Short-Term Audiometric Trajectories in Older Adults Receiving Systemic Human Umbilical Cord-Derived Mesenchymal Stem Cells: An Exploratory Single-Arm Substudy

Abstract

Purpose

To describe short-term longitudinal audiologic outcomes after systemic administration of human umbilical cord-derived mesenchymal stem cells (hUC-MSCs) in older adults with vascular aging.

Patients and Methods

This exploratory, single-arm audiologic substudy was embedded within a nonrandomized interventional parent trial of hUC-MSC administration for vascular aging. Participants were aged 60 to 80 years and met the parent study’s prespecified eligibility criteria. Audiometric outcomes were assessed repeatedly at baseline and on days 30, 90, and 180 after the first infusion. The primary exploratory audiologic outcome was the ear-level change from baseline to day 90 in the 4-frequency pure-tone average (PTA-4; 0.5, 1, 2, and 4 kHz), with negative values indicating lower hearing thresholds at follow-up. Mean changes were estimated using a linear mixed-effects model for repeated measures.

Results

Of 107 treated participants, 93 entered the hearing substudy (47 men [50.5%] and 46 women [49.5%]; mean [SD] age, 68.7 [5.1] years). After ear-level exclusions, 176 eligible ears from 91 participants comprised the analysis set. At day 90, 156 ears from 81 participants were evaluable. The model-estimated mean change in PTA-4 from baseline to day 90 was −0.77 dB HL (95% CI, −1.52 to −0.02). Among evaluable ears, the observed mean change from baseline to day 90 was −0.75 dB HL (95% CI, −1.52 to 0.02). Both estimates were less than 1 dB in magnitude and small relative to reported short-term audiometric test-retest variability. At days 30, 90, and 180, 97.2%, 95.5%, and 96.3% of evaluable ears, respectively, remained within the prespecified ±10-dB stability range.

Conclusion

In this exploratory single-arm substudy, mean PTA-4 changed little over short-term follow-up, and most evaluable ears remained within the prespecified stability range. Without a concurrent audiologic comparator, the observed trajectory cannot be attributed causally to hUC-MSC administration. These data may inform future controlled studies.

Parent Trial Registration

Chinese Clinical Trial Registry Identifier: ChiCTR2500101200; registered on April 22, 2025.

Plain Language Summary

Age-related hearing loss is common, but little is known about short-term changes in hearing tests among older adults who receive systemic human umbilical cord-derived mesenchymal stem cells (hUC-MSCs) in aging-related intervention studies. We analyzed repeated hearing tests from older adults with vascular aging who received hUC-MSCs. Average hearing thresholds changed little over 90 days, and most ears with available follow-up tests remained within a predefined stability range through day 180. Because the hearing substudy did not include an untreated group that underwent the same hearing assessments, it cannot determine whether hUC-MSCs improved or preserved hearing. The findings describe short-term hearing-test variability and may help inform the design of future controlled studies.

Introduction

Age-related hearing loss is common in later life and is associated with communication difficulties, social isolation, functional limitations, and poorer cognitive health.^1–5 Current management relies primarily on hearing devices and rehabilitation, and disease-modifying treatments remain limited.^6–8 Serial audiometry can provide clinically relevant trajectory data in studies of interventions targeting biological processes associated with aging.

Vascular and metabolic risk factors have been associated with hearing sensitivity in older adults.^9–11 Age-related strial degeneration, inflammation, and microvascular dysfunction may compromise cochlear function and increase pure-tone thresholds.^11–15 These processes overlap with vascular-aging biology and provide a rationale for longitudinal audiometric measures spanning conventional speech-frequency and higher-frequency domains. This biological overlap, however, does not establish that a systemic intervention will improve hearing.

Mesenchymal stem cells (MSCs) and MSC-derived extracellular vesicles have been investigated in preclinical models of auditory injury. Their hypothesized relevance is indirect, involving paracrine, immunomodulatory, and tissue-protective signaling rather than established cochlear regeneration.^16–21 Evidence relevant to systemic delivery remains preclinical, including intravenous administration of human MSCs after noise- or drug-induced hearing loss in rats,¹⁷ whereas the reported first-in-human auditory application of stromal cell-derived extracellular vesicles was intracochlear.²¹ These studies differ from the present population in disease context, delivery route, and outcome assessment and do not establish that systemic hUC-MSC administration alters human auditory thresholds. Clinical cell-therapy studies in aging-related conditions have also generally focused on non-audiologic outcomes.^22,23

