Areas of uncertainty in endophthalmitis care

Abstract

Endophthalmitis remains one of the most feared presentations in ophthalmic practice, not only because of its potential for rapid and irreversible vision loss but also due to the substantial diagnostic, therapeutic, and prognostic uncertainty that accompanies its management. Despite landmark trials, advances in vitreoretinal surgery, antimicrobial therapy, imaging, and molecular diagnostics, many aspects of endophthalmitis care continue to rely on clinical judgment rather than uniformly applicable, high-quality evidence. This review synthesizes contemporary areas of uncertainty spanning disease definition, diagnosis, treatment, antimicrobial resistance, and outcome prediction. We highlight the conceptual challenges in distinguishing infectious endophthalmitis from sterile or immune-mediated intraocular inflammation, particularly in culture-negative cases and post-intravitreal injection scenarios. Diagnostic ambiguity is compounded by limited microbiological yield, evolving molecular techniques with uncertain clinical interpretation, and supportive but non-specific imaging findings. Treatment-related uncertainties include the optimal timing and extent of vitrectomy in the era of small-gauge surgery, selection and dosing of intravitreal antimicrobials amid rising resistance and geographic variability, the contentious role of adjunctive corticosteroids, and management strategies for fungal and culture-negative endophthalmitis. Prognostication remains imprecise, with anatomical success often failing to predict functional recovery. Rather than reiterating established guidelines, this review focuses on real-world clinical gray zones and unresolved questions. Endophthalmitis is best viewed as a heterogeneous clinical syndrome shaped by microbial, host, and treatment-related factors. Recognizing uncertainty as an inherent feature of care underscores the need for individualized, adaptive decision-making and highlights priorities for future research, including standardized endpoints, region-specific surveillance, and integration of emerging imaging, molecular, and computational tools.

Key Words: Endophthalmitis; Diagnostic uncertainty; Intravitreal therapy; Vitrectomy timing; Antimicrobial resistance

Core Tip: Endophthalmitis remains a vision-threatening ophthalmic emergency in which clinicians must make urgent decisions despite substantial diagnostic and therapeutic uncertainty. Overlapping clinical presentations, frequent culture negativity, evolving antimicrobial resistance, and variable response to treatment complicate early management. Key controversies persist regarding the timing and extent of vitrectomy, selection and dosing of intravitreal antimicrobials, and the role of adjunctive corticosteroids. This review highlights these real-world gray zones and emphasizes the need for individualized, adaptive decision-making rather than rigid, algorithm-based approaches.

INTRODUCTION

Endophthalmitis remains one of the most feared entities in ophthalmic practice, not only because of its potential for rapid and irreversible vision loss, but also because it frequently forces clinicians to make high-stakes decisions in the setting of substantial diagnostic and therapeutic uncertainty. Despite decades of accumulated clinical experience, landmark trials, and advances in surgical techniques, imaging, and microbiological diagnostics, the management of endophthalmitis continues to rely heavily on clinical judgment and institutional practice patterns rather than uniformly applicable, high-quality evidence[1-4]. This paradox underscores a central challenge: Endophthalmitis is a well-recognized clinical entity, yet many of its foundational principles remain incompletely defined.

Traditionally viewed as an intraocular infection caused by a proliferating microorganism involving the inner coats of the eye, particularly the vitreous cavity, this conceptual framework has been increasingly challenged. Notably, a substantial proportion of clinically presumed infective endophthalmitis cases remain culture-negative despite severe intraocular inflammation, underscoring the limitations of conventional microbiological confirmation and contributing directly to diagnostic and therapeutic uncertainty[5-8]. Advances in molecular diagnostics have blurred the distinction between active infection, immune-mediated inflammation, and sterile inflammatory mimickers. Scenarios characterized by dense vitritis, culture negativity, or polymerase chain reaction (PCR)-based pathogen detection without viable organisms raise fundamental questions regarding disease classification, biological relevance, and therapeutic intent, often compelling clinicians to initiate aggressive antimicrobial therapy in the absence of microbiological confirmation[6,8].

Several long-standing management paradigms remain equally contentious, including the timing and extent of pars plana vitrectomy[2,3,9], reliance on empirical intravitreal antibiotics amidst emerging safety and resistance concerns[10-12], the role of adjunctive corticosteroids[13-15], and the optimal approach to culture-negative[6,16] or post-injection endophthalmitis[17-20] These challenges are further compounded by geographic variability in microbial profiles and resistance patterns, limiting the generalizability of standardized treatment algorithms[21]. Emerging approaches, including advanced retinal imaging biomarkers, molecular profiling, and artificial intelligence-based analytic frameworks, offer potential avenues to refine diagnosis and prognostication, although their clinical integration remains in an early and evolving phase.

This review critically examines and provides a structured synthesis of areas of uncertainty in the diagnosis, classification, management, and prognostication of endophthalmitis. By focusing on unresolved questions and real-world dilemmas rather than reiterating established guidelines, we aim to promote individualized, biologically informed decision-making and identify priorities for future research in this complex and heterogeneous syndrome. This review does not address medicolegal considerations, health-policy frameworks, or cost-effectiveness analyses, focusing instead on clinical, biological, and decision-making uncertainties relevant to routine ophthalmic practice.

