In Vivo Efficacy of LL-37 and Its Derivative Peptides Against Methicillin-Resistant Staphylococcus aureus in Animal Models: A Systematic Review and Meta-Analysis

1. Introduction

Methicillin resistant Staphylococcus aureus is one of the leading causes of severe bacterial infection worldwide, imposing substantial morbidity, mortality and health system burden and contributing importantly to the larger global crisis of antimicrobial resistance. Large-scale analyses estimate that bacterial antimicrobial resistance contributed to nearly 5 million deaths in 2019, with resistant pathogens such as Staphylococcus aureus being major contributors to this burden (Murray et al., 2022). Beyond mortality, MRSA infections impose an enormous economic burden, with healthcare costs amounting to several billion US dollars annually due to prolonged hospitalization, complex treatment regimens, and increased rates of complications (Hirabayashi et al., 2024). Alarmingly, even last-resort therapies such as vancomycin are facing reduced susceptibility, characterized by progressive increases in minimum inhibitory concentrations (“MIC creep”) (Matei & Visan, 2025), further limiting effective treatment options and underscoring the urgent need for alternative therapeutic strategies, including antimicrobial peptides such as LL-37.

Novel therapeutic approaches are therefore an urgent priority because conventional antibiotic pipelines are limited and failed therapy for complicated skin wound infections, device associated infections and invasive disease continues to produce poor clinical outcomes and high costs (Thi Phuong et al., 2022). Epidemiologic and clinical reports document persistent incidence of invasive Staphylococcus aureus infection in high income and low resource settings alike and highlight the continuing public health relevance of strategies that go beyond classical antibiotics to restore bacterial control and support tissue repair (Kourtis & Hatfield, 2019; Zhen et al., 2020).

Host defence peptides represent a promising therapeutic class because these molecules combine direct antimicrobial activity with host immune modulatory and tissue protective functions (Steinstraesser et al., 2009). The human cathelicidin LL-37 is the single human cathelicidin peptide and it acts at the interface of innate immunity and tissue repair (Dürr et al., 2006). LL-37 is expressed by neutrophils and multiple epithelial cell types and is released at sites of infection and injury where it can bind microbial surfaces, destabilize membranes and inhibit biofilm formation while also acting on host cells to modulate inflammation and recruit reparative cell types (Duplantier & van Hoek, 2013; Nijnik & Hancock, 2009).

LL-37 exerts antimicrobial effects by actively interacting with membranes, leading to pore formation and membrane permeabilization. It also neutralizes microbial ligands that trigger harmful inflammatory responses (Xhindoli et al., 2016). Structural and biophysical studies have further defined LL-37 interactions with lipid bilayers and revealed conformational properties that underlie its broad-spectrum activity. These properties permit LL-37 to kill both extracellular and intracellular Staphylococcus aureus in laboratory models and to act against biofilm embedded bacteria that are less susceptible to conventional antibiotics (Overhage et al., 2008).

Beyond microbicidal activity, LL-37 promotes key processes of wound repair including re epithelialization, granulation tissue formation and angiogenesis. Experimental studies in mammalian tissues show that LL-37 stimulates endothelial proliferation through formyl peptide receptor like 1 mediated signaling, enhances vascular growth in animal wound models and that loss of the cathelicidin ortholog impairs wound vascularization and closure (Koczulla et al., 2003). Human tissue studies also show reduced LL-37 in chronic non healing ulcers consistent with a physiologic role in repair (Heilborn et al., 2003). These combined antimicrobial and tissue protective activities make LL-37 an especially attractive template for therapies aimed at infected wounds and soft tissue infections.

Despite its therapeutic promise, LL-37 faces important limitations for clinical translation. Pathogens including Staphylococcus aureus secrete proteases that cleave and inactivate LL-37 and other host peptides, effectively neutralizing their antimicrobial functions in some infection niches. Proteolytic degradation by staphylococcal proteinases such as aureolysin and V8 protease produces cleavage fragments and can abolish bactericidal activity unless the active domain is preserved or stability is enhanced (Sieprawska-Lupa et al., 2004). Host and bacterial proteases found in wound fluid further reduce peptide half-life and bioavailability (Grönberg et al., 2011). In addition, high concentrations of native LL-37 can be cytotoxic in some contexts so strategies that balance potency and safety are required (Kai-Larsen, 2008). Additional challenges include high production cost and reduced activity in complex physiological environments (Li et al., 2024).

