adjustment (mean difference = −5.10 m/s, Bonferroni-adjusted 95% CI: −13.30 to 3.11; adjusted P = 0.815). MNCV was also significantly lower in PDN-WL than in NWL (mean difference = −21.25 m/s, Bonferroni-adjusted 95% CI: −30.24 to −12.26; adjusted P < 0.001) and in PDN-Con than in NC (mean difference = −15.96 m/s, Bonferroni-adjusted 95% CI: −26.82 to −5.10; adjusted P = 0.001). In contrast, the difference between NWL and NC did not reach statistical significance after Bonferroni adjustment (mean difference = −8.31 m/s, Bonferroni-adjusted 95% CI: −18.60 to 1.99; adjusted P = 0.221). No significant differences in amplitude were observed among the six groups (F = 0.319, P = 0.896; Table 4). The key post-hoc comparison results are presented in Table 5. For the post-hoc pairwise comparisons presented in Tables 5 and 6, both the confidence intervals for the mean differences and the P values were Bonferroni-adjusted across all 15 possible pairwise comparisons following the six-group one-way ANOVA.

Table 5.

Pairwise Comparisons of MNCV Between Key Groups

Comparison	Mean diff. [Bonferroni-adjusted 95% CI]	Cohen’s d	Bonferroni-Adjusted P
PDN-WL vs PDN-Con	−13.60 [−23.22, −3.97]	−2.89	0.002**
NPDN-WL vs NPDN-Con	−12.39 [−22.98, −1.79]	−2.63	0.013*
PDN-WL vs NPDN-WL	−5.10 [−13.30, 3.11]	−1.08	0.815
PDN-Con vs NPDN-Con	−3.89 [−15.62, 7.84]	−0.83	1.000
NWL vs NC	−8.31 [−18.60, 1.99]	−1.76	0.221
PDN-WL vs NWL	−21.25 [−30.24, −12.26]	−4.51	< 0.001***
PDN-Con vs NC	−15.96 [−26.82, −5.10]	−3.39	0.001**

Note: P values and 95% confidence intervals for the pairwise mean differences were Bonferroni-adjusted across all 15 possible pairwise comparisons following the six-group one-way ANOVA. Cohen’s d: |d| < 0.5 small, 0.5–0.8 medium, ≥ 0.8 large. Mean diff. = mean difference. *P < 0.05, ** P < 0.01, ***P < 0.001.

Table 6.

Pairwise Comparisons of SNCV Between Key Groups

Comparison	Mean diff. [Bonferroni-Adjusted 95% CI]	Cohen’s d	Bonferroni-Adjusted P
PDN-WL vs PDN-Con	−7.18[−17.59, 3.24]	−1.41	0.513
NPDN-WL vs NPDN-Con	−7.69[−19.16, 3.77]	−1.51	0.580
PDN-WL vs NPDN-WL	−0.69 [−9.57, 8.19]	−0.14	1.000
PDN-Con vs NPDN-Con	−1.21 [−13.90, 11.48]	−0.24	1.000
NWL vs NC	−7.45 [−18.59, 3.70]	−1.46	0.594
PDN-WL vs NWL	−12.53 [−22.26, −2.81]	−2.46	0.005**
PDN-Con vs NC	−12.80 [−24.55, −1.05]	−2.51	0.024*

Notes: P values and 95% confidence intervals for the pairwise mean differences were Bonferroni-adjusted across all 15 possible pairwise comparisons following the six-group one-way ANOVA. Cohen’s d: |d| < 0.5 small, 0.5–0.8 medium, ≥ 0.8 large. Mean diff. = mean difference. *P < 0.05, **P < 0.01.