LITERATURE SEARCH

A comprehensive narrative review of the English literature was conducted to identify studies addressing areas of uncertainty in the diagnosis, management, and prognosis of endophthalmitis. Electronic searches were performed using PubMed/MEDLINE, EMBASE, and the Cochrane Library, covering publications from inception to the most recent update prior to manuscript submission. Search terms included combinations of endophthalmitis, diagnostic uncertainty, culture-negative, vitrectomy, intravitreal antibiotics, antimicrobial resistance, fungal endophthalmitis, post-injection inflammation, outcomes, prognosis, imaging biomarkers, molecular diagnostics, and artificial intelligence. Reference lists of relevant reviews, landmark trials, and key original articles were manually screened to identify additional pertinent studies. Priority was given to randomized controlled trials, large observational studies, systematic reviews, consensus statements, and influential case series, with selective inclusion of emerging literature where high-level evidence was lacking. Given the narrative, concept-driven scope of this review focused on unresolved questions and clinical gray zones, formal Preferred Reporting Items for Systematic reviews and Meta-Analyses methodology and quantitative synthesis were not applied. Study selection emphasized clinical relevance, methodological rigour, and contribution to areas of uncertainty rather than exhaustive inclusion. This approach was intended to capture both established evidence and evolving perspectives relevant to real-world endophthalmitis care.

DEFINITIONAL AND CONCEPTUAL UNCERTAINTIES

Etymologically, endophthalmitis refers to inflammation of the internal coats and intraocular spaces of the globe[22]. Clinical evaluation, therefore, begins with defining the anatomical compartments involved. Inflammation confined to the outer coats, cornea or sclera is classified as keratitis or scleritis, respectively, whereas combined involvement of inner and outer coats, often with orbital extension, constitutes panophthalmitis[22-24]. In contrast, inflammation predominantly affecting the uveal tissues and intraocular spaces, including the anterior chamber, vitreous cavity, subhyaloid, intraretinal, subretinal or choroidal spaces, or even the retinal vasculature, achieves the anatomical definition of endophthalmitis[22,25]. While conceptually clear, this anatomical framework offers limited guidance in distinguishing infectious from non-infectious intraocular inflammation.

Clinically, infective endophthalmitis typically presents with severe ocular pain, rapid visual decline, hypopyon, fibrin, dense vitritis, and a progressive course[22,25]. Non-infective endophthalmitis more often manifests with milder pain, mild-to-moderate visual impairment, absence of hypopyon, low-grade anterior chamber or vitreous inflammation, and a relatively indolent course[26,27]. Importantly, overlap between these phenotypes is common, particularly in early presentations, partially treated infections, or immunocompromised hosts, limiting the reliability of clinical severity alone as a discriminating criterion[26].

In real-world ophthalmic practice, the term endophthalmitis is pragmatically used to denote a microbial infection within the eye, most commonly bacterial or fungal, characterized by intraocular inflammation, progressive vitritis, and a substantial risk of irreversible visual loss, often necessitating urgent intervention[25]. However, clinicians frequently encounter scenarios in which vitreous involvement is minimal or absent or microbiological confirmation is lacking[5,22]. Concurrently, several infective intraocular inflammatory entities, despite sharing overlapping clinical features such as vitritis, are not conventionally classified as endophthalmitis.

For instance, conditions such as acute retinal necrosis and toxoplasma retinochoroiditis, although caused by infectious agents and often associated with marked vitritis, are excluded from the endophthalmitis spectrum because the causative organisms do not freely proliferate within the vitreous cavity[28-30]. In these entities, organisms are typically not recoverable using routine smear or culture techniques, and diagnosis relies instead on antigen or nucleic acid detection using targeted PCR-based assays. In contrast, ocular toxocariasis has historically been described as an endophthalmitis-like presentation in some patients, as Toxocara larvae can invade intraocular tissues and incite sustained vitreous-centred inflammation, supported by characteristic clinical findings and serology, and occasionally direct intraocular identification/removal of larval material[31,32]. Similarly, tuberculosis-related retinal vasculitis or intermediate uveitis represents a hypersensitivity response to systemic mycobacterial antigens, wherein ocular samples usually fail to demonstrate viable organisms[33,34]; however, true tubercular endophthalmitis is recognized when hematogenous dissemination results in intraocular abscess formation with microbiological confirmation from ocular specimens[23].

Structural breaches such as vitreous wick, posterior capsule rupture, thin sclera in high myopia, filtering blebs, and sutureless vitrectomy ports may increase susceptibility to infection by facilitating microbial ingress. Minor penetrating injuries, including plant thorn or pencil tip trauma in children, further complicate early diagnosis due to delayed presentation and atypical microbial profiles[35,36]. This discordance between traditional definitions and contemporary clinical presentations highlights a fundamental conceptual controversy: What should be labelled as endophthalmitis, and on what biological or clinical basis? Thus, for routine clinical usage, the term endophthalmitis is best reserved for intraocular inflammation involving the inner coats of the eye in which a microbiological etiology is presumed, biologically plausible, and potentially demonstrable, typically through direct organism proliferation within intraocular spaces and isolation using conventional microbiological testing of ocular samples.

Assessment of a breach in ocular integrity provides additional etiologic context. Recent intraocular surgery, intravitreal injection, or trauma favours an exogenous process, which may be infectious, toxic, or immune-mediated. In the absence of an identifiable breach, endogenous causes, including hematogenous spread of infection or masquerade syndromes such as vitreoretinal lymphoma, must be considered. Moreover, immune-mediated entities such as tubercular retinal vasculitis or intermediate uveitis, post-fever retinitis, and white-dot syndromes may closely simulate infective endophthalmitis despite the absence of active intraocular microbial proliferation. Collectively, these ambiguities support viewing endophthalmitis as a heterogeneous syndrome rather than a single disease entity[25,26]. A structured, stepwise assessment integrating anatomical involvement, clinical phenotype, breach analysis, and inferred pathogenic mechanism is therefore essential. No single clinical or investigative feature is pathognomonic in isolation, and diagnostic uncertainty is intrinsic to many presentations. Figure 1 presents a pragmatic framework to navigate these gray zones and clarify areas of overlap encountered in routine practice.

