Formulation-Driven Innovation in Antifungal Therapy: From Nanotechnology to AI-Assisted Design

Abstract

Invasive fungal infections are a major global health threat, contributing substantially to morbidity and mortality, particularly in immunocompromised populations. Despite the availability of four major antifungal drug classes—azoles, polyenes, echinocandins, and flucytosine—current therapies remain constrained by dose-limiting toxicity, suboptimal pharmacokinetics, and the emergence of multidrug resistance. This review critically evaluates formulation-driven strategies to optimize the therapeutic performance of existing antifungal drugs. It discusses the biological, biopharmaceutical, and clinical barriers that limit treatment outcomes and the formulation approaches developed to overcome these challenges. Particular emphasis is placed on nanotechnology-based drug delivery systems, biofilm-targeting strategies, stimuli-responsive platforms, and combination therapies. The review further highlights the emerging role of artificial intelligence (AI)-driven formulation design, predictive modeling, and precision therapeutics in advancing antifungal treatment. The integration of nanotechnology, AI-driven computational modeling, and precision therapeutics offers new opportunities to improve efficacy, reduce toxicity, and enable more personalized antifungal treatment. Together, these advances position formulation science as a key driver of innovation in next generation of antifungal therapy.

Introduction

Invasive fungal infections (IFIs) are responsible for more than 3.8 million deaths annually, underscoring their substantial and escalating health burden.¹ The importance of fungal diseases in public health was highlighted by the World Health Organization’s publication of the fungal priority pathogen list in 2022.² The rising incidence has been driven by an increasing number of immunocompromised patients, the overuse of broad-spectrum antibiotics and environmental changes such as climate change and urbanization.^3,4

Despite this growing burden, the clinical use of antifungal pharmacotherapy remains limited to four main drug classes: azoles, polyenes, echinocandins, and pyrimidine analogues. These classes are associated with substantial drawbacks, including toxicity, poor pharmacokinetic profiles, and narrow activity spectra.⁵ The impact of these limitations is affected by the emergence of antifungal resistance. For instance, Candida auris isolates have demonstrated fluconazole resistance in 93% of cases and amphotericin B resistance in 35%.⁶ This challenge is further compounded by the eukaryotic nature of fungal cells, which share many biochemical pathways with human cells, making the development of selective agents more challenging.⁷

In response to these challenges, drug formulation science has emerged as a transformative approach to revitalize existing antifungal drugs. Rather than serving as passive delivery vehicles, new formulations are designed to overcome biopharmaceutical barriers by improving solubility, permeability, and target selectivity.⁸ In particular, nanotechnology-based drug delivery systems have shown considerable potential to enhance the bioavailability of poorly soluble drugs and to address resistance mechanisms.⁹ Recent advancements in stimuli-responsive nanomaterials enable smart drug delivery in response to infection-specific cues, thereby maximizing efficacy while minimizing systemic exposure.¹⁰ Additionally, sustained-release systems, hydrogels, and bioadhesive formulations enhance local drug retention and maintain therapeutic concentrations at infection sites.¹¹ These innovations highlight the critical role of formulation science in improving the performance of antifungal drugs and shaping the future of antifungal pharmacotherapy.

This review examines how modern formulation approaches aim to improve the efficacy and safety of antifungal drugs. Specifically, it addresses: (1) the current clinical challenges, including drug resistance, pharmacokinetic limitations, and toxicity; (2) advanced formulation and delivery platforms designed to overcome these barriers, with a focus on targeted, sustained, and enhanced antifungal delivery; and (3) emerging opportunities in artificial intelligence (AI)-driven drug discovery, computational formulation design, and model-informed precision dosing (MIPD) for personalized antifungal therapy. Finally, the review discusses the progress made in these areas and their significance for clinical translation and the development of next-generation antifungal therapy.

Pharmacological and Biopharmaceutical Barriers in Antifungal Therapy

Effective antifungal formulations must overcome multiple biological and pharmacological barriers that limit therapeutic efficacy (Figure 1).

Figure 1.

Pharmacological and Biopharmaceutical Barriers to Effective Antifungal Therapy.

Drug Resistance

The effectiveness of current antifungal agents is increasingly undermined by complex genetic, physiological, and microenvironmental resistance mechanisms. These adaptations present substantial challenges in treatment, limiting the efficacy of traditional dose-escalation or single-target approaches.

Antifungal resistance can be intrinsic or acquired and involves various mechanisms, such as modification of target proteins, decreased intracellular drug concentration, adaptive stress signaling, and enhanced resistance under growth conditions such as biofilm formation.¹² These mechanisms predominantly affect the three major antifungal drug classes: azoles, echinocandins, and polyenes, each with distinct yet overlapping resistance mechanisms. For example, azoles face widespread resistance, particularly in Candida and Aspergillus species. This resistance is driven by qualitative and quantitative alterations of the target enzyme, lanosterol 14α-demethylase (Erg11/Cyp51). Clustered point mutations near the heme-binding domain weaken azole binding affinity. Additionally, transcriptional upregulation of ERG11, often mediated by gain-of-function mutations in the transcription factor Upc2, necessitates higher intracellular drug concentrations for effective inhibition.^13–15 Genomic plasticity further enhances resistance; for example, Candida albicans can form isochromosomes, and Cryptococcus neoformans exhibits chromosome duplication-mediated heteroresistance, which reversibly increases expression of ERG11 and efflux transporter genes.¹⁶ These dynamic adaptations highlight the importance of formulations that maintain effective intracellular drug levels despite varying genetic backgrounds.

Beyond target modification, efflux-mediated resistance is a widespread pharmacological defense across fungal pathogens. Efflux pumps are membrane proteins that confer drug resistance by actively exporting antifungal agents out of the cell, reducing their intracellular concentration below lethal levels. The two main types of efflux pumps are ATP-binding cassette (ABC) transporters, which use ATP hydrolysis to facilitate drug export, and major facilitator superfamily (MFS) transporters, which couple drug efflux with the inward flow of protons down their electrochemical gradient across the plasma membrane.¹⁷ The broad substrate specificity of these efflux systems allows cross-resistance to structurally diverse azoles, underscoring the need for strategies that either bypass transporter recognition or enable co-delivery with chemosensitizing agents.¹⁸

Polyene resistance, although less prevalent, arises from the ability of fungi to remodel their membrane composition. Loss-of-function mutations in ergosterol biosynthesis genes such as ERG3 or ERG6 reduce ergosterol content and impair polyene binding, thereby decreasing drug susceptibility.¹⁹ These biological constraints, combined with the intrinsic toxicity and poor aqueous solubility of polyene compounds, further reinforce the need for formulation strategies that improve drug solubility and target selectivity. At the molecular level, global stress response regulators such as Hsp90 and calcineurin stabilize signaling networks that are essential during stress induced by antifungals.^20,21 These pathways help buffer the deleterious effects of resistance-conferring mutations and facilitate rapid adaptation, further complicating the pharmacological control of infections. Additionally, biofilm growth represents a unique and significant pathophysiological barrier. Biofilms formed by Candida, Fusarium, and other fungi are surrounded by a dynamic extracellular matrix that entraps antifungal drugs, limits their penetration, and protects metabolically dormant persister cells.^22–24 The multifactorial nature of biofilm resistance, including matrix-mediated sequestration, efflux activation, and stress signaling, renders conventional antifungal therapy largely ineffective.