Sensory Nerve Conduction Velocity (SNCV)

SNCV differed significantly among the six groups (F = 11.171, P < 0.001; Table 4). Bonferroni-adjusted post-hoc comparisons showed that SNCV was lower in the NPDN-WL group than in the NPDN-Con group [(19.55 ± 5.04) m/s vs (27.24 ± 4.86) m/s; mean difference = −7.69 m/s, Bonferroni-adjusted 95% CI: −19.16 to 3.77; adjusted P = 0.580]. Similarly, SNCV was lower in the PDN-WL group than in the PDN-Con group [(18.86 ± 4.73) m/s vs (26.04 ± 4.66) m/s; mean difference = −7.18 m/s, Bonferroni-adjusted 95% CI: −17.59 to 3.24; adjusted P = 0.513]. Neither of these two comparisons reached statistical significance after Bonferroni adjustment. The corresponding Cohen’s d values were −1.51 and −1.41, respectively, indicating large standardized differences; however, these effect-size estimates should be interpreted as exploratory because of the small sample sizes and wide confidence intervals.

SNCV did not differ significantly between PDN-WL and NPDN-WL (mean difference = −0.69 m/s, Bonferroni-adjusted 95% CI: −9.57 to 8.19; adjusted P = 1.000) or between PDN-Con and NPDN-Con (mean difference = −1.21 m/s, Bonferroni-adjusted 95% CI: −13.90 to 11.48; adjusted P = 1.000). In contrast, SNCV was significantly lower in PDN-WL than in NWL (mean difference = −12.53 m/s, Bonferroni-adjusted 95% CI: −22.26 to −2.81; adjusted P = 0.005) and in PDN-Con than in NC (mean difference = −12.80 m/s, Bonferroni-adjusted 95% CI: −24.55 to −1.05; adjusted P = 0.024). SNCV was also lower in NWL than in NC, although this difference did not reach statistical significance after Bonferroni adjustment (mean difference = −7.45 m/s, Bonferroni-adjusted 95% CI: −18.59 to 3.70; adjusted P = 0.594). No significant differences in sensory nerve action-potential amplitude were observed among the six groups (F = 1.341, P = 0.281; Table 4). The key post-hoc comparison results are presented in Table 6.

Histopathological Examination Results

Following completion of nerve conduction velocity and MWT measurements, three rats from each group were randomly selected, euthanized, and the left sciatic nerve was isolated for HE staining. In addition, skin tissue from the plantar surface of the hind foot was collected for intraepidermal nerve fiber density (IENFD) detection.

Pathological Features of Sciatic Nerve HE Staining

HE staining showed varying degrees of histopathological abnormalities in the sciatic nerves across the experimental groups (Figure 5A–F).

Figure 5.

HE staining of the sciatic nerve after weight loss intervention. Each panel contains two micrographs at ×100 and ×200 magnification, designated with Roman numerals (eg, Figure 5Ai, Figure 5Aii). Hematoxylin and eosin (HE) staining revealed pathological characteristics of the sciatic nerve in each group. (A) NC group, nerve fibers were regularly arranged with intact myelin sheaths. (B) NWL group, mild nerve fiber injury was observed. (C) NPDN-Con group, uneven myelin density and axonal swelling were noted. (D) NPDN-WL group, aggravated demyelination (indicated by red arrows) and prominent vacuolar degeneration (indicated by green arrows). (E) PDN-Con group, disorganized nerve fiber architecture with focal demyelination. (F) PDN-WL group: severe demyelination (indicated by red arrows) and marked axonal swelling (indicated by green arrows).

NC group: Nerve fibers were arranged in a neat and compact manner with intact axonal structure, uniform myelin sheath thickness, regularly distributed Schwann cell nuclei, and no obvious abnormalities.

NWL group: Mild nerve fiber damage was observed compared to the NC group, mainly characterized by slight loosening of individual fiber arrangement, though the overall structure remained essentially intact.

NPDN-Con group: Nerve fiber structure exhibited mild disorganization with uneven myelin sheath density, and partial axons exhibited swelling.

NPDN-WL group: More prominent histopathological abnormalities were observed than in the NPDN-Con group. Manifestations included severe disorganization of nerve fiber arrangement, extensive myelin sheath loss (indicated by red arrows), obvious axonal swelling and degeneration, accompanied by widespread vacuolar degeneration (indicated by green arrows).