Figure 1 Stepwise clinical framework for the diagnosis of endophthalmitis. AC: Anterior chamber; TB: Tuberculosis.

DIAGNOSTIC UNCERTAINTIES

The definitional ambiguities surrounding endophthalmitis are closely mirrored by persistent diagnostic uncertainties, largely stemming from the absence of any single clinical, microbiological, or imaging feature that reliably distinguishes infective from non-infective intraocular inflammation at presentation. In routine practice, the diagnosis of endophthalmitis is therefore largely clinical, frequently made under time pressure, with incomplete information, and in the context of potentially irreversible visual consequences if treatment is delayed[37,38]. The multifactorial contributors to diagnostic ambiguity, spanning clinical presentation, microbiological yield, imaging findings, and systemic context, are summarized in Table 1[6].

Table 1 Domains of diagnostic uncertainty in endophthalmitis.

Domain	Key sources of uncertainty	Clinical implications
Clinical	(1) Overlap in presenting features between infective and non-infective intraocular inflammation; (2) Variable severity at presentation (early, indolent, or partially treated infections); (3) Atypical presentations in immunocompromised or elderly patients; and (4) Rapid inflammatory responses in toxic or immune-mediated conditions mimicking infection	(1) Clinical severity alone cannot reliably distinguish infection from sterile inflammation; (2) Risk of delayed treatment in true infection or overtreatment in non-infective cases; and (3) Necessitates probabilistic decision-making rather than binary diagnosis
Microbiological	(1) High rates of culture negativity despite clinically presumed infection; (2) Prior antibiotic exposure reducing yield; (3) Fastidious or low-virulence organisms; (4) PCR positivity without viable organisms; and (5) Unclear significance of low microbial DNA copy numbers	(1) Absence of culture growth does not exclude infection; (2) PCR results require cautious interpretation and clinical correlation; and (3) Microbiological data often lag behind therapeutic decision-making
Imaging	(1) B-scan ultrasonography lacks specificity for infective vs sterile vitritis; (2) OCT findings reflect secondary inflammatory damage rather than etiology; (3) Widefield imaging documents extent but not cause of inflammation; and (4) Limited ability to predict microbial virulence or disease trajectory	(1) Imaging serves as an adjunct rather than a diagnostic arbiter; (2) Risk of over-interpreting nonspecific structural changes; and (3) Imaging cannot replace clinical and microbiological synthesis
Systemic/etiologic context	(1) Absence of identifiable ocular breach does not exclude infection; (2) Presence of a breach does not exclude toxic or immune-mediated reactions; (3) Subclinical or occult systemic infections; and (4) Masquerade syndromes (e.g., vitreoretinal lymphoma)	(1) Difficulty distinguishing endogenous infection from non-infective masquerade syndromes; (2) Requires parallel systemic evaluation and longitudinal reassessment; and (3) Misclassification may lead to inappropriate therapy or delayed diagnosis

PCR: Polymerase chain reaction; OCT: Optical coherence tomography.

Microbiological confirmation, traditionally regarded as the diagnostic cornerstone, is often elusive. Vitreous and aqueous cultures remain negative in a substantial proportion of clinically presumed infective endophthalmitis cases, even in the presence of severe inflammation[6,38]. Prior antibiotic exposure, low intraocular microbial burden, fastidious organisms, and limitations in sampling and laboratory processing all contribute to this gap. Molecular techniques such as PCR have improved pathogen detection rates but introduce new interpretive challenges[5,39]. Detection of microbial DNA does not necessarily indicate active intraocular infection, particularly in the absence of viable organisms or progressive inflammation, and may reflect residual genetic material, contamination, or transient intraocular exposure. Accordingly, PCR positivity cannot be equated with microbiological causality, nor does PCR negativity reliably exclude infection[39-41].

Clinical phenotype remains a central diagnostic guide but is inherently imperfect. Classical features of infective endophthalmitis, such as severe pain, hypopyon, fibrin, dense vitritis, and rapid clinical deterioration, may be absent in early disease, indolent infections, partially treated cases, or immunocompromised patients[37]. Conversely, non-infective entities, including toxic anterior or posterior segment syndromes, immune-mediated uveitis, and masquerade syndromes, may present with marked inflammation that closely mimics infection[42,43]. Disease tempo, while informative, is similarly unreliable, as certain infections evolve slowly, whereas toxic or inflammatory reactions may present abruptly[37]. Subclinical ocular surface disease, including severe blepharitis, purulent conjunctivitis, nasolacrimal duct obstruction, and periocular infections, may serve as under-recognized risk modifiers, particularly in patients undergoing repeated intravitreal injections[44]. Failure to identify and address these factors may contribute to diagnostic delay or misclassification of early infection vs sterile inflammation.

Ocular imaging serves as an important adjunct but remains supportive rather than definitive. B-scan ultrasonography is invaluable when media opacity limits fundus visualization but lacks specificity for differentiating infective from sterile inflammation[37,45,46]. Optical coherence tomography may demonstrate secondary retinal changes such as macular edema, subretinal fluid, or retinal necrosis; however, these findings are not pathognomonic and typically reflect downstream inflammatory injury rather than primary etiology[37]. Widefield imaging assists in documenting the extent of posterior segment involvement, particularly in endogenous or atypical cases, but does not independently resolve diagnostic ambiguity[47].

The distinction between exogenous and endogenous endophthalmitis further complicates diagnostic reasoning. While a recent breach in ocular integrity favours an exogenous process, its absence does not exclude infection, particularly in patients with systemic sepsis, indwelling catheters, or immunosuppression[47]. Conversely, the presence of a breach does not preclude non-infective mechanisms, as toxic or immune-mediated reactions may follow surgical or pharmacologic interventions[42,43].