Serial audiometric data in older adults after systemic hUC-MSC administration are therefore limited. The audiologic analysis plan, finalized before formal statistical analysis, selected day 90 4-frequency pure-tone average (PTA-4; 0.5, 1, 2, and 4 kHz) as the primary exploratory outcome because it provides a conventional summary across these frequencies; the greater baseline high-frequency impairment was examined separately in exploratory band-specific analyses. We sought to determine whether mean PTA-4 remained broadly stable or showed a directional change by day 90 and to describe supportive trajectories at days 30 and 180. Without a concurrent audiologic comparator, this substudy was not designed to test a causal treatment effect.

Materials and Methods

Study Design, Setting, and Ethics

This exploratory hearing substudy was embedded within the prospectively registered parent study, Clinical Research on the Evaluation of the Efficacy and Safety of Mesenchymal Stem Cell Therapy for Vascular Aging (ChiCTR2500101200), conducted at Chinese PLA General Hospital and the National Clinical Research Center for Geriatric Diseases.²⁴ The parent study evaluated the efficacy and safety of systemic hUC-MSC administration in adults with vascular aging, whereas the present substudy focused specifically on exploratory audiologic outcomes derived from serial pure-tone audiometry. Serial audiometry was implemented only in the hUC-MSC-treated cohort; therefore, the current substudy had no concurrent audiologic comparator. Audiometric testing was conducted as an additional exploratory measurement within the approved study procedures, and the resulting audiologic outcomes were not enumerated among the registered outcomes of the parent study.²⁴ The audiologic analysis plan was finalized before formal statistical analysis.

The parent-study protocol was approved by the Ethics Committee of Chinese PLA General Hospital (Approval No. S2025-195-01). All participants provided written informed consent for participation in the parent study. The hearing substudy and audiometric assessments were covered by the stated Ethics Committee approval and informed consent process. Reporting followed the Transparent Reporting of Evaluations with Nonrandomized Designs (TREND) statement.²⁵

Participants and Analysis Sets

The parent trial enrolled 107 adults aged 60 to 80 years who met the prespecified eligibility criteria for vascular aging between May and November 2025. Vascular-aging eligibility required at least one of the following: stable coronary artery disease, prior myocardial infarction with at least one coronary artery stenosis greater than 50%, or brachial-ankle pulse wave velocity greater than 1800 cm/s, together with the remaining protocol-defined clinical and safety requirements. Detailed participant eligibility criteria and data collection procedures are provided in the Supplementary Methods. Audiometric assessments were scheduled at baseline and on days 30 (±3 days), 90 (±7 days), and 180 (±7 days) after the first infusion.

Ear-level eligibility criteria were specified in the audiologic analysis plan finalized before formal statistical analysis. Ears were excluded if they had conductive hearing loss, mixed hearing loss, a history of sudden sensorineural hearing loss, or childhood-onset hearing loss. Participants could contribute one or two eligible ears. The eligible-ear analysis set included all participants with at least one eligible ear and all corresponding eligible ears. Visit-specific evaluable sets included eligible ears with the 4-frequency pure-tone average (PTA-4; 0.5, 1, 2, and 4 kHz) measurements at baseline and the corresponding follow-up visit. The complete-case set included eligible ears with PTA-4 measurements at baseline and all three follow-up visits. For ear-level analyses, changes from baseline were calculated by comparing measurements from the same ear across visits.

hUC-MSC Administration

Participants received four intravenous infusions of Amimestrocel Injection (Platinum Life Excellence Biotech Co., Ltd., Beijing, China), an allogeneic hUC-MSC product, at a dose of 1×10⁶ cells/kg per infusion. Use of Amimestrocel for vascular aging was investigational and outside its approved indication. Product handling and quality-control procedures are described in the Supplementary Methods.