PDN-Con group: Exhibited disorganized nerve fiber structure with partial myelin sheath loss, with injury severity slightly more pronounced than in the NPDN-Con group.

PDN-WL group: Showed the most severe nerve damage among all groups. Severe myelin sheath loss was evident (indicated by red arrows), with highly swollen axons and loss of structural integrity (indicated by green arrows).

Changes in Intraepidermal Nerve Fiber Density (IENFD) at the Plantar Surface

Immunohistochemical staining with PGP9.5 of plantar skin revealed the following differences in morphology and density of intraepidermal nerve fibers (IENF) among groups: NC group showed nerve fibers with clear trajectory, uniform thickness, normal density, and uniform staining intensity. All experimental groups exhibited varying degrees of degenerative changes, characterized primarily by fiber shortening, non-uniform thickness, increased fiber breakage, and reduced staining intensity. Morphological observation revealed varying degrees of degenerative changes in all experimental groups, with more pronounced abnormalities in nerve fiber morphology and density observed in the weight-loss groups (NPDN-WL and PDN-WL; Figure 6).

Figure 6.

PGP9.5 immunohistochemical staining shows the characteristics of nerve fibers in each group. Each panel contains two micrographs at ×100 and ×400 magnification, designated with Roman numerals (eg, (A)i, (A)ii). (A) NC group, nerve fibers are clearly oriented, with uniform thickness. (B) NWL group, nerve fibers are clearly oriented, with uniform thickness. (C) NPDN-Con group, nerve fibers are shortened, with relatively uniform thickness. (D) NPDN-WL group, nerve fibers are obviously shortened, with uneven thickness. (E) PDN-Con group, nerve fibers are shortened, with relatively uniform thickness. (F) PDN-WL group, nerve fibers are obviously shortened, with uneven thickness, showing the most severe damage. Intraepidermal nerve fiber density in the plantar skin.

Descriptive analysis of IENFD showed that values in both the NPDN-WL and PDN-WL groups were lower than those in their respective control groups (Table 7), which is consistent with the behavioral and electrophysiological findings.

Table 7.

Comparison of IENFD in Rats Across Groups

Group	n	IENFD (Fibers/mm)
NPDN-WL	3	17.70 (15.28, 22.52)
NPDN-Con	3	32.18 (26.54, 37.81)
PDN-WL	3	19.31 (18.50, 25.74)
PDN-Con	3	25.74 (21.72, 37.00)
NWL	3	24.13 (20.11, 52.28)
NC	3	38.61 (18.50, 47.46)

Note: IENFD: Intraepidermal nerve fiber density of the plantar. Differences between groups were tested using non-parametric tests, and the differences were not statistically significant (P > 0.05).

Experimental Animal Attrition and Sample Size Distribution

Details of animal usage throughout the entire experiment are provided in Table S1.

During the first phase (diabetes modeling period), a total of three rats (identified as 19,20, and 22) died, all due to laboratory-related accidents: inadequate ventilation caused by the cage box failing to be retracted into its slot resulted in asphyxia deaths in all three rats. These fatalities occurred prior to randomization.

After successful modeling in the second phase (Week 11), a total of 42 rats survived, including 31 diabetic rats (17 in the spontaneous PDN group and 14 in the painless diabetes group) and 11 normal control rats. Two rats died during this stage: one in the normal control group exhibited ocular and nasal hemorrhage, likely due to internal bleeding resulting from fighting among co-housed rats (ID 30); one in the Spontaneous PDN group died despite maintaining normal diet without surface wounds or bleeding (ID 39). In the third phase (rapid weight-loss intervention period, Weeks 12–21), initial grouping was as follows: the PDN group (17 rats) was divided into PDN-WL (9 rats) and PDN-Con (8 rats); the NPDN group (14 rats) was divided into NPDN-WL (7 rats) and NPDN-Con (7 rats); and the normal group (11 rats) was divided into NWL (6 rats) and NC (5 rats).