Collectively, these resistance mechanisms suggest that antifungal resistance typically does not result from a single target alteration, but rather from the complex interplay of multiple biological defense pathways acting in concert. This complexity offers a rationale for developing advanced formulations. Approaches such as nanocarriers, co-formulation systems, and biofilm-penetrating technologies can enhance intracellular drug delivery, bypass efflux pumps, disrupt stress responses, and ultimately restore antifungal efficacy.

Suboptimal Pharmacokinetics

Many antifungal agents exhibit poor pharmacokinetic profiles that undermine their clinical potential. One of the major challenges is poor aqueous solubility, particularly among polyenes and azoles, which leads to erratic or incomplete gastrointestinal absorption.^25,26 As a result, inadequate systemic exposure can lead to subtherapeutic drug concentrations at the site of infection, potentially causing treatment failure. For example, the oral absorption of itraconazole and posaconazole is substantially influenced by gastric pH, food intake, and gastrointestinal integrity, leading to variability in plasma concentrations.^27,28 Permeability is another key factor affecting the absorption and bioavailability of antifungal drugs. The ability of a molecule to permeate a lipophilic membrane is determined by several factors, including its structure, molecular weight, and physicochemical properties.²⁹ Antifungal agents with high molecular weight, such as amphotericin B, exhibit poor intestinal permeability, resulting in low oral bioavailability. This limitation poses a substantial challenge for the development of effective oral formulations of amphotericin B.³⁰

Following absorption, distribution is frequently restricted by extensive plasma protein binding. Many azoles and echinocandins exhibit protein binding exceeding 90%, reducing the free drug fraction available for pharmacological activity.³¹ Limited tissue penetration further complicates antifungal therapy. The blood-brain barrier (BBB) limits the entry of large molecules, such as echinocandins and amphotericin B, into the central nervous system (CNS), complicating the management of fungal meningitis.³² Additionally, drug delivery into dense fungal biofilms or poorly vascularized tissues, such as the nail bed in onychomycosis, remains a challenge, often requiring long-term treatment.⁸

Metabolic variability also contributes to pharmacokinetic complexity, especially among azole antifungals. Genetic polymorphisms in hepatic cytochrome P450 (CYP) enzymes underlie nonlinear pharmacokinetics and marked interindividual variability in systemic drug exposure. Furthermore, many azole antifungals function as both substrates and inhibitors of CYP enzymes, thereby altering the metabolism of co-administered medications and giving rise to clinically significant drug–drug interactions. For example, voriconazole is metabolized mainly by CYP2C19, CYP2C9, and CYP3A4, and it also acts as a potent inhibitor of these enzymes, increasing the risk of drug-drug interactions and necessitating dose adjustment for concomitant medications.³³ Many antifungals have a narrow therapeutic window, where exceeding specific concentrations significantly increases the risk of toxicity.³⁴ In contrast, subtherapeutic exposure is associated with resistance and treatment failure.³⁵ Furthermore, antifungal drugs with short plasma half-life, such as 5-flucytosine (3–4 hours), present challenges in maintaining adequate drug exposure, requiring frequent dosing and potentially compromising patient adherence.³⁶

Collectively, these pharmacokinetic limitations provide a strong justification for the development of novel formulation strategies that improve absorption, enhance tissue distribution, overcome physiological barriers, and maintain effective drug concentrations at infection sites.

Host Toxicity

A major concern in antifungal therapy is achieving selective toxicity against the fungal pathogen without harming the human host. This task is intrinsically challenging, as fungi and mammalian cells share many molecular similarities, substantially limiting pathogen-specific drug targets.³⁷

Amphotericin B, the prototype polyene, remains essential in the management of severe cryptococcosis owing to a broad-spectrum fungicidal activity. Its mechanism of action involves binding to ergosterol and pore formation in fungal membranes.³⁸ However, because fungal ergosterol and cholesterol in mammalian cell membranes share structural similarities, amphotericin B can also bind to cholesterol in host cell membranes, thereby leading to toxic effects in human cells.³⁹ Nephrotoxicity is of particular concern, resulting from renal vasoconstriction, tubular membrane injury, and reduced blood flow, which can lead to elevated serum creatinine and electrolyte disturbances.⁴⁰

Azoles exert their antifungal activity by inhibiting lanosterol 14-α-demethylase, a cytochrome P450-dependent enzyme essential for ergosterol synthesis in fungal cell membranes. However, their incomplete selectivity also leads to the inhibition of mammalian P450 enzymes. In particular, ketoconazole is prone to inhibiting mammalian isoenzymes, disrupting steroid hormone synthesis, and potentially causing endocrine disorders. This inhibition can also alter the metabolism of co-administered drugs, increasing the risk of drug-drug interactions and potential hepatotoxicity.⁴¹ Mitochondrial damage is another mechanism contributing to hepatotoxicity. Both ketoconazole and posaconazole have been shown to decrease the mitochondrial membrane potential and impair the activity of electron transport chain enzyme complexes in liver cells, leading to the accumulation of reactive oxygen species (ROS) and apoptosis.⁴² Furthermore, prolonged exposure to voriconazole has been associated with phototoxicity and an increased risk of cutaneous squamous cell carcinoma.^43,44 Although echinocandins generally exhibit a favorable safety profile, they have been associated with abnormal liver function tests.⁴⁵

Overall, these toxicity concerns, combined with challenges in drug delivery, underscore the gap between antifungal efficacy and clinical outcomes. These challenges highlight the importance of formulation strategies that minimize off-target effects and host toxicity, ultimately enhancing therapeutic efficacy.

Formulation Strategies and Emerging Platforms for Antifungal Drug Delivery

Identifying the key challenges in antifungal therapy highlighted above has spurred considerable research interest in a variety of formulation technologies. The following sections systematically review these strategies based on the primary therapeutic limitations they are designed to address, while recognizing that some approaches address multiple therapeutic limitations (Figure 2).

Figure 2.

Formulation Strategies for Optimizing Antifungal Therapy.

Strategies to Overcome Drug Resistance

Given that drug resistance is driven by multiple mechanisms, various formulation strategies have been developed to combat the resistance of antifungal drugs. Selected examples are summarized in Table 1 and discussed in detail.^46–57

Table 1.

Summary of Recent Formulation Approaches for Overcoming Antifungal Drug Resistance

Approach	Formulation Platform	Payload(s)	Key Outcome(s)	Ref.
Biofilm-penetrating and matrix-disrupting systems
Surface modification with biomatrix degradation enzymes	β-1,3-glucanase-functionalized chitosan NPs	Amphotericin B	Superior efficacy in removing mature biofilms compared with the free drug and non-functionalized NPs by disrupting the biofilm matrix and eliminating cells directly. Eradication of biofilm in clinical isolates	[48]
	Alcalase-coated shellac NPs	Amphotericin B	92% reduction in biofilm mass, 7% reduction in biofilm viability compared with the growth control in C. albicans, and 6-fold reduction in fungal CFU after 180 min	[49]
Multimodal functionalization	Kappa-carrageenan capped-silver NPs	Silver (Ag)	80% biofilm inhibition and eradication, membrane disruption and pore formation in C. albicans and C. glabrata, ROS generation, and inhibition of ECM enzymes	[50]
	Tyrosol-functionalized chitosan gold NPs	Tyrosol	Complete eradication of mature biofilms, downregulation of cell wall biosynthesis genes, ROS generation, ergosterol depletion, and reduction of protein content in ECM	[51]
2Synergistic drug combination
Multimodal synergism	Nanosuspension	Curcumin + fluconazole	Disruption of biofilms and hyphal growth, reduction of efflux activity, downregulation of adhesion-related genes ALS1, ALS3, HWP1, and EFG1	[54]
	PLGA polymeric NPs	Voriconazole + cyclosporine A	4-fold increase in antifungal activity compared with the combination of the two free drugs, and effective inhibition of C. albicans biofilm growth compared with free drug counterparts	[56]
Combination with efflux pump inhibitors	Vaginal hydrogel	Fusidic acid + fluconazole	Synergistic restoration of fluconazole efficacy via efflux pump inhibition in Candida albicans by fusidic acid, downregulation of CDR1, CDR2, and MDR1	[52]
	Vaginal hydrogel	Celastrol + fluconazole	Chemosensitization of azole-resistant isolates, efflux inhibition, and downregulation of efflux genes	[57]

Abbreviations: NPs, Nanoparticles; CFU, Colony-forming unit; ROS, Reactive oxygen species; ECM, Extracellular matrix; PLGA, Poly(lactic-co-glycolic acid).

