Clinical efficacy of intrastromal voriconazole injection in fungal keratitis involving different corneal stromal depths

Abstract

Purpose

To evaluate the clinical efficacy of intrastromal injection of 0.5% (5 mg/mL) voriconazole in fungal keratitis (FK) involving different corneal stromal depths.

Methods

This was a non-randomized, retrospective study based on patient treatment preferences. A total of 53 patients (53 eyes) with confirmed fungal keratitis treated between March 2021 and December 2023 were enrolled. Based on anterior segment optical coherence tomography (AS-OCT) findings, lesions were stratified by depth into superficial stromal involvement (Group A, n = 29) and deep stromal involvement (Group B, n = 24). Each group was further subdivided according to treatment modality: non-injection group (conventional antifungal therapy) (subgroups A1 and B1) versus injection group (conventional antifungal therapy combined with intrastromal voriconazole injection) (subgroups A2 and B2). The following clinical parameters were compared among the four subgroups: reduction ratio of corneal epithelial staining area at 7 days post-treatment initiation, best-corrected visual acuity (BCVA), corneal epithelial healing time, clinical cure time, and final clinical outcomes.

Results

All groups demonstrated significant reduction in ulcer area after 7 days of treatment (P < 0.05). In superficial stromal fungal keratitis, ulcer improvement rates were comparable between groups A1 (37.19%) and A2 (36.11%) (P = 0.838). In deep stromal fungal keratitis, group B2 (41.30%) showed significantly greater improvement compared to B1 (14.95%) (P < 0.0001). Final BCVA improved significantly in all groups, with no significant intergroup differences. Epithelial healing time and clinical cure time showed no significant differences across groups. The overall clinical cure rate was 98.1%, achieved through the non-injection group in 79.2% of cases and the injection group in 18.9% of cases. Treatment failure occurred in one case (1.9%).

Conclusions

Intrastromal voriconazole injection provides no additional benefit for superficial stromal fungal keratitis compared with conventional topical therapy and is therefore not recommended as routine treatment given its invasive nature. For deep stromal fungal keratitis, intrastromal injection demonstrates superior early lesion control but does not reduce overall disease duration or improve final visual outcomes. Further large-scale prospective studies are warranted to establish its optimal indications and clinical value.

What is known

What is new

(1) Fungal keratitis is primarily managed with pharmacological therapy; however, the availability of antifungal agents for ocular use remains limited. These agents demonstrate poor tissue penetration and insufficient efficacy against deep corneal infections, frequently resulting in delayed ulcer healing, infection recurrence, and corneal perforation. (2) Voriconazole, a second-generation triazole antifungal agent, exhibits superior antifungal activity and corneal stromal penetration compared to conventional agents. However, its penetration into deep stromal layers remains limited, potentially leading to residual lesions or recurrence.

(1) For patients with superficial stromal fungal keratitis, the addition of intrastromal voriconazole injection to conventional antifungal therapy demonstrated no significant therapeutic advantage. Given the invasive nature of this procedure, intrastromal voriconazole injection is not recommended for superficial stromal infections. (2) For patients with deep stromal fungal keratitis, combined therapy with intrastromal voriconazole injection and conventional antifungal agents resulted in more effective reduction of corneal epithelial staining area at 7 days post-treatment compared to conventional therapy alone. However, this approach did not significantly shorten clinical cure time or improve final clinical outcomes.

Introduction

Fungal keratitis (FK) is a highly recalcitrant and potentially blinding corneal infectious disease that can lead to permanent vision loss in severe cases. Primary predisposing factors include ocular trauma, prolonged contact lens wear, extended use of broad-spectrum antibiotics or corticosteroids, ocular surface disorders, and compromised host immunity [1, 2]. In regions with large agricultural populations, the risk of trauma-related fungal ocular exposure is particularly elevated [3].

Currently, FK is primarily managed with pharmacological therapy; however, the availability of topical antifungal agents for ophthalmic use remains limited in China, with only natamycin and fluconazole eye drops commercially available. These agents demonstrate poor corneal penetration and limited efficacy against deep stromal infections, frequently resulting in delayed ulcer healing, recurrent infection, and corneal perforation [4–6], ultimately necessitating therapeutic keratoplasty. However, corneal donor shortage, high surgical costs, and poor compliance with long-term immunosuppressive regimens significantly limit the accessibility of corneal transplantation in developing countries [7].

In recent years, intrastromal injection of antifungal agents has emerged as a focal point of clinical research. Voriconazole, a second-generation triazole antifungal agent, exhibits superior antifungal activity and corneal stromal penetration compared to conventional topical agents. However, even with topical voriconazole therapy, penetration into the deep stromal layers remains suboptimal, potentially leading to persistent infection or disease recurrence [8]. To enhance local drug concentration and improve therapeutic efficacy in deep fungal infections, intrastromal voriconazole injection has been explored as an adjunctive treatment approach.

Nevertheless, systematic studies evaluating the efficacy of this intervention stratified by lesion depth remain limited. This study aimed to investigate the clinical efficacy of intrastromal voriconazole injection in treating fungal keratitis at different stromal depths, thereby providing evidence-based guidance for individualized treatment strategies.