Audiologic Assessments

At each scheduled assessment, the same experienced senior technician, blinded to previous audiometric results, performed pure-tone audiometry in a sound-isolated booth with background noise below 30 dB(A), using the same MADSEN Astera audiometer, TDH-39 headphones, and bone-conduction vibrator throughout the study. Hearing aids, when used, were removed immediately before testing. Thresholds were obtained using a modified Hughson–Westlake ascending procedure, and masking was applied when indicated; full masking procedures²⁶ are provided in the Supplementary Methods. Routine air- and bone-conduction testing was performed in accordance with GB/T 16296.1–2018. Full equipment, calibration, and test-environment details are provided in the Supplementary Methods. At baseline and each follow-up visit, otoscopy and tympanometry were performed to assess middle-ear status. For baseline ear-level etiologic classification, a clinically significant conductive component was defined as an air-bone gap of ≥15 dB HL at two or more tested frequencies (0.25, 0.5, 1, 2, and 4 kHz); the complete definitions used to classify conductive and mixed hearing loss are provided in the Supplementary Methods. PTA-4 was the mean threshold at 0.5, 1, 2, and 4 kHz.²⁷ Low-, mid-, and high-frequency PTA values were calculated as the mean thresholds at 0.125, 0.25, and 0.5 kHz; 0.5, 1, and 2 kHz; and 4 and 8 kHz, respectively. Hearing-loss severity categories and changes in severity for each ear were derived from PTA-4 using World Health Organization-based cut points.²⁸

Outcomes

The primary exploratory audiologic outcome was the change in PTA-4 from baseline to day 90, summarized principally as the model-estimated adjusted mean change. Supportive day 90 summaries included the observed change among day 90 evaluable ears. PTA-4 changes from baseline to days 30 and 180 were supportive outcomes. For descriptive analyses, response categories were defined using a study-specific 10-dB threshold: an ear was classified as improved if PTA-4 decreased by at least 10 dB, stable if the absolute change was less than 10 dB, or progressed if PTA-4 increased by at least 10 dB. This threshold was interpreted in the context of short-term test-retest variability reported in other audiometric settings.^29,30

Other outcomes included frequency-band and single-frequency changes, hearing-loss severity transitions, word recognition score (WRS), missingness assessments, sensitivity analyses, and exploratory subgroup summaries. Definitions and methods for these outcomes are provided in the Supplementary Methods.

Statistical Analysis

Continuous variables were summarized as mean (SD) or median (interquartile range), as appropriate, and categorical variables as number and percentage. Missing values and varying denominators were reported where relevant. No formal sample-size calculation was performed because the sample was determined by available serial audiometry. Accordingly, the analyses were exploratory, and the available sample was not designed to establish or exclude a clinically meaningful treatment effect. Estimates were reported with 95% confidence intervals (CIs), and P values were descriptive.

The primary analysis used a linear mixed-effects model for repeated ear-level PTA-4 measurements across visits.^31,32 Fixed effects were visit, ear side, age, and sex. Random intercepts were included for participants and for ears nested within participants to account for within-participant bilateral correlation and repeated measurements within ears.^31,33 Participants represented the independent observational units, with ear-level measurements modeled as clustered within participants. Baseline PTA-4 formed part of the repeated outcome trajectory and was not entered as a separate covariate. The primary contrast was the model-estimated mean change from baseline to day 90; estimates were reported with 95% CIs.

Supportive day 90 analyses included the observed change among evaluable ears and an adjusted day 90 change-score model restricted to ears with PTA-4 measurements available at both baseline and day 90. Two participant-level sensitivity analyses reduced bilateral ear data to one contribution per participant: eligible-ear PTA-4 values were averaged within participant at each visit before longitudinal modeling, and eligible-ear day 90 changes were averaged within participant for a one-sample Wilcoxon signed-rank analysis. Model details are provided in the Supplementary Methods. Band-specific adjusted models estimated changes at days 30, 90, and 180 among ears with baseline PTA of at least 20 dB HL in the corresponding frequency band.

The primary linear mixed-effects model used all available repeated observations and did not impute missing day 90 values. This observed-data, likelihood-based approach is consistent with general guidance for longitudinal analyses with incomplete data, while its validity depends on assumptions about the missing-data mechanism.^34–37 Sensitivity analyses included complete-case, baseline observation carried forward (BOCF), subject-level, cluster-robust mean, participant-level Wilcoxon, and follow-up availability analyses. Exploratory subgroup estimates were descriptive, and no confirmatory interaction tests were performed. Full model specifications and sensitivity procedures are provided in the Supplementary Methods.

Safety events were summarized descriptively for all 107 hUC-MSC-treated participants using parent-study records collected through active inquiry at study visits and review of hospitalization and medical records. Investigators assessed event relatedness by clinical judgment. The comprehensive parent-trial safety analysis will be reported separately. All analyses were performed using R version 4.4.1.