Biofilm-Penetrating and Matrix-Disrupting Systems

Biofilms are highly organized microbial communities that adhere to surfaces and are encased in a self-generated protective extracellular matrix. Within these communities, microorganisms exhibit a substantial reduction in susceptibility to antimicrobial agents, up to 1000-fold less sensitive than their planktonic counterparts.⁴⁶ Fungal biofilms limit antifungal efficacy through extracellular matrices that restrict drug diffusion and harbor persister cells that can tolerate high drug concentrations.⁴⁷ To overcome these structural barriers and achieve deep biofilm penetration, researchers have engineered nanocarriers that incorporate matrix-degrading enzymes (β-1,3-glucanase, protease) and are tailored to their surface properties, yielding positive outcomes.^48,49 For instance, Tan et al fabricated chitosan nanoparticles loaded with amphotericin B and functionalized with β-1,3-glucanase.⁴⁸ These nanoparticles effectively inhibited planktonic cell growth and biofilm formation, demonstrating superior efficacy in eradicating mature biofilms compared with free drug or non-functionalized nanoparticles. These nanoparticles penetrated the biofilm structures, disassembled the matrix, and targeted embedded cells, effectively removing biofilms from clinical isolates. Similarly, multimode action-modulating nanoparticles, such as tyrosol-functionalized chitosan gold nanoparticles and kappa-carrageenan capped silver nanoparticles (CRG-AgNps), have been shown to substantially eradicate biofilm.^50,51 Gupta et al developed CRG-AgNps that inhibited and eradicated 80% of Candida biofilms at approximately 300 μg/mL.⁵⁰ This effect was attributed to multiple mechanisms, including membrane disruption and pore formation. In addition, biofilm-forming ability was further impeded by the generation of ROS and the inactivation of hydrolytic enzymes.⁵⁰ From a formulation perspective, immobilizing enzymes on nanomaterials is a promising strategy for enhancing their stability against environmental stresses, such as temperature and pH fluctuations.⁵⁸ Accordingly, Li et al⁵⁹ demonstrated that immobilization of alginate lyase (Aly08) on low-molecular-weight chitosan nanoparticles (AL-LMW-CS-NPs) significantly enhanced its thermal stability and reusability.⁵⁹ Nevertheless, long-term enzyme stability remains dependent on the formulation and storage conditions and should be systematically evaluated for individual enzyme-functionalized nanocarriers. Collectively, these findings highlight the importance of disrupting biofilm architecture through advanced drug delivery strategies to combat drug resistance.

Co-Formulation Systems

Expanding beyond conventional monotherapy, several studies have explored the use of non-antifungal drugs as resistance modifying agents. This strategy leverages existing pharmacological compounds to restore susceptibility to antifungal drugs by targeting specific resistance mechanisms, such as efflux pump inhibition and cellular sensitization.^52–57

Previous studies demonstrated that combining fusidic acid with fluconazole can overcome drug resistance in Candida albicans.⁵² Fusidic acid exhibited synergistic activity by inhibiting efflux pumps and downregulating the resistance-associated genes CDR1, CDR2, and MDR1. The combination was formulated as a vaginal Carbopol 940 hydrogel, which substantially reduced fungal burden in a mouse model with minimal histopathological changes in vaginal tissue.⁵² Similarly, a recent study explored the use of amantadine hydrochloride, an antiviral and anti-Parkinson’s drug, as a sensitizer to enhance the effectiveness of azoles against resistant C. albicans. Laboratory tests and a Galleria mellonella infection model demonstrated that the drug combination significantly improved survival rates and reduced fungal tissue invasion. The synergy was attributed to amantadine’s ability to block efflux pumps, inhibit early biofilm formation, and reduce the secretion of extracellular phospholipase, a key virulence factor.⁵³

Zhang et al incorporated curcumin nanosuspension to enhance fluconazole’s activity by downregulating adhesion-related genes, suppressing efflux pumps, and inhibiting hyphal formation.⁵⁴ Furthermore, incorporating essential oils or natural compounds with antifungal activity may restore azole susceptibility. Ahmad et al demonstrated that the essential oil carvacrol showed synergistic activity with fluconazole, enhancing efficacy against 44 of 49 Candida strains tested, including azole-resistant isolates.⁵⁵ Collectively, the findings from these diverse co-formulation approaches highlight their potential to mitigate antifungal drug resistance.

Despite these advantages, dual-drug combinations may still permit the emergence of resistance during prolonged treatment. This risk may be reduced by selecting drug pairs with independent and complementary mechanisms, thereby limiting fungal escape through a single adaptive pathway.⁶⁰ Pharmacokinetic optimization to minimize exposure within the mutant selection window may also help suppress resistant subpopulations.⁶¹ Susceptibility-guided combination selection and longitudinal monitoring of resistance-associated phenotypes or molecular markers may further enable timely adaptation of therapy. Thus, rational combination design, pharmacokinetic optimization, and resistance surveillance are important for preserving the long-term efficacy of dual-drug antifungal strategies.

Strategies to Overcome Pharmacokinetic Limitations

The pharmacokinetic limitations discussed earlier substantially reduce the clinical efficacy of many antifungal agents. Consequently, various formulation strategies have been developed to overcome these issues. Representative methods to overcome major pharmacokinetic challenges are summarized, with relevant studies overviewed in Table 2.^62–84

Table 2.

Formulation Approaches Addressing PK Limitations of Antifungal Drugs

PK Limitation	Approach	Payload(s)	Key Outcome(s)	Ref.
Solubility	Co-crystallization	Ketoconazole	100-fold increase in solubility, 12.79-fold increase in oral bioavailability	[62]
	Inclusion complex	Griseofulvin	477-fold increase in solubility, 90% in vitro drug release in 10 min, 177% increase in relative bioavailability	[67]
	Lyophilized nanosuspension	Luliconazole	Improved solubility by >370, >700, >1250, and >1150-fold at pHs 1.2, 4.6, 6.8, and 7.4, respectively	[69]
	Amorphous solid dispersions	Posaconazole	7.1-fold increase in aqueous solubility compared with physical mixture using Kollidon K30	[82]
	Hybrid aerogel	Clotrimazole	3-fold increase in drug release rate achieved at 180 min	[83]
Permeability	Spanlastics gel	Voriconazole	1.3-fold enhancement in vaginal mucosal permeation compared with the conventional gel	[73]
	Invasomes gel	Luliconazole	2.47-fold higher skin permeation than pure luliconazole gel	[74]
	Microemulsion-based gel	Efinaconazole	Enhanced permeation across the nail plate with no observable lag time	[75]
	Polymeric lipid hybrid nanoparticles (Lipomer)	Itraconazole	1.476-fold increase in intestinal permeability compared with the pure drug solution	[84]
Duration of action	Chitosan/sodium-alginate nanoparticles	Isavuconazole	Sustained drug release over 24 hours following the Higuchi model	[78]
	Sustained−release liposomes	Amphotericin B	Prolonged precorneal residence (4.5 hours vs 1 hour for marketed products), reduction in dosing frequency (from 24 to 6 instillations daily)	[79]
	Hybrid degradable nanofibers	Fluconazole, vancomycin and ceftazidime	Sustained release for approximately 30 days (in vitro) and approximately 56 days (in vivo)	[81]

