A comparative meta-analysis of amniotic membrane preservation methods: Balancing biological potency and clinical feasibility in ophthalmology

Protocol and reporting

The review was reconstructed and reported in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020.^[3] The analytical plan pre-specified direct comparative human AM evidence, separation of biological from technical replication, and cautious quantitative synthesis when at least two studies reported sufficiently comparable outcomes.

Search strategy

PubMed/MEDLINE was searched on August 6, 2026, for publications dated January 1, 2000, through March 10, 2026. Medical Subject Headings and free-text terms combined amnion or AM with preservation terms (cryopreservation, freeze-drying, lyophilization, dehydration, drying, fresh tissue, preservation, or storage) and eligible outcomes (growth factors, cytokines, cell survival or viability, extracellular matrix, histology, ultrastructure, biomechanics, or structure). The complete executable search and dated PubMed export were retained in the audit workbook. Reference lists of retrieved studies were checked for contextual evidence, but no additional database yield was added to the PRISMA count.

Eligibility and study selection

Primary studies were eligible when they examined human AM and directly compared at least two preservation or preparation conditions. Eligible outcomes included growth-factor or cytokine content, endogenous membrane-cell viability, histological or ultrastructural integrity, extracellular matrix characteristics, mechanical properties, and relevant ophthalmic outcomes. Reviews, editorials, isolated case reports, studies without a direct preservation comparator, and experiments in which tissue composition differed between comparison groups without an interpretable preservation contrast were excluded. Animal or seeded-cell findings could inform the narrative synthesis only when the underlying human AM preparations were directly compared; they were not pooled with endogenous human-tissue outcomes.

The PubMed records underwent rule-assisted title and abstract screening followed by manual verification of the direct-comparison shortlist and full-text eligibility decisions. A single primary reason was assigned to each full-text exclusion. The full screening register, including unretrieved reports, was retained for audit. We do not characterize this workflow as duplicate independent human screening.

Data extraction and methodological assessment

Extracted variables included citation, donor or placenta denominator, anatomical region, processing conditions, comparator, assay, numerical outcome, dispersion measure, and unit of analysis. Standard errors (SEs) were converted to standard deviations (SDs) using SD=SE×√n. Technical replicates, image fields, or multiple fragments from one placenta were not treated as independent biological observations. Values available only in figures were digitized from high-resolution renderings and flagged as approximate.

Because the evidence was predominantly ex vivo, conventional clinical-trial instruments were not directly applicable. Methodological limitations were assessed across biological replication, group comparability, intervention definition, outcome measurement, completeness of reporting, unit-of-analysis integrity, and conflicts of interest. Judgments were low concern, some concerns, or high concern; they guided interpretation rather than a numeric quality score.

Quantitative synthesis

Positive continuous outcomes measured on different scales were summarized using the natural logarithm of the ratio of means and back-transformed for presentation. Ratios above 1.0 indicate a higher value in the numerator group. Random-effects models used restricted maximum likelihood estimation. Modified Hartung–Knapp 95% confidence intervals (CIs) were used because each synthesis contained only three studies. For paired donor designs without a reported SD of the within-donor difference, sampling variances assumed an intradonor correlation of 0.50; sensitivity analyses used correlations of 0, 0.25, and 0.75. Heterogeneity was described by I² and τ². Leave-one-out analysis was performed for epidermal growth factor (EGF). Funnel plots and statistical tests for publication bias were not used because fewer than 10 studies contributed to each synthesis. Correlated placental and reflected regions from the same Moraes donors were evaluated in alternative models and never entered simultaneously.

Study selection

The search identified 497 unique PubMed records. After title and abstract screening, 48 reports were sought; two could not be retrieved, and 46 were assessed in full text. Two reports were excluded after full-text assessment, leaving 44 studies in the systematic review. Six unique studies contributed to quantitative synthesis and 38 to narrative synthesis [Figure 1]. The characteristics of the six studies included in the quantitative synthesis are summarized in Table 1.

Close

Figure 1: PRISMA 2020 flow diagram.