A total of 12 rats were lost during the third phase, with mortality rates as follows for each group:

Two rats in the PDN-WL group were lost (from an initial total of 9 to 7): rats numbered 34 and 35 both died due to improper gastric intubation. On the day following intubation, they exhibited significant food refusal and lethargy; additionally, blood-tinged foam was observed in their oral cavities during the procedure, suggesting pulmonary injury caused by accidental airway entry.

The PDN-Con group lost four rats (from an initial cohort of eight to four at the end): rats numbered 24,29,41, and 47 died in a laboratory accident due to a nighttime malfunction in the animal housing air conditioning system. The repair took approximately 4–5 hours, during which sustained elevated ambient temperatures caused these rats to die from heat stress.

The NPDN-WL group showed no loss in sample size (initial: 7 samples → final: 7 samples).

The NPDN-Con group lost four units (from an initial total of seven to three at the end): all four died due to the same air conditioning failure event and perished in the same batch as those in the PDN-Con group, with unit numbers 5, 9, 10, and 23.

The NWL group lost one animal (from an initial cohort of six to five at the end): Animal No.6 died due to inadvertent tracheal intubation; it exhibited significantly reduced food intake and lethargy the day after intubation.

The NC group lost one animal (from an initial total of 5 to 4 at the end): one animal died due to improper gastric intubation procedure, identified as number 38.

In summary, the causes of death in the third phase were as follows: improper gastric intubation in 4 cases; and laboratory-related incidents (high ambient temperature due to air conditioning malfunction) in 8 cases. These unexpected deaths were promptly reported to the Ethics Committee for Experimental Animals of Guangzhou University of Chinese Medicine in accordance with institutional requirements.

Final sample size analysis: By the end of the experiment (week 21), a total of 30 rats were included in the final data statistics and pathological analysis, with the sample sizes per group as follows: PDN-WL group (7 rats), PDN-Con group (4 rats), NPDN-WL group (7 rats), NPDN-Con group (3 rats), NWL group (5 rats), and NC group (4 rats).

Analysis of PDN Model Success Rate and Conversion Rate

Spontaneous Model Success Rate

At the end of the spontaneous PDN phenotype observation and identification period (Week 11), 31 diabetic model rats were assessed by MWT testing. Results demonstrated that 17 rats met the PDN diagnostic criteria (MWT decline ≥ 50% from baseline), achieving a spontaneous success rate of 54.8% (17/31). The remaining 14 rats exhibited NPDN phenotype and entered the subsequent intervention phase.

Painful Phenotype Conversion Induced by Rapid Weight Reduction

Following 10 weeks of rapid weight reduction intervention, all 7 rats in the NPDN-WL group showed significantly decreased MWT compared to pre-intervention baseline (P < 0.01) and converted to PDN phenotype, achieving a complete phenotypic conversion (100%). In striking contrast, the 3 rats in the NPDN-Con group showed no conversion to painful phenotype during the same period (0% conversion rate). Fisher’s exact test confirmed that the conversion rate was significantly higher in the NPDN-WL group than in the NPDN-Con group (P = 0.008). These results indicate that the combined weight-loss intervention was associated with an increased proportion of NPDN rats meeting the PDN phenotype criterion. However, the small control-group sample size and the combined use of food restriction and metformin preclude definitive causal attribution to rapid weight loss alone.

Exploratory Comparison Between Spontaneous PDN and Weight-Loss-Induced TIND-Like PDN

To compare the two routes to the painful neuropathy phenotype, we performed an exploratory comparison between rats with spontaneous PDN (PDN-Con group, n = 4) and rats with weight-loss-induced TIND-like PDN (NPDN-WL group, n = 7; Table 8). Because of the small and unequal group sizes, the quantitative variables were summarized as median (25th percentile, 75th percentile) and compared using the two-sided Mann–Whitney U-test.

Table 8.