Solubility

To address solubility issues with antifungal drugs, several approaches have been employed, leading to valuable outcomes. Among these approaches, crystal engineering via co-crystallization has proven effective in enhancing drug solubility. Baldea et al reported a 100-fold increase in solubility for a ketoconazole-fumaric acid cocrystal,⁶² while Yu et al observed a 1800-fold increase using glutaric acid.⁶³ Martin et al demonstrated that a ketoconazole-p-aminobenzoic acid cocrystal resulted in a 10-fold solubility enhancement.⁶⁴ In addition, altering crystalline states through amorphization can effectively improve drug solubility. Dhodi et al showed that converting posaconazole into a co-amorphous system with carboxylic acids significantly increased solubility and in vivo absorption.⁶⁵ In a similar approach, França et al reported that a griseofulvin-tryptophan co-amorphous system improved kinetic solubilization and prevented recrystallization for 12 months.⁶⁶ Ding et al employed cyclodextrin inclusion complexes with 2-hydroxypropyl-γ-cyclodextrin, assisted by supercritical carbon dioxide, achieving a 477-fold increase in water solubility of griseofulvin and 90% drug release in 10 minutes.⁶⁷

Other solubility enhancement strategies such as microemulsions, nanosuspensions, and self-nanoemulsifying drug delivery systems (SNEDDS), have also been employed to improve drug solubilization, release, and antifungal efficacy of various hydrophobic compounds.^68–70

Permeability

One of the major pharmacokinetic restrictions of antifungal treatment is the limited permeability across biological membranes. Formulation approaches that can transiently modify biological barriers or enhance drug permeability are essential for improving drug absorption. To address the challenges associated with oral amphotericin B delivery, considerable efforts have been made to develop nanocarrier systems that enhance intestinal drug permeability. Jain et al developed polymer-lipid hybrid nanoparticles (PLNs) consisting of a lecithin core and a protective gelatin shell to improve the oral bioavailability of amphotericin B.⁷¹ This formulation increased the drug’s permeability in Caco-2 cell models by 5.89-fold and its oral bioavailability in rats by 4.69-fold.⁷¹ In a study by Nimtrakul et al, copolymeric micelles were investigated as a delivery system for amphotericin B.⁷² These amphotericin B-loaded polymeric micelles were prepared using polyvinyl caprolactam-polyvinyl acetate-polyethylene glycol copolymer (Soluplus^®). The micelles protected the drug from acidic degradation in simulated gastric fluid and cellular uptake studies in Caco-2 cells demonstrated a 6-fold increase in drug uptake, driven by enhanced permeability across Caco-2 cell monolayers.

Ultradeformable vesicular systems are also effective for enhancing mucosal and transdermal penetration. For example, voriconazole-loaded spanlastics developed by Sheta et al showed 1.3-fold enhancement in penetration through the vaginal mucosa of rats compared with a conventional voriconazole gel formulation.⁷³ In a similar study, Kumari et al developed a Carbopol 934-based invasomes gel that improved the transdermal delivery of luliconazole by re-organizing the stratum corneum.⁷⁴ The optimized invasome gel showed approximately 2.47-fold higher skin permeation than the pure luliconazole gel. Emulsion-based nanocarriers have also been developed to enhance drug penetration through keratinized barriers. For instance, efinaconazole microemulsions with globule size less than 100 nm exhibited substantially higher transungual drug flux than conventional topical solution formulation, facilitating penetration through the dense keratin network of the nail plate.⁷⁵

Duration of Action

As mentioned earlier, some antifungal agents are rapidly cleared from the body, resulting in short half-lives. These pharmacokinetic properties necessitate frequent administration, which is often associated with poor patient compliance. To overcome these shortcomings, sustained-release drug delivery systems have been developed to maintain therapeutic drug levels for prolonged periods following a single administration.⁷⁶ This approach reduces dosing frequency and improves patient adherence.⁷⁷

Polymeric nanoparticle systems are commonly used for sustained antifungal delivery. In one recent study, Khan et al developed chitosan/sodium alginate nanoparticles loaded with isavuconazole, achieving sustained drug release with 94.58% of the drug released over 24 h.⁷⁸ Skin permeation studies using rat skin showed gradual and controlled drug transport, with enhanced antifungal activity against C. albicans. Moreover, lipid-based vesicular systems have also been applied to enhance local drug exposure. Mishra et al formulated a corneal targeted liposomal amphotericin B for the treatment of fungal keratitis.⁷⁹ The liposomes were prepared by dissolving Phospholipon-90H, stearylamine, and amphotericin B in organic solvents, evaporating the mixture to form a thin lipid film, and then hydrating it with water. This liposomal formulation significantly increased precorneal residence time, with therapeutic drug concentrations persisting for up to 4.5 h, compared with approximately 1 hour for the marketed formulation. The enhanced retention was attributed to the interaction between the positively charged liposomal surface and the negatively charged mucin layer on the cornea, which may help sustain drug release.⁷⁹ Extending this approach to systemic delivery, researchers developed voriconazole sustained-release tablets via wet granulation using natural semi-synthetic polymers. The optimized formulation (1:1 Karaya gum: hydroxypropyl methylcellulose (HPMC) K100M) released 99.82% of the drug within 12 hours, thereby reducing dosage frequency.⁸⁰

For long-term localized therapy, implantable or depot-forming systems can provide extended antifungal release. For example, fluconazole-loaded hybrid PLGA nanofibers developed by Hsu et al showed sustained in vitro release for over 30 days and in vivo for up to 56 days, highlighting their potential for the treatment of polymicrobial osteomyelitis.⁸¹

Strategies to Minimize Toxicity

Site-specific targeting of antifungal agents localizes therapeutic action at disease sites, thereby improving drug efficacy while minimizing drug-related toxicity.⁸⁵ This section reviews selected formulation approaches for organ-specific drug delivery (eg, CNS, pulmonary, ocular, and dermal), as well as additional strategies enabling site-specific drug activation (Table 3).

Table 3.