Endogenous endophthalmitis represents a distinct diagnostic and therapeutic gray zone, often characterized by delayed recognition, subtle early ocular signs, and competing systemic priorities[47,48]. Unlike exogenous disease, ocular findings may be the first or only manifestation of systemic infection, and microbiological confirmation is frequently delayed or absent. Heterogeneous microbial profiles, frequent immunosuppression, and the need for coordinated systemic source control further compound uncertainty in the timing of vitrectomy, duration of antimicrobial therapy, and prognostication.

Collectively, these factors underscore that diagnostic uncertainty is intrinsic to endophthalmitis rather than exceptional. No single clinical sign or investigation is sufficient in isolation; instead, diagnosis relies on an evolving synthesis of anatomical involvement, clinical behaviour, investigative findings, and response to therapy[37,38]. Recognizing this uncertainty is essential, as premature diagnostic closure may result in delayed antimicrobial therapy in true infection or unnecessary intervention in non-infective inflammation[37]. These diagnostic uncertainties described above directly influence early therapeutic decisions, particularly regarding the timing of vitrectomy and escalation of antimicrobial therapy[38]. A structured, iterative diagnostic approach rather than a binary classification remains central to contemporary endophthalmitis care[6]. The following section examines the uncertainties surrounding treatment in the context of evolving surgical capabilities and contemporary clinical practice.

RISK STRATIFICATION AND HOST-RELATED MODIFIERS

Beyond microbial virulence and procedural factors, host-related and environmental modifiers play an under-recognized role in influencing susceptibility to endophthalmitis and contributing to diagnostic uncertainty[38]. Poor personal hygiene, including compromised oral health, excessive sweating, and periocular contamination, may alter conjunctival flora and increase the risk of peri-procedural microbial inoculation, particularly in tropical climates and high-volume injection settings[49,50]. Although direct causality is difficult to establish, these factors may act as facilitators rather than primary causes of infection. Systemic host factors further complicate risk stratification. Patients receiving chronic systemic antibiotics, high-dose corticosteroids, chemotherapy, or those admitted to intensive care units may exhibit atypical presentations due to altered immune responses and suppressed inflammatory signs[23,47]. In such settings, infection may progress with minimal early clinical warning, contributing to delayed diagnosis and higher rates of culture negativity. Emerging associations, including device-related airflow exposure (e.g., continuous positive airway pressure, respirators), tracheostomy care, and nutritional deficiencies, have been proposed as potential contributors to periocular microbial exposure; however, robust evidence remains limited, and these factors currently warrant awareness rather than protocolized intervention[47,51]. Recognizing these host-related modifiers is essential for maintaining a heightened index of suspicion and guiding individualized surveillance and counselling strategies.

TREATMENT-RELATED UNCERTAINTIES

Timing and extent of vitrectomy

The timing and extent of pars plana vitrectomy remain among the most debated aspects of endophthalmitis management. Traditionally, vitrectomy decisions in endophthalmitis have been guided by the Endophthalmitis Vitrectomy Study (EVS), which identified presenting visual acuity, specifically light perception, as the key threshold at which immediate vitrectomy conferred a meaningful advantage[1,2,52]. However, extrapolating EVS conclusions to contemporary practice warrants caution. EVS was conducted in an era of 20-gauge vitrectomy, limited visualization, lower cut rates, and restricted intraoperative control, factors that constrained both the safety and completeness of surgical intervention[9,52]. Consequently, the study’s emphasis on conservative surgical indications reflected not only disease biology but also technological limitations of the time. In contrast, the Complete Early Vitrectomy for Endophthalmitis (CEVE) paradigm emerged in the context of small-gauge, wide-angle, high-cut-rate vitrectomy systems, allowing earlier and more effective microbial load reduction, removal of inflammatory mediators, improved intravitreal antibiotic distribution, and enhanced diagnostic yield[3,53]. CEVE shifts the decision framework from presenting visual acuity to fundus visibility and disease severity, advocating early vitrectomy when posterior segment assessment is compromised, even in eyes with vision better than light perception[54,55]. Reported outcomes from the CEVE series suggest favourable anatomical and functional results, although these data are derived from retrospective cohorts rather than randomized trials[3]. The key differences between the EVS and CEVE paradigms are summarized in Table 2.

Table 2 Key differences between Endophthalmitis Vitrectomy Study and Complete Early Vitrectomy for Endophthalmitis.

Domain	EVS	CEVE
Ref.	Chen et al[10], 2021	Dib et al[3], 2020
Study design	Prospective, randomized multicenter trial	Retrospective cohort/treatment paradigm
Era and technology	20-gauge vitrectomy, limited visualization	Small-gauge, wide-angle, high-cut-rate vitrectomy
Primary trigger for vitrectomy	Presenting visual acuity (LP vs HM or better)	Fundus visibility and disease severity
Timing of surgery	Immediate vitrectomy is mainly for LP eyes	Early vitrectomy when the posterior view is obscured
Extent of vitrectomy	Core vitrectomy; no routine PVD induction	Maximal safe vitrectomy; cortical purulence removal
Treatment philosophy	Conservative, protocol-driven thresholds	Proactive, low threshold for surgery and retreatment
Outcome interpretation	The benefit is limited to the LP subgroup	Favourable outcomes reported across the wider VA spectrum
Generalizability today	Foundational but era-specific	Reflects contemporary surgical capability

EVS: Endophthalmitis Vitrectomy Study; CEVE: Complete Early Vitrectomy for Endophthalmitis; LP: Light perception; HM: Hand movements; PVD: Posterior vitreous detachment; VA: Visual acuity.