To overcome these barriers investigators, have designed LL-37 derived peptides and synthetic analogs that preserve antimicrobial and host beneficial functions while improving resistance to proteolysis and lowering cytotoxicity. These modifications aim to improve peptide stability, reduce toxicity, and retain or enhance antimicrobial potency, e.g., through amino-acid substitutions, truncations, cyclization, or fatty-acid conjugation (Julia et al., 2023; Li et al., 2024). Lead candidates include SAAP148 which is derived from the LL-37 sequence and optimized for stability and potency (de Breij et al., 2018). Preclinical studies demonstrate that SAAP148 eradicates persister cells and mature MRSA biofilms in vitro and that a topical SAAP148 formulation eradicated MRSA in mouse skin infection models and in ex vivo human wound tissue. These results illustrate that rationally designed LL-37 derived peptides can outperform the native peptide in challenging infection models and support their candidacy for translational development (Scheper et al., 2021). Preclinical animal models generally report reductions in MRSA burden and improved healing with LL-37-based treatments, though heterogeneity in designs and outcomes limits firm conclusions (Forde et al., 2024).

Preclinical evidence evaluating LL-37 and its derivative peptides against MRSA remains heterogeneous in experimental design, outcome reporting, and comparator selection, and a focused quantitative synthesis is lacking. Although multiple animal studies have demonstrated antibacterial and wound-healing benefits of LL-37-based therapies, there is no clear consensus regarding the optimal delivery vehicle, and dosage regimens are inconsistently reported across studies, limiting cross-study comparability and translational interpretation. There is therefore a clear need to systematically compile and evaluate animal studies testing LL-37 and its derivative peptides in MRSA infection models to characterize effect sizes for bacterial clearance, wound healing, and survival, and to identify factors influencing efficacy and safety. Addressing these questions through a systematic review and meta-analysis of controlled animal studies will help prioritize lead molecules, refine dosing and delivery strategies, and support rational decision-making for early-phase clinical evaluation.

This systematic review evaluates whether LL-37 and its derivatives improve bacterial clearance, enhance wound healing, and increase survival compared to placebo or standard treatment in animal models of MRSA infection, and aims to identify gaps that must be addressed to advance clinical translation.

2. Methods

2.1. Study Design and Registration

This systematic review and meta-analysis were conducted in accordance with the PRISMA 2020 guidelines (Page et al., 2021). The review protocol was prospectively registered in the PROSPERO database with ID number CRD420251129262. Eligibility criteria and search strategies were structured using the Population-Intervention-Comparator-Outcome-Study design (PICOS) framework, with the study design restricted to preclinical in vivo animal models of MRSA infection to ensure relevance, transparency, and reproducibility.

2.2. Protocol

A Population-Intervention-Comparator-Outcome-Study design (PICOS) model was defined to guide data extraction and ensure consistency across included studies.

2.3. Search Strategy

The literature search was last updated on April 21, 2026. Advanced search features were used to systematically query the PubMed, Scopus, and Embase databases. The search strategy combined terms related to LL-37 ("LL-37," "LL37," "human cathelicidin," and "antimicrobial peptide") with terms for methicillin-resistant Staphylococcus aureus ("methicillin-resistant Staphylococcus aureus," "MRSA") and animal studies ("mouse model," "in vivo," and "preclinical"). To ensure contemporary methodological relevance, the search was limited to studies published between January 1, 2000, and April 21, 2026. The complete Boolean search strategy and database-specific query strings are provided in Supplementary Appendix A.

2.4. Data Collection and Analysis

2.4.1. Selection of Studies

Study selection followed PRISMA 2025 guidelines. Titles/abstracts and full texts were independently screened in duplicate by three reviewers; disagreements were resolved by consensus or by a fourth reviewer.