Formulation Approaches for Organ- and Site-Specific Delivery of Antifungal Drugs

Approach		Formulation Platform	Payload(s)	Key Outcome(s)	Ref.
1. Organ-specific drug delivery
Brain	Magnetic nanocarrier for brain targeting	Magnetic Liposomes	Amphotericin B	2-fold increase in brain drug concentration under a magnetic field, demonstrating enhanced brain targeting with increased localized drug accumulation	[86]
	Nose-to-brain delivery system	Nanostructured Lipid Carriers	Ketoconazole	Enhanced brain delivery via intranasal administration, bypassing the BBB through the olfactory pathway, improved drug accumulation in brain tissue	[87]
Lung	Micellar formulation	Deoxycholate micelles	Amphotericin B	2.8-fold increase in lung drug concentration compared with the commercial liposomal formulation, reduced nephrotoxicity in murine aspergillosis	[88]
	Aerosolizable LNVs (inhalation)	Phosphatidylcholine-based LNVs	Voriconazole	Enhanced cellular uptake in lung tissue with prolonged pulmonary retention	[89]
Eye	Hybrid nanocarrier ocular system	Bioadhesive niosomes incorporated into pH-sensitive in-situ gel	Itraconazole	Enhanced ocular targeting with increased corneal permeation and retention, resulting in improved localized antifungal activity	[90]
	Extended ocular delivery device	Drug-eluting contact lenses	Econazole	Sustained ocular drug delivery via antifungal contact lenses, maintaining fungicidal activity against Candida albicans for up to 3 weeks at the corneal site	[91]
Skin and Nail	Microporation-assisted topical nail delivery	Aqueous gels and thermogels (HPMC/Poloxamer)	Ciclopirox	Increased drug retention within the nail plate and enhanced delivery across the nail, decreased dose frequency to fortnightly administration	[92]
	Microneedle patch	Dissolving microneedle patches with PLGA NPs	Amphotericin B	Enhanced delivery into deep dermal tissues with effective inhibition of fungal cell growth and sustained drug release over 4 days	[93]
2. Fungal cell wall-targeted nanocarriers
Peptide-mediated targeting		Penetratin-decorated liposomes	Posaconazole	More than 80% increase in liposome-fungal cell interaction for C. albicans and C. auris	[94]
Chitin-specific targeting		CFW-PE-coated ethosomes	Voriconazole	Specific chitin-mediated binding to fungal cell walls, enables selective targeting over mammalian cells	[95]
3. Stimuli-responsive systems
pH-responsive intracellular release		pH-sensitive liposomes	Nystatin	Acidic pH-triggered destabilization of liposomes with site-specific intracellular drug release within lysosomes	[96]
Enzyme triggered release		Lipase-responsive PCL nanoparticles	Amphotericin B	Lipase-triggered degradation of PCL enabled on-demand drug release specifically in the presence of fungal lipase, resulting in enhanced antifungal efficacy with reduced cytotoxicity	[97]
4. Prodrug
Stimuli-responsive prodrugs		Telodendrimer micelles using boronate “click” chemistry	Amphotericin B	Targeted drug release in response to acidic pH and ROS in the infected tissue	[98]
		Lipidic prodrug co-crystals	Bifonazole + NSAIDs (eg, Aspirin)	Dual-action synergistic therapy with site-specific release triggered by the acidic and oxidative environment of fungal biofilms	[99]

Abbreviations: BBB, Blood-brain barrier; LNVs, Lipid-Nanovesicles; HPMC, Hydroxypropylmethylcellulose; PLGA, Poly(lactic-co-glycolic acid); NPs, Nanoparticles; CFW-PE, Calcofluor white-phosphatidylethanolamine; PCL, Polycaprolactone; ROS, Reactive oxygen species.

Organ-Specific Drug Delivery

CNS-Targeted Delivery Systems

Effective management of CNS infections remains a major global health challenge. Despite the availability of treatments, poor drug delivery is considered a primary concern.¹⁰⁰ The BBB poses a major challenge to effective pharmacotherapy by limiting the access of a wide range of therapeutic agents to the brain, necessitating the development of new formulations to specifically target the brain.¹⁰¹

One formulation approach to increase target selectivity involves using nanoparticles decorated with ligands that bind to specific receptors expressed on the luminal (blood-facing) membrane of the endothelial cells that form the BBB. The ligand-receptor binding triggers internalization of the receptor-ligand complex into intracellular vesicles, which then undergo endosomal sorting and eventually fuse with the abluminal (brain-facing) membrane to release the drug into the brain parenchyma.¹⁰² For example, Tang et al reported improved brain delivery of amphotericin B with prolonged survival in a murine meningitis model using OX26-modified poly(lactic acid) (PLA)-polyethylene glycol (PEG) nanoparticles.¹⁰³ The PLA-PEG copolymer was first conjugated with the OX26 anti-transferrin receptor antibody, which binds the transferrin receptor abundantly expressed on BBB endothelial cells. The interaction of OX26 with the target receptors triggered receptor-mediated transcytosis, enabling the nanoparticles to cross the BBB and resulting in brain drug concentrations 6.7-fold higher than those of free amphotericin B, while substantially reducing fungal burden.¹⁰³ Similarly, Shao et al demonstrated that itraconazole accumulated more effectively in the CNS and was released into cells upon loading into GLUT1-targeted dehydroascorbic acid (DHA)-functionalized micelles.¹⁰⁴ Despite these benefits, receptor-mediated targeting may also pose off-target safety concerns. Unintended ligand interactions with non-target receptors or expression of the target receptor in peripheral tissues may promote off-target accumulation and cellular uptake. For example, transferrin receptor 1 (TfR1), a widely exploited target for BBB transport, is also expressed in peripheral tissues, and TfR-targeted brain-shuttle biologics have been associated with dose-dependent reticulocyte depletion related to peripheral TfR engagement.¹⁰⁵ Such off-target interactions may increase local exposure to both the nanocarrier and encapsulated antifungal agent, potentially leading to tissue-specific toxicity. Therefore, careful optimization of ligand specificity, affinity, and surface density, together with comprehensive biodistribution and off-target toxicity studies, is essential to minimize these risks and support the safe translation of receptor-targeted CNS nanocarriers.

Borneol- and PEG-modified bovine serum albumin nanoparticles have also been shown to improve itraconazole brain distribution. Additionally, polysorbate-80-functionalized alginate nanoparticles carrying miltefosine have demonstrated enhanced BBB translocation and reduced fungal burden in the brain in cryptococcosis models.^106,107 Zhao et al employed an external magnetic field to improve brain targeting, developing amphotericin B magnetic liposomes that encapsulate superparamagnetic ferroferric oxide (Fe₃O₄) nanoparticles.⁸⁶ The liposomes were fabricated using a film dispersion-ultrasonication method, yielding particles with an approximate size of 240 nm. Their study demonstrated enhanced brain targeting under magnetic field influence, and a substantial reduction in drug toxicity compared with free amphotericin B.⁸⁶

Beyond systemic delivery strategies, nose-to-brain delivery provides a non-invasive method that bypasses the BBB through the olfactory pathway. Du et al indicated that nanostructured lipid carriers (NLCs) loaded with ketoconazole effectively delivered the drug to the brain via the nasal cavity, thereby enhancing antifungal activity and reducing fungal burden in murine models of cryptococcal meningoencephalitis.⁸⁷

Pulmonary Drug Delivery System

Targeting the lungs, the primary entry point for many fungal pathogens, allows for direct delivery of drugs to the site of infection, resulting in high local drug concentrations and minimized systemic exposure and related toxicities.^108,109

Formulation-dependent differences in amphotericin B aggregation state and particle size influence biodistribution. A study using a murine pulmonary aspergillosis model found that a polyaggregated amphotericin B-sodium deoxycholate micellar formulation achieved approximately 2.8-fold higher lung drug concentrations than liposomal amphotericin B.⁸⁸ The study also reported lower kidney exposure, suggesting enhanced pulmonary targeting and reduced nephrotoxicity. Similarly, lipid-based nanocarriers using lung-endogenous phospholipids can enhance pulmonary drug delivery by improving compatibility with pulmonary surfactants. For instance, Kaur et al developed lipid nanovesicles loaded with voriconazole containing dipalmitoylphosphatidylcholine (DPPC).⁸⁹ Following nebulization, these nanovesicles demonstrated prolonged lung residence and approximately 4-fold higher pulmonary exposure than free drug, with only modest increases in systemic drug levels. Alternatively, polymeric nanoparticles offer complementary advantages for pulmonary antifungal delivery. Paul et al demonstrated that chitosan-coated PLGA nanoparticles exhibited mucoadhesion-mediated retention in the lungs, resulting in sustained drug release and improved bioavailability of voriconazole following inhalation.¹¹⁰ Additionally, inhaled PC945 (opelconazole) has shown promising results in localized antifungal therapy. It achieves high lung concentrations with prolonged pulmonary retention, low systemic exposure at steady state, and minimal risk of drug-drug interactions.¹¹¹ These pulmonary formulation strategies highlight the potential of site-specific delivery to enhance antifungal efficacy while substantially improving safety.