Comparison of Spontaneous PDN and Weight-Loss-Induced (TIND-Like) PDN

Indicator	Spontaneous PDN (PDN-Con; n = 4) M(P25, P75)	Weight-Loss-Induced PDN (PDN-WL; n = 7) M(P25, P75)	r	P
MWT at intervention endpoint (g)	6.99 (6.60,7.90)	0.60(0.60,0.70)	0.837	0.006**
MWT decline from baseline (%)	−10.78(−33.00,28.00)	−89.65(−92.60,-87.20)	0.798	0.008**
MNCV (m/s)	35.67(33.75,40.58)	22.00(20.75,25.33)	0.798	0.008**
SNCV (m/s)	24.92 (22.41,30.77)	18.67 (15.95, 22.80)	0.627	0.038*
Sciatic nerve histology (HE)	Mild–moderate disorganization; focal myelin loss	Severe demyelination, axonal swelling, vacuolar degeneration		–
Body weight changesdecline from intervention (%)	18.66(10.70,26.50)	−24.79(−28.70,-18.70)	0.798	0.008**
Random blood glucose at endpoint (mmol/L)	19.50(16.70,21.70)	10.30(9.30,12.30)	0.798	0.008**

Notes: Data are presented as median (25th percentile, 75th percentile). Spontaneous PDN was represented by the PDN-Con group, whereas weight-loss-induced TIND-like PDN was represented by the PDN-WL group. Quantitative variables were compared using the two-sided Mann–Whitney U-test. Effect size was calculated as |r| = |Z|/√N, where N represents the total number of animals included in the two-group comparison. The reported P values are unadjusted and should be interpreted as exploratory because multiple outcomes were examined. Negative percentage changes indicate decreases from baseline, whereas positive values indicate increases. Histopathological findings were compared descriptively and were not subjected to inferential statistical testing. *P < 0.05, **P < 0.01.

Abbreviations: MWT, mechanical withdrawal threshold; MNCV, motor nerve conduction velocity; SNCV, sensory nerve conduction velocity; HE, hematoxylin and eosin.

At the intervention endpoint, MWT was lower in the PDN-WL group than in the PDN-Con group [0.60 (0.60, 0.70) g vs 6.99 (6.60, 7.90) g; P = 0.006]. The percentage change in MWT from baseline was also greater in magnitude in the PDN-WL group [−89.65% (−92.60%, −87.20%)] than in the PDN-Con group [−10.78% (−33.00%, 28.00%); P = 0.008].

MNCV was lower in the PDN-WL group than in the PDN-Con group [22.00 (20.75, 25.33) m/s vs 35.67 (33.75, 40.58) m/s; P = 0.008; |r| = 0.798]. SNCV was also lower in the PDN-WL group [18.67 (15.95, 22.80) m/s] than in the PDN-Con group [24.92 (22.41, 30.77) m/s; P = 0.038; |r| = 0.627].

The median percentage change in body weight was −24.79% (−28.70%, −18.70%) in the PDN-WL group, indicating weight loss, compared with 18.66% (10.70%, 26.50%) in the PDN-Con group, indicating weight gain (P = 0.008). Endpoint random blood glucose was also lower in the PDN-WL group than in the PDN-Con group [10.30 (9.30, 12.30) mmol/L vs 19.50 (16.70, 21.70) mmol/L; P = 0.008].

Descriptively, sciatic nerve sections from the PDN-WL group showed more prominent demyelination, axonal swelling, and vacuolar degeneration than those from the PDN-Con group. However, because no formal histopathological severity score was obtained, this qualitative difference was not subjected to inferential statistical testing. Given the small sample sizes and the examination of multiple outcomes, these comparisons should be regarded as exploratory.