2.4.2. Data Collection and Management

Two reviewers independently extracted data using a standardized Microsoft Excel form; disagreements were resolved by a third reviewer through re-examination of the study. Extracted variables included:

• Study identifiers: authors, year, title, abstract
• Animal model details: species, strain, sex, age/weight
• MRSA infection model: strain, inoculum, infection site
• Experimental design: number of groups, sample size
• Intervention: LL-37 or derivative sequence, dose, formulation, route, treatment duration
• Comparator details: placebo, untreated control, or standard antibiotic (name, dose)
• Outcomes: primary (quantitative bacterial counts: mean ± SD/SEM, CFU per tissue), secondary (wound healing metrics, survival/mortality)

When outcome data were only presented in graphical form, means and standard deviations were extracted using a digital data extraction tool called WebPlotDigitizer (Rohatgi, 2024). For studies with multiple time points, data from the point of maximal reported effect (typically final CFU counts) were extracted. Four studies (including one with three independent experiments) contained sufficient data for inclusion in the meta-analysis.

2.4.3. Assessment of Risk of Bias (ROB) in Included Studies

Two independent reviewers assessed the risk of bias using the SYRCLE Risk of Bias (ROB) tool, which is specifically designed for animal intervention studies (Hooijmans et al., 2014). This tool consists of ten questions that evaluate various types of bias, including selection, performance, detection, attrition, and reporting biases. Each item was classified as having a “low,” “high,” or “unclear” risk of bias. Any disagreements in ratings were resolved through consensus between the reviewers. The risk of bias ratings informed the interpretation of the findings but were not used as exclusion criteria (Hooijmans et al., 2014).

2.4.4. Measures of Treatment Effect

Studies with at least three animals per group and an untreated or placebo control were considered eligible for quantitative synthesis. For bacterial clearance (continuous outcome), colony-forming unit (CFU) data were extracted directly or converted from log-transformed values here necessary (Rohatgi, 2024). Where only graphical results were provided, numerical data were extracted using digital extraction tools. Comparisons between LL-37 (or derivative peptide) groups and controls were made by calculating Hedges’ g, with standardized mean differences (SMDs) used as the effect measure to account for varying measurement scales and to correct for small-sample bias (<20 animals per group) using the standard correction factor (Vesterinen et al., 2014). Hedges’ g was selected over Cohen’s d because Cohen’s d tends to overestimate effect sizes in studies with small sample sizes, a characteristic of the included animal experiments. For wound healing (binary outcome), treatment effects were expressed as pooled risk ratios (RR) with corresponding 95% confidence intervals (CI). Similarly, survival outcomes (binary outcome) were analyzed using pooled risk ratios with 95% CIs. All analyses employed a random-effects model to account for between-study variability. Heterogeneity was assessed using the I² statistic, with values <60% considered indicative of low-to-moderate heterogeneity and >70% as evidence of substantial heterogeneity (Higgins, et al., 2009). Forest plots were generated using RevMan version 7.2 (Review Manager, The Nordic Cochrane Centre, The Cochrane Collaboration). Statistical procedures followed established guidance for meta-analysis of data from animal studies by Vesterinen et al (2014).

3. Results

3.1. PRISMA Summary of Results

Figure 1 presents the PRISMA flow of the search process. A total of 1,366 records were retrieved from three databases. After removal of 368 duplicates through EndNote, 998 records remained for screening. Title and abstract screening excluded 934 studies, leaving 64 articles for full-text review. Of these, 48 were excluded for not meeting the eligibility criteria, resulting in 16 studies that were included in the systematic review.