Ocular and Corneal Delivery Systems

Ocular fungal infections can lead to visual impairment or blindness and are increasingly prevalent. The limited bioavailability of topical antifungal formulations has led to growing interest in novel drug delivery systems to improve ocular retention, corneal permeation, and therapeutic efficacy.¹¹²

Several studies have focused on developing solutions to address these challenges. For example, natamycin-loaded solid lipid nanoparticles (SLNs) with a mean particle size of approximately 42 nm were developed and exhibited a positive zeta potential due to the use of stearyl amine as a charge modifier. This formulation demonstrated improved corneal permeation and antifungal activity compared with the unformulated natamycin.¹¹³ Similarly, voriconazole-loaded zein-pectin-hyaluronic acid nanoparticles (ZPHA-VRC NPs) exhibited desirable physicochemical properties and antifungal activity comparable to free voriconazole. In vivo studies using Galleria mellonella and Caenorhabditis elegans demonstrated reduced systemic toxicity and improved survival, indicating the potential of these nanoparticles as a safer antifungal delivery system for voriconazole.¹¹⁴ In a recent study, Badran et al developed itraconazole-loaded bioadhesive niosomes coated with chitosan and hyaluronic acid, incorporated into a pH-sensitive in situ gel.⁹⁰ This formulation showed enhanced ex vivo corneal permeability and antifungal action compared with the control formulation. Additionally, therapeutic contact lenses have been explored as a non-invasive platform for prolonged antifungal delivery. Contact lenses containing econazole-impregnated poly(lactic-co-glycolic) acid (PLGA) films encapsulated within a poly(hydroxyethyl methacrylate) (pHEMA) matrix were fabricated and demonstrated sustained antifungal activity for up to 21 days.⁹¹

Dermal and Nail Delivery Systems

Topical antifungal therapy for dermatophyte infections and onychomycosis requires formulations capable of penetrating the stratum corneum or nail plate, both of which act as barriers that limit drug permeation.^115,116

Recent advancements in transungual delivery have employed a Quality-by-Design (QbD) strategy to optimize econazole nitrate nail lacquers. Puri et al developed Eudragit RSPO-based nail lacquers (10% w/w) containing Kolliphor CS 20 as a permeation enhancer.¹¹⁷ In vitro drug release and ex vivo nail permeation studies demonstrated faster and more sustained drug delivery compared with control lacquers without the enhancer. In another study, microporation was used to physically bypass the nail barrier. When combined with aqueous ciclopirox gels, this approach significantly enhanced drug delivery in human nails, outperforming commercial lacquers.⁹²

Beyond transungual applications, strategies designed to modify the skin barrier have garnered considerable research interest for addressing cutaneous fungal infections. Cheng et al used transfersomes, known for their inherent deformability and membrane flexibility, to develop a fluconazole-loaded transfersomal gel with enhanced skin permeation and retention. This formulation outperformed both pure fluconazole gel and marketed formulations.¹¹⁸ For deeper cutaneous infections, dissolving microneedle patches loaded with amphotericin B PLGA nanoparticles demonstrated successful penetration of the stratum corneum and drug delivery to deep skin layers.⁹³

Fungal Cell Wall-Targeted Nanocarriers

The fungal cell wall, rich in mannoproteins, β-glucans, and chitin, provides highly specific targets for selective drug delivery.¹¹⁹ Numerous studies have demonstrated that functionalizing nanocarriers with recognition moieties (eg, C-type lectins such as Dectin-2, chitin-binding calcofluor white, or cell-penetrating peptides) can enhance receptor-mediated accumulation at infection sites. This approach has been shown to substantially improve binding affinity and reduce the effective doses of agents such as amphotericin B.^94,95,120 LaMastro et al developed penetratin-decorated liposomes encapsulating the antifungal drug posaconazole.⁹⁴ These liposomes were prepared using a thin-film hydration method followed by extrusion, and then penetratin peptide was conjugated to the surface via thiol-maleimide chemistry using a poly(ethylene glycol) linker. Penetratin improved liposome interaction with Candida cells, enhancing posaconazole delivery. The addition of penetratin also improved retention within biofilms by increasing interactions with biofilm cells and negatively charged components of the biofilm matrix. These decorated liposomes inhibited planktonic fungi at concentrations up to 8-fold lower than those of non-decorated versions and prevented biofilm formation at concentrations up to 1300-fold lower than the free drug. Antifungal agents themselves can also serve as dual-purpose functional components. For example, Yu et al reported that surface-functionalized amphotericin B can target intracellular fungi via ergosterol binding, while also promoting macrophage uptake.¹²¹ Further advancing this strategy, Yu et al developed amphotericin B-functionalized polymeric nanoparticles in which amphotericin B was both surface-decorated and encapsulated, enabling combined targeting and drug delivery.¹²¹ Surface-presented amphotericin B facilitates specific recognition of intracellular fungi by binding to ergosterol in the fungal membrane, while also enhancing uptake by macrophages. Following cellular internalization, the nanoparticles accumulate within infected cells, releasing amphotericin B locally and effectively eliminating intracellular C. neoformans.

In general, fungal cell wall-targeted nanocarriers demonstrate substantial potential for receptor-guided delivery, concentrating antifungal agents selectively at infection sites.

Stimuli-Responsive Systems

Spatiotemporal control of antifungal release can be achieved using stimuli-responsive formulations that utilize either endogenous fungal cues or externally applied triggers. Researchers have developed formulations that leverage infection‑specific conditions, such as acidic microenvironments (eg, pH‑sensitive liposomes) and fungal-secreted aspartic proteases (eg, enzyme‑cleavable hydrogels), to achieve localized, infection‑driven delivery.^96,122 For example, Nasti et al developed pH-sensitive liposomes for nystatin delivery using 1,2-dioleoylphosphatidylethanolamine (DOPE) and cholesteryl-hemisuccinate (CHEMS) as the pH-sensitive liposomal components (Figure 3A).⁹⁶ These liposomes destabilize in acidic intracellular compartments, allowing site-specific nystatin release and enhancing antifungal efficacy against intracellular pathogens. The liposomal formulation achieved 80% survival in murine cryptococcosis models, compared with only 20% with the free drug, attributed to its site-specific delivery to acidic lysosomes. Similarly, Uroro et al developed lipase-responsive polycaprolactone (PCL) nanoparticles encapsulating amphotericin B using an emulsion solvent evaporation method (Figure 3B).⁹⁷ This system achieved a triggered drug release specifically in the presence of lipase, an enzyme secreted by fungal pathogens such as Candida albicans, demonstrating significantly higher antifungal potency and reduced toxicity compared with the free drug.⁹⁷

Figure 3.