Results

Participant Flow and Baseline Characteristics

Figure 1 shows participant flow. Of 107 participants in the parent study, 93 entered the hearing substudy and 14 did not because the substudy had not yet begun. After excluding 2 ears from 1 participant without baseline audiometry and 8 ears from 7 participants for prespecified etiologic reasons, 176 eligible ears from 91 participants comprised the eligible-ear full analysis set.

Figure 1.

Hearing Substudy Participant Flow. The diagram shows the derivation of the hearing substudy, eligible-ear full analysis set, and visit-specific evaluable sets. Each evaluable set required PTA-4 at baseline and the corresponding follow-up. PTA-4 indicates 4-frequency pure-tone average. Bold text indicates participant and ear counts. hUC-MSC indicates human umbilical cord-derived mesenchymal stem cells.

Of the 93 participants, the mean (SD) age was 68.7 (5.1) years; 47 (50.5%) were men and 46 (49.5%) were women. Among 176 eligible ears, the mean baseline PTA-4 and high-frequency PTA were 27.9 dB HL and 46.8 dB HL, respectively. Some degree of hearing loss was present in 135 ears (76.7%) and was most commonly mild (91 ears [51.7%]) (Table 1 and Supplementary Table 1).

Table 1.

Baseline Participant and Audiologic Characteristics

Characteristic	Value
Participant characteristics (n = 93)
Age, mean (SD), y	68.7 (5.1)
Sex, No. (%)
Female	46 (49.5)
Male	47 (50.5)
Body mass index, mean (SD)	24.3 (2.6)
Hypertension, No. (%)	53 (57.0)
Diabetes, No. (%)	29 (31.2)
Hyperlipidemia, No. (%)	56 (60.2)
Smoking history, No. (%)	15 (16.1)
Noise history, No./total No. (%)	10/90 (11.1)
Self-reported hearing loss, No./total No. (%)	49/92 (53.3)
Tinnitus, No./total No. (%)	27/92 (29.3)
Baseline audiologic characteristics (n = 176 ears)
Ear side, No. (%)
Right	90 (51.1)
Left	86 (48.9)
PTA-4a, mean (SD), dB HL	27.9 (10.8)
Low-frequency PTAb, mean (SD), dB HL	22.4 (8.5)
Mid-frequency PTAb, mean (SD), dB HL	23.9 (10.1)
High-frequency PTAb, mean (SD), dB HL	46.8 (19.4)
Word recognition score, mean (SD), %	96.0 (6.8)
Baseline hearing-loss severityc, No. (%)
Normal hearing	41 (23.3)
Mild hearing loss	91 (51.7)
Moderate hearing loss	38 (21.6)
Moderately severe hearing loss	6 (3.4)

Notes: Percentages are based on nonmissing data. ^aPTA-4 is the mean threshold at 0.5, 1, 2, and 4 kHz. ^bLow-, mid-, and high-frequency PTA use 0.125/0.25/0.5, 0.5/1/2, and 4/8 kHz, respectively. ^cHearing-loss severity was defined using WHO-based PTA-4 cut points: normal, <20; mild, 20 to <35; moderate, 35 to <50; and moderately severe, 50 to <65 dB HL.

Abbreviations: dB HL, decibels hearing level; PTA, pure-tone average; PTA-4, 4-frequency pure-tone average; SD, standard deviation.

At day 90, 156 ears from 81 participants were evaluable; follow-up completeness was lower at day 30 (106 ears from 54 participants) and day 180 (109 ears from 57 participants) (Figure 1). Nonavailability was attributable mainly to missed visits (25 of 37 participants at day 30, 9 of 10 at day 90, and 12 of 34 at day 180) and declined audiologic assessment (7 of 37, 0 of 10, and 17 of 34, respectively); the remaining cases reflected scheduling or personal constraints (5 at day 30 and 1 at day 90) or loss to follow-up (5 at day 180) (Supplementary Table 2). Compared with participants without evaluable data, those evaluable at day 30 were younger (mean, 67.7 vs 70.3 years), more often female (59.3% vs 37.8%), less often hypertensive (46.3% vs 73.0%), and had a lower mean baseline PTA-4 (26.85 vs 29.95 dB HL). At day 180, evaluable participants were older (69.5 vs 67.4 years) and more often hypertensive (64.9% vs 44.1%); day 90 groups were more similar overall, although hypertension prevalence was 59.3% versus 40.0% (Supplementary Table 3 and Supplementary Figure 1).