Discussion

Rapid Weight Reduction Significantly Promotes Painful Phenotype Conversion in Diabetic Rats

Conventionally, chronic hyperglycemia is considered the key risk factor for PDN development, with mechanisms involving advanced glycation end product (AGE) accumulation, increased oxidative stress, and impaired neuronal metabolism through classic pathways.¹¹ However, clinical observations reveal that the incidence of PDN does not show a simple linear relationship with blood glucose control levels, suggesting the existence of other key contributing factors.¹²

In recent years, treatment-induced diabetic neuropathy (TIND) has received widespread clinical attention.¹³ TIND specifically refers to newly developed or exacerbated peripheral neuropathy in some diabetic patients undergoing intensive glucose-lowering therapy with rapid blood glucose improvement. The core causative factor was traditionally attributed primarily to rapid glucose decline; however, the role of acute weight loss has increasingly gained emphasis in recent years. Notably, as a commonly used glucose-lowering and weight loss medication, metformin’s role in the development of diabetic neuropathy warrants attention.^14,15 This phenomenon provides critical clinical clues for exploring the potential adverse neural effects associated with rapid metabolic changes, and aligns well with our team’s clinically-based hypothesis of “weight loss-induced pain”.¹⁶ This study innovatively translates the clinical observations of TIND into an animal model. Through rapid weight reduction mediated by caloric restriction combined with metformin, we have, to our knowledge, at the animal level, suggested that rapid weight loss may be an important contributor to PDN. It should be emphasized that metformin in this study primarily served as a modeling tool to induce rapid weight reduction, rather than as a study of its direct pathogenic effect.

Currently, basic research on PDN mainly relies on diabetic rat models induced by high-fat diet combined with low-dose STZ to mimic the disease’s natural progression.¹⁷ However, such models commonly present with delayed appearance of painful phenotype and unstable and low spontaneous success rates.^18–20 This study showed that under hyperglycemic conditions alone, rats achieved only a 54.8% spontaneous painful phenotype conversion rate at Week 11, failing to fully replicate the clinical characteristic of some patients experiencing rapid onset of severe pain.

By introducing rapid weight reduction intervention through “caloric restriction combined with metformin,” this study reproduced several TIND-like phenotypic features at the animal level. Following intervention, all non-painful diabetic rats converted to PDN phenotype with a complete phenotypic conversion (100%), and MWT decreased further in rats with pre-existing PDN, indicating increased mechanical hypersensitivity during the intervention period.

This model innovation not only substantially improved the efficiency and reliability of PDN model construction but also provided an ideal platform for future exploration of the mechanisms by which metabolic fluctuations induce neural injury. Correlation analysis revealed that endpoint MWT was negatively correlated with the percentage of body weight loss (Spearman’s ρ = −0.713) and positively correlated with endpoint body weight (Spearman’s ρ = 0.745). These findings indicate that both the magnitude of body weight loss and endpoint body weight were associated with mechanical withdrawal threshold. However, because the analyses pooled animals from different intervention groups, the observed correlations may partly reflect between-group differences and should not be interpreted as individual-level causal relationships.

Characteristics of Neurological Dysfunction and Structural Injury Under Rapid Weight Reduction

This study showed through multi-dimensional assessment that rapid weight-reduction-induced pain hypersensitivity has a pathophysiological basis, manifesting as comprehensive damage from function to structure.

Impaired Nerve Conduction Function

Nerve conduction velocity is the gold standard for assessing large fiber function in peripheral nerves. MNCV was significantly lower in both weight-loss intervention groups than in their respective diabetic control groups. SNCV was also numerically lower in the weight-loss intervention groups, but the corresponding pairwise differences did not reach statistical significance after Bonferroni adjustment. Compound motor action potential (CMAP) amplitude, which reflects axonal number, revealed no significant differences among groups. This electrophysiological pattern suggests that early neural injury induced by rapid weight reduction appears to preferentially affect myelin integrity rather than causing widespread axonal loss.

The SNCV differences between NPDN-WL and NPDN-Con and between PDN-WL and PDN-Con were not statistically significant after Bonferroni adjustment (adjusted P = 0.580 and adjusted P = 0.513, respectively; Table 6). The corresponding Cohen’s d estimates were −1.51 and −1.41. However, these estimates are imprecise because of the small and unequal group sizes and should not be interpreted as evidence of statistically significant differences. Larger, adequately powered studies are required to determine whether the observed SNCV differences are reproducible.