3.2. Characteristics of Included Studies

Following a database search and the steps from the PRISMA framework, a number of studies were selected based on specific criteria using a PICOS model (Table 1). Table 2 shows the characteristics of included studies. 16 studies were screened for changes in bacterial counts or related outcomes in MRSA infections treated with LL-37 or its derivatives in animal models, where 12 different peptides were identified (including LL-37 itself and various derivatives such as SAAP-148, 17BIPHE2, and TAT-KR-12). Eight studies examined LL-37 directly (often in combination with other agents like vancomycin), while the remaining eight focused on LL-37 derivatives, with each derivative typically investigated in one or two studies. All studies demonstrated reduced bacterial burden, improved wound healing, or modulated immune responses compared to their respective controls (e.g., no treatment, saline, PBS, or vancomycin), except for one study (Hou et al., 2013) that primarily assessed immune modulation rather than direct bacterial clearance, and another (Pence et al., 2015) that evaluated antimicrobial peptide susceptibility indirectly via bacterial mutants. 14 out of 16 studies used mouse models (predominantly BALB/c or C57BL/6 strains), one used Wistar rats, and one incorporated diabetic minipigs alongside mice. The majority of the studies used wound or skin infection models in vivo, with MRSA strains such as USA300 LAC or ATCC 43300 commonly employed across the research.

3.3. Examples of Excluded Studies

Table 3 shows some examples of excluded studies from the full-text screening stage where studies contained most of the inclusion criteria but did not test for specific criteria. The table shows the study titles and the reasons for exclusion for studies that made it to the full-text eligibility stage of the PRISMA framework. For example, the study by Skerlavaj et al. appeared to meet the criteria set out for this study, but the evaluation of SMAP-29, a sheep-derived cathelicidin peptide not derived from human LL-37, through in vitro assays without preclinical in vivo MRSA infection models does not align with the intervention focus.

3.4. Risk of Bias

The ROB tool from SYCRYL32 was used to determine the risk of bias (ROB) in animal treatment within each study, where Yes, No, and Unclear were appropriate responses and can be seen in Supplementary Appendix B. The majority of the responses in the table are “U”, which represent unclear. Seven studies had animals housed randomly, and three studies had an allocation sequence where the responses for these categories were “Y” representing Yes as an answer and low bias for the specific question.

3.5. Effects of Intervention and Meta-Analysis

The quantitative synthesis evaluated the efficacy of LL-37 and its derivative peptides against MRSA, focusing on two major outcomes: bacterial clearance and wound healing. Data were aggregated from four eligible preclinical studies contributing six independent experimental comparisons, involving a total of 44 animals across intervention and control groups.

The pooled effect for bacterial clearance, measured as standardized mean difference (SMD, Hedges’ g), revealed a large and statistically significant improvement favoring LL-37 treatment over controls. Using a random-effects model, the pooled SMD was -7.86 (95% CI: -9.40 to -6.33; p < 0.001), indicating a strong reduction in bacterial burden within treated groups. As illustrated in the forest plot (Figure 2), all six experimental comparisons, including those from Oriana et al. (2021), Shicheng et al. (2020), Luogen et al. (2022), and Fahimirad et al. (2021), showed effects consistently favoring LL-37, with none of their confidence intervals crossing the line of no effect. This uniform direction and magnitude of outcomes demonstrate robust and reproducible antibacterial efficacy across experimental designs.

Heterogeneity assessment showed remarkable consistency among studies. The I² statistic was 0%, and τ² < 0.0001, indicating that virtually all variability in effect sizes was attributable to sampling error rather than true between-study differences. The test for heterogeneity was non-significant (p = 0.645). The 95% prediction interval (-9.74 to -5.99) further suggested that similar large and negative effects would be expected in future comparable studies, emphasizing the reproducibility of LL-37’s antibacterial activity.

For the secondary outcome of wound healing, the pooled analysis demonstrated that LL-37 substantially improved the rate of complete wound closure relative to controls. The combined risk ratio (RR) was 1.25 (95% CI: 1.16-1.36), corresponding to complete healing in approximately 89.3% of LL-37-treated wounds compared to 70.0% in controls. These results indicate a clinically meaningful enhancement of tissue repair following peptide treatment. Moderate heterogeneity was observed (I² = 45%, p = 0.12), which may reflect differences in experimental protocols such as wound models (skin versus diabetic wounds) and healing durations (10-15 days). Despite these variations, the pooled estimate consistently supports the beneficial role of LL-37 in promoting wound closure and bacterial eradication in MRSA-infected models.