Primary Exploratory PTA-4 Outcome

In the primary linear mixed-effects model, the model-estimated mean change in PTA-4 from baseline to day 90 was −0.77 dB HL (95% CI, −1.52 to −0.02) (Table 2 and Supplementary Table 4). Supporting model-estimated changes were −0.75 dB HL (95% CI, −1.61 to 0.12) at day 30 and −0.51 dB HL (95% CI, −1.37 to 0.34) at day 180 (Table 2 and Supplementary Table 4).

Table 2.

Primary and Key Supportive Audiologic Outcomes

Outcome or Analysis Set	Participants, No.	Ears, No.	Estimate (95% CI) or No. (%)
PTA-4 change from baseline, dB HL
Primary LMM: day 90 vs baseline	91	176	−0.77 (−1.52 to −0.02)
Model-estimated change at day 30	54	106	−0.75 (−1.61 to 0.12)
Model-estimated change at day 180	57	109	−0.51 (−1.37 to 0.34)
Observed change from baseline to day 90	81	156	−0.75 (−1.52 to 0.02)
Adjusted day 90 change-score model	81	156	−0.74 (−1.53 to 0.04)
Day 30 response categories
Improved by 10 dB or more	54	106	2 (1.9)
Stable within ±10 dB	54	106	103 (97.2)
Progressed by 10 dB or more	54	106	1 (0.9)
Day 90 response categories
Improved by 10 dB or more	81	156	7 (4.5)
Stable within ±10 dB	81	156	149 (95.5)
Progressed by 10 dB or more	81	156	0 (0.0)
Day 180 response categories
Improved by 10 dB or more	57	109	3 (2.8)
Stable within ±10 dB	57	109	105 (96.3)
Progressed by 10 dB or more	57	109	1 (0.9)
Frequency-band changes at day 90
Low-frequency PTA	81	156	−1.36 (−2.71 to 0.06)
Mid-frequency PTA	81	156	−0.98 (−1.82 to −0.11)
PTA-4	81	156	−0.75 (−1.52 to 0.02)
High-frequency PTA	81	156	−1.07 (−2.16 to 0.03)

Notes: Changes are follow-up minus baseline; negative values indicate lower follow-up thresholds. Visit-specific counts reflect observed data; the primary LMM used all available measurements. The primary estimate compares day 90 with baseline. The LMM adjusted for visit, ear side, age, and sex and included random intercepts for participants and ears nested within participants. The 95% CI for the observed change from baseline to day 90 was estimated using a 10,000-replicate participant-cluster bootstrap. Response categories were improved (≤−10 dB), stable (>−10 to <10 dB), and progressed (≥10 dB). Frequency-band estimates were based on observed changes from baseline to day 90 among evaluable ears. Low-, mid-, and high-frequency PTA used 0.125/0.25/0.5, 0.5/1/2, and 4/8 kHz, respectively; PTA-4 used 0.5/1/2/4 kHz.

Abbreviations: CI, confidence interval; dB HL, decibels hearing level; LMM, linear mixed-effects model; PTA, pure-tone average; PTA-4, 4-frequency pure-tone average.

Among 156 evaluable ears, the observed mean change in PTA-4 from baseline to day 90 was −0.75 dB HL (95% CI, −1.52 to 0.02). The adjusted day 90 change-score model estimated a mean change of −0.74 dB HL (95% CI, −1.53 to 0.04) (Table 2). The three day 90 estimates were directionally similar and all less than 1 dB in magnitude, but only the 95% CI from the primary model excluded 0.

Longitudinal Audiologic Patterns and Individual Responses

Mean audiometric profiles were similar across visits, with higher thresholds at 4 to 8 kHz than at lower frequencies (Figure 2A); single-frequency changes are reported in Supplementary Table 5 and Supplementary Figure 2. Observed mean low-, mid-, and high-frequency PTA and PTA-4 values were similar across visits (Figure 2B). Mild hearing loss was the most common category among observed ears at baseline and each follow-up (Figure 2C).

Figure 2.