Neuronal Structure and Myelin Injury

HE staining results of the sciatic nerve corroborate the above electrophysiological findings. The weight reduction intervention groups, particularly the PDN-WL group, exhibited severe myelin loss, vacuolar degeneration, and axonal swelling. The observed myelin abnormalities were consistent with the reductions in nerve conduction velocity. However, the present histological assessment was qualitative and cannot establish a direct mechanistic relationship between the structural abnormalities and electrophysiological changes.

Degenerative Changes in Intraepidermal Nerve Fiber Density

Small fiber neuropathy of the skin is a core feature of PDN, manifesting as loss of intraepidermal nerve fibers.²¹ Immunohistochemical observation using PGP9.5 in this study showed that rats in the weight reduction intervention groups exhibited obvious shortening and disruption of intraepidermal nerve fibers with degenerative morphological changes. Although due to sample size limitations the intergroup differences in IENFD did not reach statistical significance, the consistent declining trend correlated well with the sharp decrease in MWT.These observations suggest that the combined intervention may be associated with abnormalities involving both large myelinated fibers and the peripheral terminals of small sensory fibers. However, the IENFD differences were not statistically significant and should be interpreted cautiously.

Discussion on Potential Metabolic Mechanisms of Rapid Weight Reduction-Induced TIND

The present results suggest that painful phenotype conversion is closely associated with the magnitude of weight loss, with underlying mechanisms potentially stemming from energy crisis and adaptive disturbance induced by acute metabolic fluctuations.²²

During rapid weight reduction, organisms experience a dramatic shift from prolonged caloric excess to acute energy restriction. For peripheral nerves already under chronic hyperglycemic stress, such drastic changes in metabolic environment may trigger severe metabolic dysfunction. Axonal mitochondria, as the energy factory, may become dysfunctional due to substrate supply variability and demand pressure, leading to insufficient ATP generation and oxidative stress burst. The myelin loss and axonal swelling observed in this experiment represent morphological manifestations of how neural structural units cannot maintain their integrity and function under acute energy stress. Additionally, the healthy weight loss group (NWL) also exhibited significantly decreased mechanical withdrawal threshold (MWT) by 32.77% (P < 0.05), suggesting that rapid weight reduction itself may contribute to stress on neurons; however, this effect is substantially amplified in the diabetic background. The metabolic disturbance context combined with rapid weight reduction may constitute important conditions for painful phenotype conversion. The above propositions still require further molecular mechanistic verification. Future work will specifically target the candidate pathways of oxidative stress (eg, reactive oxygen species accumulation and lipid peroxidation), mitochondrial dysfunction (eg, impaired electron transport chain activity, ATP depletion, and dysregulated mitophagy), and neuroinflammation (eg, upregulation of TNF-α, IL-1β, and IL-6 with glial activation) in the sciatic nerve and dorsal root ganglia, to establish the molecular basis of weight-loss-induced neural injury.²³

Study Limitations

The present study has several limitations that should be acknowledged. First, substantial animal attrition occurred during Phase 3, with the PDN-Con group being most severely affected, which may limit statistical power for certain comparisons. Although effect-size estimates were consistently large, confirmation in adequately powered studies with balanced group sizes remains necessary. Causes of attrition included environmental control system failure, gavage-related procedural errors, aggressive cage-mate behavior, and one case of spontaneous death of unknown cause—all of which were considered unrelated to the experimental interventions. These findings underscore the need for improved procedural standardization and environmental monitoring. The reduced sample size should be considered when interpreting secondary outcome measures. Because attrition was disproportionate and non-random across groups (eg, PDN-Con: 8 to 4; NPDN-Con: 7 to 3), the possibility of attrition bias cannot be excluded; the direction of any resultant bias on the group comparisons is uncertain. Second, the correlation analysis between MWT and percentage body weight loss was performed on pooled data from all experimental groups (n = 30). As this analysis integrated data from different intervention groups, the observed correlation may be influenced by between-group clustering effects rather than purely reflecting individual-level causal relationships. Third, the weight-loss intervention combined caloric restriction with metformin treatment, so the independent effects of these two components are confounded and cannot be distinguished in the present design. Fourth, this study primarily focused on phenotypic characterization and preliminary mechanistic associations, without in-depth elucidation of specific molecular pathways (eg, oxidative stress, mitochondrial dysfunction, and neuroinflammation). Finally, the differences in IENFD between groups did not reach statistical significance; future studies with larger sample sizes are warranted to enhance detection power.²⁴