4. Discussion

Taken together, the body of preclinical evidence indicates that LL-37 and its derivatives exert a coherent antimicrobial effect against MRSA in diverse in vivo models and concurrently promote tissue repair processes (see Figure 2; Table 2). This pattern is biologically plausible: peptides combine direct microbicidal activity with modulation of host responses that accelerate wound resolution, as exemplified by topical and carrier-enhanced interventions (Oriana et al., 2021; Luogen et al., 2022). While the consistency of direction across pooled and non-pooled studies strengthens confidence in a true therapeutic signal, interpretation must remain cautious because methodological variability across experiments, in formulation, model choice and outcome measurement, limits immediate generalizability to clinical settings.

The consistent bacterial-clearance signal observed across studies underscores the robustness of LL-37 and its derivatives’ antibacterial efficacy, suggesting that their effects are biologically reproducible across diverse MRSA infection models (Figure 2; Table 2). Studies such as Oriana et al. (2021) and Luogen et al. (2022) show convergent outcomes despite differing peptide formulations and delivery routes, indicating that LL-37’s intrinsic membrane-targeting and anti-biofilm mechanisms underlie this stability. These mechanisms, coupled with improved peptide stability and host-modulatory actions, plausibly explain the sustained antimicrobial performance observed in preclinical models (Talapko et al., 2022).

Improved wound healing observed across included studies likely reflects a synergistic mechanism integrating both antimicrobial and host-modulatory effects of LL-37 and its derivatives. By rapidly reducing bacterial burden and preventing biofilm establishment, these peptides create a permissive environment for tissue regeneration, as evidenced by enhanced epithelialization, collagen organization, and vascularization reported in multiple experiments (Luogen et al., 2022; Nakagami et al., 2012). Beyond microbial clearance, LL-37 analogs appear to orchestrate immune responses that balance inflammation and promote angiogenesis through increased MCP-1, IL-17A, and CD31/vWF expression (Yang et al., 2020; (Zhang et al., 2021). Such pleiotropic effects align with established mechanisms for cationic host defense peptides, which not only disrupt bacterial membranes but also stimulate keratinocyte migration, endothelial proliferation, and granulation tissue formation (Mangoni, 2011). Collectively, these converging antibacterial and pro-repair activities underscore LL-37’s translational potential as a dual-function therapeutic for infection-associated wound healing, particularly in settings where microbial resistance impairs conventional antibiotic efficacy (see Figure 2; Table 2).

Delivery route and formulation markedly shape LL-37’s in vivo impact. For instance, Oriana et al. (2021) found that a combination of topical and systemic LL-37 yielded far greater MRSA clearance and wound repair (re-epithelialization, collagen organization, angiogenesis) than either route alone (Oriana et al., 2021). Likewise, encapsulation in advanced carriers can extend peptide dwell time and target release. Luogen et al. (2022) used a MIL-101 metal-organic framework to carry LL-37 and covalently attached vancomycin, achieving H₂O₂-triggered local release, NIR imaging of infection sites, and synergistic MRSA killing in mice (Luogen et al., 2022). In general, hydrogels or nanogels can load LL-37 and release it gradually: for example, crosslinked HA- or alginate-based matrices retain high water content and can be engineered to respond to pH or temperature (Cesaro et al., 2023; Nguyen et al., 2023). Indeed, microgel encapsulation of LL-37 preserved its activity while drastically reducing hemolysis (Cesaro et al., 2023; Nguyen et al., 2023). Beyond single-agent delivery, peptide-antibiotic co-therapies further improve outcomes. In an implant infection model, Riool et al. (2017) showed that LL-37-derived peptides in a polymer-lipid (PLEX) coating continuously released peptide for 30 days, yielding ≥3.5-log reductions in implant S. aureus and 1-1.5-log reductions in adjacent tissue (in contrast to ~2-log reduction by peptide injection alone) (Riool et al., 2017; Cesaro et al., 2023). These strategies have clear translational promise: AMP-loaded gels or coatings can be applied to wounds or implants without systemic toxicity. In fact, several AMP formulations (e.g. a vancomycin gel) have advanced into clinical trials precisely because sustained local delivery overcomes the instability of free peptide (Cesaro et al., 2023; Zheng et al., 2025).