Integrated Longitudinal Audiologic Profiles and Individual PTA-4 Changes. (A) shows mean pure-tone thresholds by visit; (B) observed mean low-, mid-, and high-frequency PTA and PTA-4 values, with error bars indicating 10,000-replicate participant-cluster bootstrap 95% CIs; (C) hearing-loss severity distributions with visit-specific denominators; and (D) ranked changes in PTA-4 for individual ears at days 30, 90, and 180. In (A), the visit-specific sample sizes shown in the legend apply to all frequencies except 6 kHz; at 6 kHz, 82, 36, 55, and 38 ears contributed at baseline and days 30, 90, and 180, respectively. For display in (D), change was calculated as baseline minus follow-up; positive values indicate lower thresholds at follow-up, and dashed lines mark the 10-dB improvement and progression thresholds. High-frequency PTA used thresholds at 4 and 8 kHz. In (A), line colors identify visits. In (B), colors identify PTA bands. In (C), fill colors identify hearing-loss severity categories. In (D), green, gray, and red bars denote improved, stable, and progressed ears, respectively. PTA indicates pure-tone average; PTA-4, 4-frequency PTA at 0.5, 1, 2, and 4 kHz. CI indicates confidence interval; “Mod. severe”, moderately severe hearing loss.

Among evaluable ears, 103 of 106 (97.2%) at day 30, 149 of 156 (95.5%) at day 90, and 105 of 109 (96.3%) at day 180 remained within the study-specific descriptive stability range of ±10 dB (Figure 2D and Table 2). Counts classified as improved or progressed at each visit are reported in Table 2 and Figure 2D. WRS results are reported in Supplementary Table 6 and Supplementary Figure 3.

Band-Specific Changes Across Follow-Up Visits

Among 156 evaluable ears, observed mean changes from baseline to day 90 ranged from −1.36 to −0.75 dB HL across low-frequency PTA, mid-frequency PTA, PTA-4, and high-frequency PTA (Table 2). In analyses restricted to ears with baseline PTA of at least 20 dB HL in the corresponding band, adjusted day 90 changes were −3.08 dB HL (95% CI, −4.49 to −1.68) for low-frequency PTA and −1.31 dB HL (95% CI, −2.32 to −0.30) for mid-frequency PTA. Corresponding estimates were −0.83 dB HL (95% CI, −1.67 to 0.00) for PTA-4 and −1.31 dB HL (95% CI, −2.36 to −0.25) for high-frequency PTA (Figure 3). At days 30 and 180, estimates were negative across all bands; low-frequency estimates were the most negative at both visits (Figure 3).

Figure 3.

Exploratory Band-Specific PTA Changes Among Ears With Baseline PTA of at Least 20 dB HL. The forest plot shows adjusted mean changes at days 30, 90, and 180 among ears with baseline PTA of at least 20 dB HL in the corresponding band. Blue circles, orange squares, and red triangles represent days 30, 90, and 180, respectively; horizontal bars indicate 95% CIs. The vertical dashed line denotes no change. The value n denotes ears contributing at least one postbaseline observation to the corresponding longitudinal model. Changes are follow-up minus baseline; negative values indicate lower thresholds. Low-, mid-, PTA-4, and high-frequency PTA used 0.125/0.25/0.5, 0.5/1/2, 0.5/1/2/4, and 4/8 kHz, respectively. CI indicates confidence interval; PTA, pure-tone average.

Robustness and Exploratory Subgroup Analyses

Across the day 90 robustness analyses, estimates ranged from −1.27 to −0.63 dB HL; all were negative, but several 95% CIs included 0. The subject-level longitudinal model estimated a day 90 change of −0.67 dB HL (95% CI, −1.44 to 0.10), and the participant-level Wilcoxon analysis yielded a Hodges–Lehmann pseudomedian of −0.63 dB HL (95% CI, −1.56 to 0.31); each analysis used one value per participant (Figure 4A).

Figure 4.

Day 90 Robustness Analyses and Exploratory Subgroup Summaries. (A) shows the primary day 90 PTA-4 estimate and robustness analyses; (B) shows observed PTA-4 changes by exploratory subgroup. For the cluster-robust mean, eligible-ear changes were averaged within participant. The subject-level mean and participant-level Wilcoxon analyses used one value per participant; ear counts were therefore not applicable. Age and baseline PTA-4 cutoffs were 69 years and 25 dB HL, respectively. Because subgroup analyses were ear based, the two ears of a participant could enter different baseline PTA-4 strata. Squares indicate point estimates, horizontal bars indicate 95% CIs, and vertical dotted lines denote no change; BOCF and subgroup intervals used participant-cluster bootstrap. Bold text in (B) identifies subgroup headings and does not indicate statistical significance. Changes are follow-up minus baseline; negative values indicate lower thresholds. No confirmatory interaction tests were performed. BOCF indicates baseline observation carried forward.