Despite these advances, uncertainty persists regarding patient selection for early vitrectomy. Eyes presenting with relatively preserved vision, indolent inflammation, or partial response to initial intravitreal therapy represent a persistent gray zone. While early surgical intervention may prevent disease progression in aggressive infections, it may expose some patients to unnecessary surgical risk when medical therapy alone would suffice. Conversely, delayed vitrectomy in rapidly progressive or highly virulent infections may compromise visual outcomes. The optimal extent of surgery also remains ill-defined. Whether a limited core vitrectomy is sufficient or whether a more complete vitreous removal is required, particularly in eyes with minimal vitreous opacification or predominantly anterior segment inflammation, continues to be debated[55,56].

To summarize, EVS and CEVE should not be viewed as competing doctrines but as products of different technological eras. EVS provides a foundational benchmark for evidence-based care, while CEVE reflects an adaptive response to evolving surgical capabilities. Together, they underscore that vitrectomy timing and extent in endophthalmitis should be contextualized within contemporary surgical feasibility, disease aggressiveness, and individual patient risk rather than rigid adherence to historical visual acuity thresholds[2,3,9,52].

Intravitreal antimicrobial selection and dosing

Intravitreal antimicrobial therapy remains the cornerstone of initial endophthalmitis management; however, important uncertainties persist regarding agent selection, dosing, and retreatment strategies. Empirical regimens are designed to provide broad coverage against common Gram-positive and Gram-negative bacterial or fungal organisms, yet rising antimicrobial resistance and marked geographic variability in pathogen profiles increasingly challenge the universal applicability of standardized protocols. The continued reliance on established intravitreal antibiotics reflects their proven efficacy and intraocular penetration, as well as the lack of equally effective alternatives, despite emerging concerns related to drug-specific adverse effects and resistance trends[57,58]. At present, common empiric dosing regimens continue to include intravitreal vancomycin 1.0 mg/0.1 mL with ceftazidime 2.25 mg/0.1 mL (or alternative Gram-negative coverage where appropriate)[59,60]. The role of systemic antibiotics, oral or intravenous, remains uncertain in most forms of exogenous endophthalmitis. While systemic therapy is essential in endogenous disease and selected high-risk scenarios, evidence supporting routine adjunctive use in postoperative or post-injection endophthalmitis is limited. Variability in ocular penetration, systemic toxicity, and lack of standardized treatment duration contribute to inconsistent practice patterns[61].

Dosing strategies introduce additional complexity, particularly in vitrectomised and silicone-oil-filled eyes, where intraocular pharmacokinetics may be altered, ultimately affecting the therapeutic exposure[62,63]. Decisions regarding repeat intravitreal injections are therefore frequently guided by clinical response rather than microbiological confirmation, as culture results are often delayed or negative[58]. In culture-negative endophthalmitis, the optimal duration and intensity of antimicrobial therapy remain ill-defined, complicating decisions regarding escalation, de-escalation, or cessation of treatment[5,6].

A further challenge lies in balancing antimicrobial efficacy against the risk of retinal toxicity and inflammation-related sequelae, particularly with repeated dosing[64-67]. From a broader perspective, endophthalmitis care highlights a tension between the need for aggressive empirical therapy to prevent irreversible vision loss and principles of antimicrobial stewardship aimed at minimizing resistance[57,68]. This dilemma is compounded by the absence of standardized intraocular susceptibility breakpoints, limiting the clinical translation of in vitro resistance data to intraocular efficacy[50]. Collectively, these factors underscore the need for individualized antimicrobial strategies that integrate disease severity, host factors, regional microbiology, and treatment response rather than rigid adherence to uniform dosing algorithms[6,57].

ROLE OF ADJUNCTIVE CORTICOSTEROIDS

The role of corticosteroids in infectious endophthalmitis remains contentious[69,70]. Although corticosteroids may attenuate the host inflammatory response and limit secondary tissue damage, their use raises concerns regarding impaired microbial clearance and masking of disease progression, particularly if administered prematurely or without adequate antimicrobial coverage. Consequently, evidence supporting routine steroid use is inconsistent, and clinical practice varies widely with respect to timing, route of administration, and patient selection[71].

Intravitreal corticosteroids are most often considered as adjuncts in eyes with severe intraocular inflammation, dense fibrin formation, or poor presenting visual acuity, typically after appropriate intravitreal antimicrobial therapy has been instituted[13-15,72]. However, their role in infections caused by highly virulent organisms or fungal pathogens remains particularly uncertain, where suppression of host immune response may be deleterious[73-75]. Systemic corticosteroids introduce additional complexity, including the influence of systemic comorbidities, immunosuppression, and the potential to exacerbate underlying infection[76,77].

An additional challenge relates to the impact of corticosteroids on clinical monitoring and retreatment decisions. Steroid-induced suppression of inflammation may obscure clinical endpoints traditionally used to assess treatment response, complicating decisions regarding repeat intravitreal therapy or surgical escalation. Furthermore, the absence of standardized dosing regimens, both for intravitreal and systemic corticosteroids, contributes to significant variability in practice[69,76]. In the absence of robust, disease-specific evidence, corticosteroid use in endophthalmitis remains highly individualized, guided by presumed pathogen virulence, disease trajectory, and clinician judgment rather than consensus-based protocols.

MANAGEMENT OF CULTURE-NEGATIVE ENDOPHTHALMITIS

Culture-negative endophthalmitis represents a distinct and persistent therapeutic challenge. In the absence of microbiological confirmation, clinicians must balance the risks of continuing broad-spectrum antimicrobial therapy against the possibility of non-infective or immune-mediated intraocular inflammation. Atypical mycobacterial endophthalmitis represents a particularly challenging subset of culture-negative disease, often characterized by delayed onset, indolent inflammation, poor response to conventional antibiotics, and delayed microbiological confirmation. Diagnostic delay frequently results in prolonged uncertainty and suboptimal outcomes, underscoring the need for heightened clinical suspicion in chronic postoperative inflammation[78-80]. Prolonged antimicrobial exposure may be unnecessary in sterile inflammatory conditions, whereas premature de-escalation risks undertreating occult infection with potentially severe visual consequences[6,81].