Across these studies, methodological heterogeneity dominates: small rodent samples and divergent infection models account for most variation rather than LL-37’s underlying activity. For example, Oriana et al. (2021) used adult male BALB/c mice with MRSA-infected cutaneous wounds, comparing topical and systemic LL-37 treatments and measuring both histological indices (VEGF, angiogenesis) and tissue bacterial CFUs (Oriana et al., 2021). Other groups employed implanted catheter-biofilm, osteomyelitis, or pneumonia models, varying inoculum size and pathogen strain. Follow-up ranged from ~1 to >15 days, and endpoints spanned log-CFU reduction, wound closure rates, blinded image scoring or histology. Reporting was inconsistent (often lacking variance estimates or using non-standard units), which further inflates apparent heterogeneity. Importantly, our meta-analysis found essentially no heterogeneity in bacterial-clearance outcomes but only moderate heterogeneity in wound-healing endpoints. This contrast suggests that LL-37’s antimicrobial effect is robust across models, with observed variation in healing metrics reflecting methodological differences. As Sena et al. (2014) emphasize, heterogeneity in preclinical meta-analyses often reflects “real differences in experimental design”. Vesterinen et al. (2014) likewise note that small animal studies tend to be inherently heterogeneous (Vesterinen et al., 2014). In sum, these design factors limit direct cross-study comparison, yet the consistent efficacy signal of LL-37 is robustly maintained once methodological disparities are accounted for.

Despite consistent antibacterial efficacy across models (Figure 2, Table 2), several safety and pharmacokinetic concerns remain for LL-37 derivatives. Notably, host toxicity emerges at high doses: LL-37 analogs induce dose-dependent cytotoxicity and hemolysis of mammalian cells (Voronko et al., 2025), and in vivo high-dose regimens have shown adverse outcomes. Native LL-37 is particularly vulnerable to proteolytic degradation, yielding a short plasma half-life (Voronko et al., 2025). To address this, stabilized analogs have been engineered: for example, the D-amino acid-based 17BIPHE2 (derived from LL-37) resists chymotrypsin and other proteases (Wang et al., 2014), greatly enhancing stability without losing potency. Likewise, SAAP-148 achieved complete eradication of skin infections at high topical doses (up to 1% ointment, ~2000 mg/kg) without skin irritation or systemic toxicity (de Breij et al., 2018). These findings illustrate that analog design (D-form or cyclized peptides) can improve in vivo stability, but also underscore that the therapeutic window is narrow. Rigorous preclinical dose-escalation and PK/PD studies are imperative to delineate safe-efficacy margins. Future work should focus on optimizing half-life (e.g. D-amino or backbone modifications (Wang et al., 2014)) and conducting formal safety and pharmacokinetic profiling. In summary, the robust efficacy of LL-37 derivatives (Figure 2, Table 2) is tempered by proteolytic vulnerability and toxicity at high doses, making stability and safety critical gating factors for translation.

Most antimicrobial tests of LL-37 derivatives have targeted Staphylococcus aureus (often MRSA) isolates. For instance, Oriana et al. (2021) evaluated LL-37 in a mouse model of MRSA-infected wounds, and many other studies likewise use USA300 or ATCC MRSA strains. Conversely, only isolated comparisons have included E. coli, P. aeruginosa or Candida spp.; for example, a recent LL-37 nanoparticle formulation showed activity against E. coli as well as S. aureus (Ergün et al., 2025). This narrow taxonomic focus increases relevance to staphylococcal skin/wound and implant infections but limits generalizability to polymicrobial or Gram-negative contexts. Mechanistically, LL-37’s membrane-disrupting and immunomodulatory actions are thought to constrain resistance evolution (Ridyard & Overhage, 2021; Yang et al., 2025). Nonetheless, adaptive tolerance has been reported: bacteria can secrete proteases that degrade LL-37 or alter membrane charge/composition (Martynowycz et al., 2019). For example, PhoPQ-mediated lipid A modifications in Salmonella double the MIC of LL-37 (Martynowycz et al., 2019). These considerations underscore the need for systematic evaluation of LL-37 analogues against multidrug-resistant Gram-negative bacteria and fungi to define their true clinical applicability.