Export to PPT

EGF: Dried or dehydrated versus frozen membrane

Three studies provided extractable EGF data for a dried or dehydrated versus frozen contrast.^[4-6] The pooled ratio of means was 1.54 (95% CI, 0.35–6.70; I² = 92.0%; τ² = 0.321), indicating no statistically supported difference [Figure 2]. Sensitivity analyses across assumed intradonor correlations produced point estimates of approximately 1.50–1.55 without changing the conclusion. Leave-one-out estimates were unstable; omission of Kim et al. reduced heterogeneity most clearly, emphasizing the influence of processing technology and study protocol.^[6]

Close

Figure 2: Preliminary EGF synthesis: Dried versus frozen HAM. EGF: Epidermal growth factor, HAM: Human amniotic membrane.

Export to PPT

Cell viability: Cryopreserved versus fresh membrane

Three studies contributed to an exploratory synthesis of endogenous cellular viability after cryopreservation compared with fresh or non-preserved AM.^[7-9] Moraes et al. reported placental and reflected regions from the same five donors.^[9] Using the placental-region estimate, the pooled ratio was 0.32 (95% CI, 0.03–3.11; I² = 96.2%; τ² = 0.764). Using the reflected-region estimate, the ratio was 0.44 (95% CI, 0.07–2.60; I² = 90.9%; τ² = 0.448) [Figure 3]. Both CIs included 1.0; the results indicate marked protocol-dependent variability rather than superiority of fresh or cryopreserved tissue.

Close

Figure 3: Exploratory cellular viability synthesis.

Export to PPT

Narrative structural and functional evidence

Comparative studies consistently showed that preservation can maintain a recognizable basement membrane and stromal collagen architecture while producing method-specific cellular, molecular, and mechanical changes. Early direct comparisons of fresh-frozen, freeze-dried, irradiated, air-dried, glycerol-preserved, and trehalose-treated preparations identified differences in epithelial preservation, thickness, extracellular-matrix immunoreactivity, growth-factor content, viability, and handling properties.^[10-20] Subsequent studies evaluated hypothermic or alternative-temperature storage, prolonged cryopreservation, lyopreservation, supercritical-carbon-dioxide processing, and repeated freeze-thaw procedures.^[21-26] Other direct comparisons addressed antibacterial activity, tissue-engineering properties, ultrastructure, glycerol supplementation, viability-supporting additives, decontamination, irradiation, cytokine content, and angiogenic modulation.^[27-35] Additional mechanistic studies examined inflammatory and angiogenic activity, tissue-bank storage, wound-healing mediators, cell-culture support, long-term storage, EGF distribution, ultrastructure, and fresh-versus-frozen clinical behavior.^[36-47] Recent electron-microscopy evidence based on three matched placentas found broadly preserved morphology after cryopreservation, lyophilization, and vacuum drying, but it did not provide a compatible continuous biochemical or viability endpoint.^[15]

Animal ocular-surface and seeded-cell experiments suggested that dried preparations can remain useful scaffolds, but these outcomes were not considered interchangeable with endogenous human AM viability or EGF content.^[13,14,18] A decellularized dehydrated product also showed qualitative differences from cryopreserved membrane in basement-membrane components and growth-factor immunoreactivity, illustrating that “dehydrated” products cannot be treated as a single uniform intervention.^[19]

Methodological assessment

The most frequent limitations were few independent donors, incomplete dispersion reporting, ambiguity between technical and biological replication, graphical or ordinal outcomes, and heterogeneous processing protocols. These limitations were particularly important for studies reporting multiple fragments, cells, or image fields from a small number of placentas. Quantitatively included studies were retained with explicit caution rather than excluded solely for methodological concerns [Figure 4].

Close

Figure 4: Methodological assessment of ex vivo preservation studies.

Export to PPT

Limitations

This review used one bibliographic database and may have missed studies not indexed in PubMed. Two sought reports were not retrieved. Screening and extraction were not performed by two independent human reviewers; rule-assisted organization and AI-assisted verification were followed by author review. Only six unique studies provided compatible quantitative data, and both syntheses contained three studies. Several denominators were small, paired correlations were unreported, and one regional estimate required graph digitization. The assumed intradonor correlation and alternative Moraes regions were therefore examined in sensitivity analyses. High heterogeneity, diverse assays, and product-specific processing limit generalizability. Publication bias could not be meaningfully evaluated.