Abbreviations: CI, confidence interval; LMM, linear mixed-effects model; PTA-4, 4-frequency pure-tone average at 0.5, 1, 2, and 4 kHz.

Exploratory subgroup estimates ranged from −1.55 to 1.88 dB HL (Figure 4B). The widest 95% CI was observed for ears with reported noise exposure (−0.89 to 4.46 dB HL; 14 ears). No confirmatory interaction tests were performed. Primary model diagnostic plots are shown in Supplementary Figure 4.

Safety

Among all 107 hUC-MSC-treated participants in the parent trial, two serious adverse events were reported: hospitalization for coronary heart disease and pulmonary embolism. The investigators assessed both events as possibly unrelated to the study treatment based on clinical judgment; both resulted in permanent treatment discontinuation. Both participants were clinically stable at the latest available follow-up. No deaths, persistent sequelae, additional adverse events, or infusion-related reactions were reported. This brief summary is provided for context and does not constitute the comprehensive parent-trial safety analysis, which will be reported separately.

Discussion

This exploratory, single-arm audiologic substudy characterized short-term audiometric trajectories after systemic hUC-MSC administration in older adults with vascular aging. The primary day 90 PTA-4 estimate was −0.77 dB HL; the observed and adjusted estimates were similar in direction and magnitude, but exclusion of 0 by the 95% CI was not consistently reproduced across supportive analyses. Most evaluable ears remained within the prespecified ±10-dB stability range at each follow-up. In ears with baseline PTA of at least 20 dB HL in the corresponding band, adjusted estimates were negative across all follow-up visits and were most negative for low-frequency PTA. In the absence of a concurrently assessed hearing comparator, these findings describe an observed trajectory and do not establish treatment efficacy.

The absence of a comparator limits interpretation of the near-baseline trajectory. Although the primary model’s 95% CI narrowly excluded 0, the approximately 0.8-dB day 90 estimate was small relative to reported short-term audiometric test-retest variability; statistical exclusion of 0 in this exploratory analysis should not be interpreted as clinically meaningful hearing improvement.^29,30 Because the band-specific analyses selected ears with baseline PTA of at least 20 dB HL in the corresponding band, their negative estimates may partly reflect regression to the mean. Follow-up completeness was lower at days 30 and 180, and baseline characteristics differed according to follow-up availability. These patterns raise the possibility of differential follow-up, while the missing-data analyses rely on assumptions that cannot be fully verified.^34–37 The observed changes also cannot be separated from the natural course of age-related hearing loss.^38,39 Baseline vascular and metabolic comorbidities were described, but changes in these conditions and exposure to potentially ototoxic medications during follow-up were not available as time-varying covariates; their influence on hearing thresholds therefore could not be separated from the observed trajectory. Because the hearing outcome was exploratory and was not included among the parent study registered outcomes, the analyses provide descriptive estimates and feasibility information rather than a causal treatment contrast.

Preclinical MSC studies have explored regenerative and neurotrophic effects in auditory injury models.^16–19 General MSC secretome literature describes paracrine, immunomodulatory, and tissue-protective mechanisms.⁴⁰ Within experimental auditory injury models, MSC-derived extracellular vesicles have also been reported to exert immunomodulatory and neuroprotective effects.^18,20,21 The preclinical auditory models differ from older adults with heterogeneous age-related hearing loss in disease context, delivery route, and outcome assessment. Systemic administration in a vascular aging cohort addresses a different question from locally targeted regenerative otology studies.²¹ No mechanistic or biomarker outcomes were available in this substudy, so the observed audiometric trajectories could not evaluate these proposed pathways.

The principal contribution of this substudy is the provision of short-term estimates of audiometric change, response-category distributions, and follow-up-completeness data in an older intervention cohort. These data may inform endpoint definitions, expected variability, follow-up schedules, and sample-size assumptions for future studies. Controlled studies should prespecify hearing outcomes, include a concurrently assessed audiologic comparator, and incorporate mechanistic measures when biological pathways are evaluated.