Prior antibiotic exposure is a common contributor to culture negativity and may alter both microbiological yield and clinical presentation, further complicating interpretation of treatment response[6,82]. In cases with persistent or worsening inflammation, repeat intraocular sampling or early vitrectomy may be considered for both diagnostic clarification and therapeutic benefit, although the incremental yield of repeat cultures remains variable[83,84]. Risk stratification based on disease severity, tempo of progression, and host factors is therefore essential to distinguish indolent presentations from aggressive infections that warrant continued or escalated therapy[85].

The absence of standardized management pathways for culture-negative endophthalmitis contributes to substantial variability in practice. Therapeutic decisions are typically guided by clinical trajectory, anatomical response, and evolving signs rather than predefined microbiological or temporal endpoints[6,82]. This reliance on dynamic reassessment underscores the limitations of current diagnostics and highlights the need for pragmatic, adaptable treatment algorithms. Future advances in molecular diagnostics and inflammatory biomarkers may help refine de-escalation strategies[5,8,86]; until then, individualized management with ongoing diagnostic reconsideration remains central to care.

THERAPEUTIC AMBIGUITIES IN FUNGAL ENDOPHTHALMITIS

Fungal endophthalmitis exemplifies treatment-related uncertainty owing to its often-indolent presentation, frequent diagnostic delays, and limited therapeutic options[87-89]. Management remains challenging, with ongoing debate regarding antifungal agent selection, the role and duration of systemic therapy, and the timing and extent of vitrectomy[88,89]. Species-specific differences further complicate decision-making, as infections caused by Candida species may demonstrate a more favourable response to medical therapy, whereas filamentous fungi such as Aspergillus are often associated with aggressive disease and poorer visual outcomes[90,91].

Intravitreal antifungal therapy is frequently required and may need to be repeated; however, optimal dosing intervals and treatment duration are poorly defined[87,88,92-95]. Systemic antifungal therapy introduces additional uncertainty related to variable intraocular penetration, systemic toxicity, and drug–drug interactions, particularly in medically complex patients[91,96]. Diagnostic delay, which is common in fungal disease, further adversely influences prognosis and often necessitates more aggressive intervention[48,87].

The role of vitrectomy in fungal endophthalmitis remains incompletely defined. Early surgical intervention may facilitate reduction of fungal burden, removal of vitreous abscesses, and improved antifungal penetration, yet the optimal timing relative to medical therapy is unclear[48,91,97,98]. Endogenous fungal endophthalmitis adds further complexity, requiring close coordination with systemic infection management and source control[48,91,96]. The use of corticosteroids in fungal endophthalmitis is particularly controversial. While steroids may reduce inflammation-related tissue damage, they carry a significant risk of exacerbating fungal proliferation and are generally approached with caution, if not avoided, especially in the acute phase[74]. In the absence of robust comparative data, current management relies largely on small case series and expert opinion, resulting in substantial heterogeneity in treatment strategies and outcomes[87-89].

POST-INTRAVITREAL INJECTION INFLAMMATION: A THERAPEUTIC GRAY ZONE

Post-intravitreal injection inflammation represents a particularly challenging therapeutic gray zone, as early differentiation between true infectious endophthalmitis and sterile inflammatory reactions is often unclear[17,99,100]. Overlap in presenting features, including pain, anterior chamber reaction, and vitreous haze, complicates initial management decisions, particularly when symptoms occur soon after injection[99,101]. While early-onset inflammation may favour a sterile or toxic mechanism, delayed or progressive presentations raise concern for infection; however, temporal patterns alone are insufficiently reliable to guide treatment[99,102].

The decision to initiate aggressive antimicrobial therapy vs close observation carries significant implications, especially in patients requiring ongoing intravitreal therapy for chronic retinal disease. Unnecessary intervention may lead to treatment interruption, patient anxiety, and procedural morbidity, whereas delayed therapy in true infection risks irreversible visual loss[102]. Microbiological sampling may aid diagnosis, but is often deferred in mild or improving cases due to limited yield and procedural risk[99,103].

An additional unresolved issue is the safety of re-exposure to the same pharmacologic agent following resolution. Evidence guiding rechallenge decisions is sparse, and risk stratification remains largely empirical[104]. Clear patient counselling regarding uncertainty, recurrence risk, and alternative treatment options is therefore essential. The lack of standardized definitions distinguishing infectious endophthalmitis from toxic or immune-mediated post-injection reactions further contributes to variability in practice[99,103]. Until more robust data are available, management remains individualized and heavily dependent on clinical evolution and clinician judgment[105].

TREATMENT ENDPOINTS AND THERAPEUTIC FUTILITY

One of the least defined aspects of endophthalmitis management relates to treatment endpoints. Clear, evidence-based criteria for therapeutic success, failure, or futility remain lacking[37,38]. Clinical improvement may lag behind microbiological clearance, while persistent intraocular inflammation does not necessarily indicate ongoing infection[6]. As a result, decisions to escalate therapy, proceed to surgery, or de-escalate treatment are rarely protocol-driven and instead rely on iterative clinical judgment informed by disease trajectory, anatomical response, and host factors[37]. This dynamic decision-making underscores the need for continuous reassessment rather than rigid treatment algorithms[38].