Stimuli-Responsive Drug Delivery Strategies for Site-Selective Antifungal Therapy. (A) pH-responsive, (B) enzyme-responsive, and (C) light-responsive drug delivery systems.

Although fungal enzyme- and pH-responsive systems offer selective drug release at infection sites, the reliability of these triggers may vary across patients.¹²³ Fungal enzyme activity can vary among clinical isolates and across infection conditions, while local pH may fluctuate with tissue microenvironment, potentially resulting in variable trigger sensitivity and drug release.^123–125 Interpatient differences in host inflammatory responses may further contribute to heterogeneity in the local infection microenvironment. Moreover, fungal pathogens such as Candida albicans can dynamically adapt to and modify the local pH, further affecting the consistency of pH-responsive triggering.¹²⁶ Therefore, clinically translatable stimuli-responsive systems should be designed with trigger thresholds that accommodate physiologically relevant variability and validated across heterogeneous infection conditions. Combining multiple disease-associated stimuli or incorporating complementary release mechanisms may further improve the robustness of drug release across diverse patient populations.

Externally triggered modalities, such as photodynamic therapy combined with light‑activated photosensitizers, including methylene blue or phthalocyanine, have shown enhanced antifungal efficacy when combined with gold- or chitosan-based nanoparticles (Figure 3C). This enhanced efficacy arises from improved photosensitizer delivery and uptake.^127,128

Prodrug Design for Selective Activation

The prodrug approach involves designing an inactive compound that is converted into its active form through specific chemical or enzymatic processes at the site of action. This bio-reversible modification is an effective strategy for optimizing pharmacokinetic and biopharmaceutical parameters while mitigating adverse effects.^129,130

Guo et al developed the amphotericin B-telodendrimer (AmB-TD) prodrug, which self-assembles into monodispersed micelles.⁹⁸ This formulation was developed by introducing multiple phenylboronic acid (PBA) moieties into a linear dendritic scaffold, enabling the successful conjugation of amphotericin B via a catalysis-free, fast, and dialysis-free “click” boronate chemistry drug-loading method. The resulting boronate linkage is reversible, facilitating accelerated drug release in response to acidic pH and ROS found at sites of fungal infection and inflammation. The prodrugs demonstrated substantially reduced nephrotoxicity and haemolytic activity compared with Fungizone^® and exhibited improved antifungal activity in both immunocompromised and immunocompetent mouse models. In another study, Nowak et al designed enzymatically cleavable antifungal conjugates using a “Trojan horse” strategy.¹³¹ In this approach, 5-fluorocytosine (5-FC) was linked to fatty acid residues (C2-C18) that served as molecular carriers via a trimethyl lock (TML) intramolecular linker.¹³¹ After cellular uptake, intracellular esterases hydrolyze the ester bond between the fatty acid and the TML linker, triggering rapid lactonization of the TML system and the concomitant release of 5-FC. The resulting conjugates demonstrated strong in vitro antifungal activity against Candida species.

Recent advances have expanded the concept of prodrug toward dual-delivery systems. Liu et al developed lipidic prodrug co-crystals of bifonazole-aspirin as a dual delivery platform that responds to the acidic and oxidative microenvironments of Candida biofilms.⁹⁹ Following topical administration to the perianal region of infected mice, this formulation reduced the fungal burden by more than four orders of magnitude, while simultaneously restoring tissue integrity and immune homeostasis.⁹⁹

Prodrugs can also be engineered to be activated by enzymes at the infection sites to minimize systemic exposure. Mercer et al reported that in superficial mycoses caused by dermatophytes, the inactive coumarin glycone esculin functions as a water-soluble prodrug.¹³² It is hydrolyzed in situ by β-glucosidases produced by dermatophytes and certain skin microbiota, yielding the active antifungal aglycone, esculetin.

Future Perspectives

The landscape of antifungal therapy is rapidly evolving, driven by advancements in technology and a deeper understanding of fungal pathogenesis. Several emerging areas have the potential to transform the development and application of antifungal drugs (Figure 4).

Figure 4.

AI-Driven Innovation in Antifungal Drug Development and Therapy.

AI-Driven Antifungal Therapy

Identification of New Antifungal Drugs

AI and machine learning (ML) are accelerating antifungal drug discovery by analyzing large biological and chemical datasets to identify and optimize new drug candidates. These approaches enable the screening and identification of promising compounds that might remain overlooked by traditional methods.¹³³ For example, researchers have used a Random Forest binary classification model to identify five key features predictive of the antifungal activity of synthetic polymers against Candida albicans. These features include a hydrophilic composition of at least 30%, a calculated partition coefficient (cLogP) of 0 to +0.5, a hydrophobic composition limited to 20%, a degree of polymerization of ≤18, and a cationic composition limited to 50%.¹³⁴ AI-based latent diffusion models have also been used to generate new antimicrobial peptides. Out of 40 synthesized candidates, 25 exhibited antifungal or antibacterial activity, including peptides active against C. glabrata.¹³⁵

Furthermore, ML classifiers have enhanced our understanding of antifungal resistance by identifying mutations that increase drug resistance in Candida auris.¹³⁶ The adoption of computational tools provides substantial benefits in addressing the complexities of fungal diseases and predicting fungal adaptation, thereby accelerating the discovery and development of new antifungal drugs.

Drug repurposing is a strategic approach to identifying new antifungal drugs among existing drugs. This method substantially reduces the time and cost associated with developing new drugs, as these compounds have already undergone thorough safety and pharmacological evaluations. AI-driven platforms expedite drug repurposing by analyzing large-scale clinical data alongside related scientific literature through natural language processing (NLP) techniques. This process helps uncover potential new applications of existing drugs.¹³³ For example, Gao et al trained a chemical-descriptor ML model to identify inhibitors of CaFKS1, an essential subunit of 1,3-β-glucan synthase, which is crucial for fungal cell wall biosynthesis and found exclusively in fungi.¹³⁷ The model demonstrated high classification accuracy, identifying goserelin and icatibant as promising candidates for drug repurposing with high confidence. Literature evidence supports experimental Candida inhibition for these predicted drugs, demonstrating the potential for ML to prioritize candidates for laboratory validation. Additionally, deep learning is being integrated into antifungal repurposing pipelines alongside structure-based methods to identify candidates with plausible target engagement. Joshi et al screened a library of 1930 FDA-approved drugs against Candida albicans dihydrofolate reductase (CaDHFR) using a comprehensive workflow that combined deep learning, molecular docking, X-score, similarity search, and molecular dynamics simulations to refine and rank candidates.¹³⁸ Post-molecular dynamics simulation identified paritaprevir, lumacaftor, and rifampin as promising candidates with favorable interaction stability and binding free energy relative to a reference ligand. Based on these findings, the authors proposed further exploration of these drugs as repurposed inhibitors of CaDHFR in the treatment of candidiasis.