In our systematic review, methodological deficiencies were widespread in LL-37 peptide studies (SYRCLE assessments were predominantly “unclear” for randomization, allocation concealment and blinding). Only seven studies documented random housing and just three described group allocation; virtually none reported blinding. For example, no study described allocation concealment (leaving group comparability uncertain), and none provided a priori sample-size calculations (suggesting most experiments were underpowered). For context, a recent audit found randomization in only 0-63% of in vivo studies and blinding in just 11-71% (Steele et al., 2025). Consequently, many outcomes lacked reported variance, hampering quantitative synthesis and reducing confidence in pooled estimates. This mirrors broader trends: absence of bias controls and low statistical power inflate efficacy estimates (Romantsik et al., 2024). Sena et al. (2014) likewise noted that animal studies often “lack methodological rigor,” leading to inflated effects; indeed, meta-analyses repeatedly show that omission of randomization/blinding overestimates treatment effects (Vesterinen et al., 2014). To address these shortcomings, we echo calls for preregistered protocols, ARRIVE 2.0 adherence, data sharing, power calculations, and clear reporting of randomization, allocation, and blinding (Percie et al., 2020). Enhanced rigor and transparency are essential to reduce waste. Without these improvements, pooled efficacy estimates must be interpreted with caution and their generalizability to clinical settings is limited.

To advance LL-37 toward clinical use, future research must focus on rigorous translational studies. Firstly, detailed pharmacokinetic/pharmacodynamic (PK/PD) profiling and formal toxicity/dose-finding studies are essential (Negus & Banks, 2016). For example, Zhang et al. (2021) conducted graded dose-toxicity assays of LL-37 derivatives in mice, highlighting how systematic dose-ranging can identify effective yet safe regimens. Stabilization strategies should be pursued: incorporation of D-amino acids (as in Zhang’s KR12 derivative) and peptide encapsulation in nanoparticles or polymers can markedly extend in vivo half-life. In fact, LL-37 nanogel encapsulation significantly increased pulmonary retention and reduced off-target clearance (Kłodzińska et al., 2025), illustrating one such approach. Translational models also need to scale up: promising candidates should be evaluated in clinically relevant disease models and larger animals (e.g. diabetic wound or osteomyelitis models as explored by Nakagami and Machado) before human trials. Combination or carrier-enhanced therapies may further improve outcomes, for instance, metal-organic framework carriers co-loaded with LL-37 and antibiotics have shown synergistic antibacterial effects. Standardized efficacy endpoints (with typical 10-14-day follow-up) are critical to enable cross-study comparisons. Importantly, pathogen testing should expand beyond MRSA to include Gram-negative bacteria and fungi. A systematic, harmonized approach on these priorities will generate robust data and facilitate translation of LL-37 peptides into clinical antimicrobials (Zhang et al., 2021; Kłodzińska et al., 2025).

In summary, LL-37 and its derivatives demonstrate consistent antimicrobial and tissue-reparative efficacy across preclinical models, particularly against MRSA. However, unresolved issues regarding safety, stability, and pharmacokinetic behavior necessitate targeted preclinical optimization before any clinical translation can be responsibly pursued.

5. Conclusions

LL-37 and its derivatives demonstrate strong antimicrobial and tissue-reparative efficacy in preclinical MRSA and implant infection models, underscoring their therapeutic potential. However, translational readiness remains limited by uncertainties in safety, stability, pharmacokinetics, and methodological rigor. Focused preclinical studies addressing these gaps, through standardized protocols, dose optimization, and advanced model validation, are essential prerequisites before clinical evaluation can be responsibly pursued.