To address these limitations, future investigations will implement the following improvements: (1) expand sample sizes, particularly for control groups, to ensure adequate statistical power; (2) strengthen laboratory quality control, including refinement of gavage technique protocols, enhancement of environmental monitoring systems, and establishment of contingency plans for facility failures; (3) adopt a factorial (2 × 2: caloric restriction × metformin) design with independent control arms to explicitly separate the individual and combined effects of caloric restriction and metformin, thereby resolving the confounding of the two components in the current intervention; (4) quantify oxidative stress markers (eg, reactive oxygen species, malondialdehyde, superoxide dismutase, glutathione), assess mitochondrial function (eg, electron transport chain complex activities, ATP content, and mitophagy markers such as PINK1/Parkin), and profile inflammatory mediators (eg, TNF-α, IL-1β, IL-6, NF-κB) in the sciatic nerve and dorsal root ganglia, using molecular biology, proteomics, and other advanced techniques to further explore the molecular mechanisms underlying metabolic fluctuation-induced neural injury; and (5) validate the findings in multiple diabetes models to improve reproducibility and generalizability. These analyses will help distinguish primary pathogenic mechanisms from secondary adaptive responses.

Conclusion

This preliminary study suggests that a combined intervention consisting of food restriction and metformin treatment can reproduce several TIND-like features in diabetic rats, including marked reductions in MWT, impaired nerve conduction, and histopathological abnormalities. The findings indicate an association between the intervention and the development or worsening of a painful neuropathy phenotype; however, the independent contributions of food restriction, metformin treatment, weight reduction, and changes in blood glucose cannot be determined from the current design. The small and unequal control-group sample sizes further limit the precision and generalizability of the findings. Larger, adequately powered studies with separate food-restriction and metformin groups are required to validate the model and clarify the underlying mechanisms. The potential clinical relevance of rapid metabolic and weight changes should be evaluated in appropriately designed clinical studies.

Funding Statement

This work was supported by the Guangzhou Science and Technology Key R&D Program (No.2025B03J0049), Guangdong Provincial Administration of Traditional Chinese Medicine Research Project (No. 20251133), Guangdong Provincial Hospital of Traditional Chinese Medicine In-House Formulation R&D Special Project (No. YN2024MS040) and Guangdong Provincial Hospital of Chinese Medicine Chronic Disease Management Special Project (YN2024MB024). The funding was used primarily for animal purchase and maintenance, purchase of experimental consumables, and payment of laboratory facility fees. The funder had no role in the design, data collection, data analysis, and reporting of this study.

Data Sharing Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

Statement of Ethics

This study protocol was reviewed and approved by the Ethics Committee for Experimental Animals of Guangzhou University of Chinese Medicine (Ethics Approval Number: 20240402009). All animals were handled in accordance with the Regulations for the Administration of Affairs Concerning Experimental Animals (State Council of the People’s Republic of China). Euthanasia was performed by exposure to an overdose of isoflurane followed by cervical dislocation, consistent with the recommendations of the AVMA Guidelines for the Euthanasia of Animals.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; have agreed on the journal to which the article has been submitted; reviewed and agreed on all versions of the article before submission and during revision, gave final approval of the version to be published, and agreed on any significant changes introduced at the proofing stage; and agree to be accountable for all aspects of the work.

Disclosure

The authors report no conflicts of interest in this work.