Predictive Formulation Design Through AI and Computational Modeling

Conventional formulation optimization has often relied on laborious and resource-intensive trial-and-error approaches, typically involving the variation of one parameter at a time. In contrast, computer-aided and AI-assisted approaches enable predictive, in silico drug formulation development that accounts for the complex, nonlinear relationships between formulation variables.¹³⁹ One application of AI-assisted optimization in antifungal formulations involves ML-optimized mucoadhesive NLCs for fluconazole, intended to improve local therapy for oral candidiasis. Elkomy et al developed fluconazole-loaded NLCs and applied a joint optimization strategy using a Box-Behnken design, artificial neural networking, and variable-weight desirability functions.¹⁴⁰ This approach optimized the impact of drug, solid lipid, and surfactant concentrations on particle size and entrapment efficiency (EE). Experimental data from the optimized formulation (335 ± 13.5 nm, 73.1 ± 4.9% EE) closely matched the values predicted by the optimization model (approximately 330 nm, 75% EE).¹⁴⁰ The formulation exhibited sustained release, minimal histopathological effects on rabbit oral mucosa, and enhanced antifungal inhibition compared with a fluconazole solution. These findings underscore the value of ML-enhanced multi-factor optimization for antifungal formulation development.

However, a major challenge in applying predictive modeling to natural polymer-based formulations is the inherent batch-to-batch variability in their physicochemical properties, which can substantially affect critical quality attributes (CQAs), such as particle size, encapsulation efficiency, and drug release. For chitosan, key sources of variability include molecular weight distribution, degree of deacetylation, and impurity profiles. Current AI approaches can account for such variability by incorporating raw-material properties as critical material attributes (CMAs) within a Quality by Design (QbD) framework. Supervised learning models trained on historical data on CMAs and process analytical technology (PAT) data can predict formulation performance across different polymer batches and support the establishment of a robust design space, while anomaly detection may help identify atypical batches before manufacturing.^141,142 However, model generalizability remains constrained by limited datasets and variability arising from new polymer sources. Continuous model updating and integration with real-time PAT data are therefore important for improving batch-to-batch prediction and control.¹⁴³

Looking forward, AI, in combination with automation, robotics, and laboratory information management systems (LIMS), is enabling the development of autonomous “self-driving laboratories”. These platforms use AI to continuously design, execute, and iteratively learn from experiments, progressively refining the formulation strategies with minimal human involvement.¹⁴⁴ These developments highlight the transformative potential of AI-powered, autonomous systems in drug formulation, making the process more predictive, data-driven, and self-optimizing.

Precise and Personalized Antifungal Therapy

Similar to oncology, there is growing recognition that a one-size-fits-all approach to antifungal therapy is suboptimal. Individual treatment based on the patient and infecting organism can lead to better outcomes.¹⁴⁵ The effectiveness of therapy is influenced by factors such as the immune status of the patient, comorbidities, the site of infection, concomitant therapies, and the specific fungal species or strain.³¹ In this context, future antifungal treatment will likely incorporate principles of precision medicine. One aspect of precision medicine is MIPD, in which population pharmacokinetic models, along with patient-specific data such as therapeutic drug monitoring (TDM) and individual patient covariates, are used to individualize therapy. Using non-linear mixed-effects (NLME) models and Bayesian forecasting, prior population information can be integrated with individual observations to estimate individual pharmacokinetic parameters and adjust dosing. The approach has been successfully applied to optimize the dosing of antifungals, such as posaconazole and voriconazole, to improve treatment outcomes.^146,147 Precision nanomedicine further advances antifungal therapy by enabling individualized, controlled delivery strategies. Stimuli-responsive systems that leverage infection-specific cues or external triggers, such as light, can achieve localized antifungal release, thereby further advancing the potential for personalized therapy.

As the range of available formulations increases, clinicians will have multiple treatment options from which to choose, allowing for decisions based on patient-specific considerations. We anticipate interdisciplinary collaboration among infectious disease specialists, pharmacists, and data scientists to implement this personalized approach.

Challenges in AI-Driven Antifungal Therapy

While AI holds great promise in the discovery of antifungal drugs, several challenges limit its application. The most prominent of these is the scarcity of high-quality data, an inherent characteristic of the drug discovery and development process, exacerbated by financial constraints, time limitations, and confidentiality concerns. Deep learning models trained on small datasets are prone to overfitting and poor generalization, whereas traditional ML models remain restricted by their reliance on manually designed features, limiting their ability to fully capture complex biological relationships.¹⁴⁸ In antifungal AI application, particularly in antifungal peptides (AFPs) prediction models, challenges such as model complexity, limited data size, and decision-making biases can negatively affect model performance.¹⁴⁹ Furthermore, Ghislat et al identified issues such as bias, inconsistency, skewness, irrelevance, small dataset size, and high dimensionality, which undermine the reliability of AI models.¹⁵⁰ While many models perform well on retrospective benchmarks, few have demonstrated prospective value in identifying potent drug leads. Moreover, Li et al reported variability in ML model performance across different antifungal drug classes for C. auris resistance prediction, with higher accuracy for azoles and echinocandins but weaker performance for polyenes such as amphotericin B.¹³⁶ Beyond data-related constraints, regulatory hurdles pose a substantial barrier to AI integration in antifungal therapy. Regulatory agencies require extensive validation of AI models to ensure their safety and efficacy. The dynamic nature of biological systems and the unpredictability of drug interactions further limit AI’s predictive reliability in real-world scenarios. Ethical concerns, such as data privacy, algorithmic bias, and unequal access, also affect public trust and consent, thereby impeding the acceptance and integration of AI technologies.¹³³ The rapid evolution of pathogens presents another challenge to the sustained effectiveness of AI and ML systems. To maintain accuracy, these models must be frequently updated to capture new strains and resistance patterns, which necessitates a flexible, responsive approach to data collection and algorithm adjustment.¹⁵¹ Additionally, large language models (LLMs) are prone to “hallucinations” or confabulation. These models function by predicting the next word or sequence of words, but when they lack sufficient training on medical information, they can produce incorrect or misleading responses.¹⁵²

Conclusion

Invasive fungal infections (IFIs) remain a formidable, evolving, and persistently underestimated global health challenge. Addressing this challenge requires not only continued scientific efforts but also innovative strategies in therapeutic design. The limitations of conventional antifungal agents, such as drug resistance, poor pharmacokinetics, organ toxicity, and inadequate tissue penetration, highlight the insufficiency of traditional treatment approaches. However, these limitations present a compelling opportunity for modern formulation science to address.

Advanced drug delivery platforms, including nanotechnology-based systems, stimuli-responsive carriers, targeted nanocarriers, and prodrug strategies, have demonstrated exceptional potential in overcoming these complex and multifaceted barriers. When combined with AI, computational modeling, and precision medicine, these innovations promise a transformative shift from broad-spectrum, one-size-fits-all regimens to individualized, site-specific antifungal therapy.

However, translating these laboratory successes into clinical reality remains a substantial challenge. From a regulatory perspective, complex nanocarrier systems, including oral lipid-based formulations, often require product-specific evaluation because the carrier can alter the pharmacokinetics, biodistribution, and safety profile of the incorporated drug. Establishing reliable nonclinical-to-clinical translation and consistent systemic exposure is therefore critical for defining an appropriate starting dose and benefit–risk profile in early clinical trials. Timely engagement with regulatory agencies may further help clarify product-specific evidence requirements and facilitate clinical translation. Overcoming this gap will require integrated collaboration among formulation experts, data scientists, clinical pharmacologists, and regulatory scientists. This interdisciplinary collaboration will be pivotal in shaping next-generation antifungal therapeutics, ultimately transforming the treatment landscape and providing hope for patients confronting IFIs.

Funding Statement

This research was supported by National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (RS-2024-00336812) and the Industrial Technology Innovation Program (RS-2024-00439225) funded by the Ministry of Trade, Industry and Energy (MOTIE, Korea).

Data Sharing Statement

Data sharing is not applicable to this article as no new data were created or analyzed in this research.

Disclosure

All authors declare no conflicts of interest.