Strengths included repeated audiometric assessments of individual ears across three follow-up time points, an analysis plan finalized before formal statistical analysis that defined the eligible-ear population and the primary exploratory day 90 audiologic outcome, models that accounted for bilateral-ear and repeated-measure correlations, and multiple robustness analyses. Recalculation of PTA measures from frequency-specific thresholds and reporting of estimates with 95% CIs further supported analytic transparency.

In addition to the interpretive constraints arising from the single-arm design, measurement variability, and differential follow-up discussed above, this study has several further limitations. First, no formal sample-size calculation was performed for the hearing substudy; the available sample therefore limited precision and treatment-effect inference. Second, the hearing outcome was exploratory relative to the parent study registration, and the 180-day follow-up did not characterize longer-term change. Third, participants were selected from a single-center vascular aging trial, which may limit generalizability to broader populations of older adults with age-related hearing loss. Finally, the study-specific response categories describe threshold change and should not be interpreted as validated definitions of hearing preservation or clinical benefit.

Conclusions

In this exploratory single-arm audiologic substudy of older adults with vascular aging who received systemic hUC-MSC administration, mean PTA-4 remained broadly stable over short-term follow-up. The observed mean day 90 change was very small (−0.75 dB HL), and most evaluable ears remained within the study-specific descriptive stability range. Without a concurrently assessed audiologic comparator, causality cannot be inferred, and the observed trajectory should not be interpreted as evidence that hUC-MSC administration preserved or improved hearing. These hypothesis-generating data may inform the design of future studies. Controlled studies with longer follow-up and prespecified audiologic outcomes are required to determine whether such trajectories differ from those expected without the intervention.

Funding Statement

This study was supported by grants from the National Key Research and Development Program of China (2023YFC2508400, 2022YFC3602404, and 2023YFF1203504) and the National Natural Science Foundation of China (82671505). The investigational product used in the parent study was supplied by Platinum Life Excellence Biotech (Beijing) Co., Ltd. The governmental funders had no role in the design or conduct of the hearing substudy; collection, management, analysis, or interpretation of the audiologic data; preparation, review, or approval of the manuscript; or the decision to submit the manuscript for publication. Platinum Life Excellence Biotech (Beijing) Co., Ltd. manufactured and supplied the investigational product for the parent study but had no institutional role in protocol design, participant recruitment, data collection, audiologic data management, statistical analysis, interpretation of the audiologic findings, or the decision to submit the manuscript. Xiaoqiang Hou, an employee of the company, undertook manuscript review and final approval in his personal capacity as an author; these author responsibilities do not represent institutional participation by the company.

Abbreviations

BOCF, baseline observation carried forward; CI, confidence interval; dB HL, decibels hearing level; hUC-MSC, human umbilical cord-derived mesenchymal stem cell; LMM, linear mixed-effects model; PTA, pure-tone average; PTA-4, 4-frequency pure-tone average; SD, standard deviation; TREND, Transparent Reporting of Evaluations With Nonrandomized Designs; WHO, World Health Organization; WRS, word recognition score.

Data Sharing Statement

Consistent with the parent-trial registration, deidentified participant-level data necessary to reproduce the reported audiologic findings, together with the data dictionary, study protocol, and final audiologic substudy statistical analysis plan, are planned to be available from May 1, 2028, through April 30, 2033, via the ResMan Clinical Trial Management Public Platform (www.medresman.org.cn). Access will be considered for qualified researchers submitting a methodologically sound proposal consistent with participants’ informed consent and will be subject to institutional, ethical, privacy, and platform requirements, including a data-use agreement where required. Requests should be submitted through ResMan or directed to the corresponding author at caofeng@301hospital.com.cn and wqjavm301@sina.com. Analytic code will not be publicly available but may be provided confidentially to the journal for verification, subject to institutional approval.

Ethics Approval and Informed Consent

Ethics approval for the parent vascular-aging study was provided by the Ethics Committee of Chinese PLA General Hospital (Approval No. S2025-195-01). The Ethics Committee–approved protocol and written informed-consent process covered the audiometric assessments and the research use of the resulting data reported in this exploratory hearing substudy. The study was conducted in accordance with the Declaration of Helsinki and applicable institutional requirements.

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; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors report support from the National Key Research and Development Program of China, the National Natural Science Foundation of China, and Platinum Life Excellence Biotech (Beijing) Co., Ltd., during the conduct of the study. Xiaoqiang Hou is an employee of Platinum Life Excellence Biotech (Beijing) Co., Ltd. The authors report no other conflicts of interest in this work.