An additional challenge lies in defining meaningful outcomes. Anatomical control of infection does not always translate into functional visual recovery, and patient-reported visual function and quality-of-life outcomes are infrequently incorporated into therapeutic decision-making[77,106]. Imaging biomarkers, including serial optical coherence tomography and ultrasonographic assessment, may provide adjunctive information regarding treatment response, but their role in defining endpoints remains evolving[37,45,46]. The absence of standardized, time-based reassessment frameworks further contributes to variability in practice[6]. Collectively, these uncertainties reflect the biological heterogeneity of endophthalmitis and the limitations of existing evidence. Rather than representing deficiencies in care, they highlight the necessity for individualized, adaptive management strategies[38]. Future research should prioritize consensus definitions of treatment success and failure, integrate patient-centred outcomes, and develop pragmatic frameworks that better align therapeutic decisions with real-world clinical complexity[6].

ANTIMICROBIAL RESISTANCE AND GEOGRAPHIC VARIABILITY

Antimicrobial resistance represents an increasingly important yet under-recognized source of uncertainty in the management of endophthalmitis. Empirical intravitreal regimens have traditionally relied on relatively predictable susceptibility patterns, particularly for Gram-positive organisms that predominate in postoperative disease. However, evolving resistance profiles, both across regions and over time within the same region, are increasingly challenging the assumption that standard antibiotic combinations provide uniform and durable coverage[10,57,107-111].

Geographic variability in causative organisms further complicates empirical decision-making. The microbial spectrum of endophthalmitis varies substantially between regions, influenced by local surgical practices, healthcare infrastructure, antibiotic utilization patterns, and population-level comorbidities[57,58]. While coagulase-negative staphylococci predominate in many high-income settings, Gram-negative organisms, virulent streptococcal species, and fungal pathogens are encountered more frequently in other regions[12,57]. Applying treatment algorithms derived from geographically distinct cohorts may therefore result in suboptimal empirical coverage or delayed therapeutic adjustment[58]. Geographic variability in microbial spectrum, resistance patterns, and their clinical implications in endophthalmitis are summarized in Table 3[52-54,112-118].

Table 3 Geographic variability in endophthalmitis: Microbiology, resistance patterns, and clinical implications.

Region/setting	Ref.	Dominant microbial pattern	Key resistance/risk signal	Practical clinical implication
North America/Western Europe	Flynn et al[52], 2008	Coagulase-negative Staphylococcus, Staphylococcus aureus	Rising fluoroquinolone resistance; preserved susceptibility to intravitreal first-line agents	Standard empirical regimens generally effective, but ongoing surveillance required
South Asia	Topol et al[112], 2019	Higher burden of Gram-negative organisms (Pseudomonas, Klebsiella) and fungi	Increasing resistance to fluoroquinolones and cephalosporins	Broader empirical coverage; early consideration of fungal infection
East Asia	Wong et al[113], 2000	Mixed Gram-positive/Gram-negative spectrum; higher streptococcal prevalence	Marked regional variability; limited standardized resistance data	Local microbiological data essential for empirical decision-making
Middle East	Shams Abadi et al[114], 2023; Falavarjani et al[115], 2012	Predominantly Gram-positive; Gram-negative organisms in trauma	Emerging multidrug-resistant Gram-negative isolates	Trauma-associated cases may warrant early surgery and tailored therapy
Africa	Solomon et al[116], 2025	Higher proportion of post-traumatic and endogenous infections	Sparse resistance data; limited laboratory infrastructure	Empirical management common; need for regional registries and surveillance
Latin America	Morris et al[53], 2021	Heterogeneous spectrum (Gram-positive, Gram-negative, fungal)	Increasing resistance reported in tertiary centers	Empirical therapy often requires regional customization
Post-traumatic Endophthalmitis (Global)	Mitra et al[117], 2021; Long et al[118], 2014	Bacillus spp., Gram-negative organisms	High virulence predominates over resistance	Early vitrectomy and aggressive therapy prioritized
Endogenous Endophthalmitis (Global)	Kuhn et al[54], 2006	Candida, Aspergillus, Gram-negative bacteraemia-associated organisms	Resistance influenced by prior systemic antimicrobial exposure	Requires systemic source control and prolonged, coordinated therapy

Bacillus spp.: Bacillus species.

Resistance among Gram-negative organisms, particularly to fluoroquinolones and cephalosporins, has emerged as a growing concern, especially in post-traumatic and endogenous endophthalmitis[12,68]. Although intravitreal antibiotic concentrations often exceed minimum inhibitory concentrations, the clinical relevance of in vitro resistance remains incompletely understood[57]. This uncertainty is compounded by the absence of standardized intraocular susceptibility breakpoints, limiting meaningful interpretation of microbiological data and complicating retreatment decisions when clinical response is suboptimal[57,58].

The tension between aggressive empirical therapy and antimicrobial stewardship is particularly pronounced in endophthalmitis. Given the catastrophic consequences of undertreatment, clinicians often favour broad-spectrum coverage and repeated intravitreal dosing, even in the absence of microbiological confirmation[58]. However, prolonged systemic therapy in endogenous cases, repeated intravitreal exposure, and widespread prophylactic antibiotic use may contribute to resistance at both individual and population levels[10,57]. Collectively, these challenges highlight the need for region-specific and temporally updated microbiological surveillance, standardized resistance reporting, and multicenter registries capable of informing adaptive, evidence-based treatment strategies[57,68]. Integrating antimicrobial stewardship principles into endophthalmitis care without compromising urgency will be essential to addressing antimicrobial resistance as a persistent and evolving gray zone in clinical practice[58].

PROGNOSTIC UNCERTAINTY AND OUTCOME PREDICTION

Despite timely and appropriate treatment, predicting anatomical and functional outcomes in endophthalmitis remains inherently challenging. This uncertainty reflects the multifactorial nature of the disease, in which presenting severity, host immune response, microbial virulence, timing of intervention, and surgical management interact in complex and often unpredictable ways. Consequently, prognostication in endophthalmitis is frequently retrospective rather than reliably predictive at the time of presentation[37,38].

Several clinical variables have been associated with outcomes, yet their utility at the individual patient level is limited. Presenting visual acuity, long regarded as a key prognostic marker, provides only a coarse estimate of disease severity and may be confounded by media opacity, inflammatory burden, or pre-existing ocular pathology rather than irreversible retinal damage[2,52]. Importantly, early clinical response and disease trajectory over time often provide greater prognostic insight than baseline features alone, underscoring the value of temporal evolution as a prognostic signal[37]. Microbiological factors further complicate outcome prediction. Although infections caused by highly virulent organisms are generally associated with poorer visual outcomes, culture negativity does not reliably confer a favourable prognosis, nor does organism identification consistently predict treatment response[6,57]. Delays in microbiological confirmation and reliance on empirical therapy limit the real-time prognostic relevance of microbiological data[58].

Imaging biomarkers have emerged as potential adjuncts for outcome prediction, particularly with advances in optical coherence tomography. Findings such as macular edema, subretinal fluid, retinal necrosis, or photoreceptor disruption may offer insight into structural damage; however, these changes often reflect secondary inflammatory injury and evolve dynamically with treatment[37]. Serial imaging, rather than isolated baseline findings, may therefore provide more meaningful prognostic information[45,46]. Notably, anatomical success following vitrectomy does not necessarily translate into functional visual recovery, especially in eyes with extensive retinal or choroidal insult[77].

Host-related factors, including age, immune status, systemic comorbidities, and baseline ocular health, further influence outcomes but are difficult to integrate into standardized prognostic models[57]. Patient-reported visual function and quality-of-life measures are also infrequently assessed, despite their relevance to real-world recovery and expectation management[106]. Collectively, these limitations highlight the need for composite, dynamic prognostic frameworks that integrate clinical course, imaging evolution, and microbiological context[6,38]. Until such models are validated, clinicians must rely on probabilistic counselling aligned with evolving disease response rather than definitive early predictions[37].

FUTURE DIRECTIONS

Reducing prognostic uncertainty in endophthalmitis will require a shift from static, single-time-point assessment toward dynamic, data-integrated prediction frameworks.[37,38] Advances in retinal imaging, molecular diagnostics, and computational analytics offer the potential to move beyond coarse clinical predictors toward individualized risk stratification; however, translating these tools into clinically meaningful models remains challenging[6]. Imaging biomarkers represent a promising avenue for outcome prediction. High-resolution optical coherence tomography allows detailed assessment of retinal architecture, including photoreceptor integrity, retinal necrosis, and macular involvement, which may correlate with functional recovery[37]. Importantly, longitudinal imaging rather than isolated baseline findings is likely to provide greater prognostic value by capturing treatment response and inflammatory resolution over time[45,46]. Ultra-widefield imaging may further assist in quantifying disease extent and peripheral involvement, particularly in endogenous or atypical presentations[47]. Standardization of image acquisition, feature definition, and reporting will be essential before imaging biomarkers can be reliably incorporated into prognostic algorithms[46].

Molecular diagnostics may complement imaging by providing insights into pathogen burden, virulence, and host-pathogen interaction. Quantitative PCR metrics, inflammatory cytokine profiling, and host immune signatures have theoretical potential to distinguish self-limited inflammation from aggressive infection, although their clinical applicability remains investigational[5,6,8]. Emerging metabolomic approaches add a biologically relevant dimension by characterizing intraocular metabolic signatures associated with microbial activity, host inflammatory response, and tissue injury. Metabolic alterations related to oxidative stress, immune activation, and cellular energy pathways may help differentiate infectious from sterile inflammation and serve as early indicators of treatment response[107-109]. While metabolomics remains largely confined to research settings, its integration with molecular and imaging biomarkers holds promise for refining diagnostic and prognostic precision.

Artificial intelligence provides a unifying framework to integrate heterogeneous data streams, including clinical variables, imaging features, microbiological findings, molecular profiles, and temporal disease evolution. Although machine-learning models may uncover latent prognostic patterns, the rarity and heterogeneity of endophthalmitis pose challenges for model training, validation, and generalizability[110-112]. Multicenter collaboration and standardized outcome definitions will therefore be critical. Ultimately, future prognostic tools should aim to provide probabilistic guidance that augments rather than replaces clinical judgment. By integrating biomarker-driven and AI-enabled approaches within a clinically grounded framework, the field may begin to meaningfully reduce, though not eliminate, the uncertainty that defines endophthalmitis care.

CONCLUSION

Endophthalmitis remains one of the most challenging entities in ophthalmic practice, defined as much by uncertainty as by urgency. Despite advances in surgical techniques, antimicrobial therapy, imaging, and molecular diagnostics, many aspects of its diagnosis, classification, management, and prognostication continue to rely on probabilistic judgment rather than definitive evidence. This review emphasizes that endophthalmitis is best regarded as a heterogeneous clinical syndrome shaped by microbial virulence, host response, route of entry, and timing of intervention. Diagnostic gray zones, evolving antimicrobial resistance, geographic variability, and limited prognostic markers constrain the universal applicability of standardized treatment algorithms. Importantly, uncertainty should be recognized not as a failure of care but as an inherent clinical reality that necessitates individualized, adaptive decision-making. Emerging tools, including advanced imaging, molecular and metabolic biomarkers, and artificial intelligence, offer promise for refining risk stratification, but their value will depend on rigorous validation and thoughtful integration. Future progress will require collaborative efforts to build consensus around definitions and endpoints while incorporating patient-centred outcomes alongside anatomical success.