Methods

Study design and participants

This was a non-randomized, patient-preference–based retrospective study. Treatment allocation was determined by patients’ informed choice after explanation of potential benefits and risks, rather than by random assignment. We reviewed consecutive patients clinically diagnosed with fungal keratitis presenting to the Department of Ophthalmology, People’s Hospital of Guangxi Zhuang Autonomous Region, between March 2021 and December 2023. Inclusion criteria comprised patients aged 18–75 years with a confirmed fungal keratitis diagnosis, with informed consent provided by both patients and their families. Exclusion criteria included corneal perforation or impending perforation, concurrent active endophthalmitis, pregnancy or lactation, hepatic or renal insufficiency, mixed infections, antifungal drug allergy, prior ocular surface surgery, drug resistance confirmed by susceptibility testing, and non-adherence to medical advice. Patients with poorly controlled systemic diseases such as diabetes mellitus, as well as those with severe immunocompromised conditions, were excluded.

Grouping and treatment protocol

Patients were stratified by ulcer infiltration depth into a superficial stromal group (Group A, ≤ 50% corneal thickness) and a deep stromal group (Group B, > 50% corneal thickness without penetrating Descemet’s membrane) based on AS-OCT findings. Lesion depth was determined by AS-OCT imaging at the site of maximum stromal infiltration. Total corneal thickness was measured from the anterior epithelial surface to the posterior endothelial surface, and the depth of infiltration was measured from the anterior surface to the deepest point of stromal opacity. The ratio of infiltration depth to total corneal thickness was calculated for group assignment. All measurements were performed independently by two experienced ophthalmologists, with discrepancies resolved by a third senior physician to ensure measurement repeatability. Superficial stromal fungal keratitis was defined as lesions involving ≤ 50% of stromal thickness, whereas deep stromal fungal keratitis was defined as lesions involving > 50% of stromal thickness without penetration of Descemet’s membrane. In routine clinical practice, after comprehensive explanation of the risks and benefits associated with both treatment strategies, treatment allocation was determined by patients’ informed preference in this non-randomized design. Within each depth-stratified group, non-injection group (A1/B1) or injection group (A2/B2). Group A comprised 29 cases (A1, 15; A2, 14), and Group B comprised 24 cases (B1, 12; B2, 12).

Conventional antifungal drug therapy included: 2% voriconazole eye drops (20 mg/mL) administered hourly [9, 10]; 0.15% amphotericin B eye drops hourly; 5% natamycin eye drops hourly; levofloxacin eye drops four times daily; and oral voriconazole tablets (200 mg twice daily, with a doubled loading dose).

Intrastromal injection technique: Following topical surface anesthesia with 0.5% proparacaine hydrochloride and eyelid speculum placement, the conjunctival sac was irrigated with 0.5% povidone-iodine. Using the three-zone method [11], the cornea was divided into: an entry zone (2-mm annular region near the limbus), an injection zone (central 5–6 mm diameter area designated as the prohibited entry zone), and remaining areas as permissible injection zones (Figs. 1 and 2). A 0.5% voriconazole injection solution was prepared (Pfizer Inc., dissolved in 10 mL sterile water for injection, then diluted) [12]. Using a 30-gauge needle inserted horizontally through the entry zone into the stroma, the solution was slowly injected into the injection zone after complete penetration of the needle tip, ensuring drug distribution to cover the ulcer and surrounding tissues. Multiple injection points were utilized with a total volume of 0.1–0.3 mL. For eccentric ulcers, injection points adjacent to the lesion were selected with appropriately increased injection pressure. Successful injection was confirmed by stromal edema and haziness surrounding the lesion. Post-procedure antibiotic ointment was applied with eye patching. Each patient received a single intrastromal voriconazole injection during the treatment course; repeat or scheduled reinjections were not performed.

Fig. 1.

Schematic diagram of three-zone method for corneal intrastromal injection

Fig. 2.

Schematic diagram of corneal epithelial staining area measurement using ImageJ software after fluorescein sodium staining. A: Linear scale bar setting, with corneal horizontal diameter standardized to 11.5 mm; B: Manual delineation of the stained area following fluorescein sodium staining; C: Automated calculation of the delineated stained area; D: Representative results of corneal epithelial staining area measurement

Outcome measures

Baseline data collected included age, gender, affected eye, causative factors, systemic diseases, disease duration, previous antifungal treatment history, and presence of hypopyon. Fungal culture and in vivo confocal microscopy (IVCM) results were documented. Corneal epithelial staining area was measured using fluorescein sodium staining with slit-lamp photography and quantified precisely using ImageJ software (Fig. 3). Baseline corneal epithelial staining area was recorded for all patients and used as an objective surrogate of ulcer size. Baseline ulcer area was compared between corresponding treatment subgroups to ensure comparability. Visual acuity was recorded using decimal charts or hand motion/counting fingers and converted to LogMAR values [13, 14]. The epithelial staining improvement ratio was calculated as: Improvement ratio = (Pre-treatment area - Post-treatment area at 7 days) / Pre-treatment area × 100%.

Fig. 3.

Comparison of the improvement ratio of corneal epithelial staining area before and after treatment among the four groups

Efficacy evaluation

Treatment outcomes were classified as clinical cure or treatment failure [15]. Medical cure was defined as ulcer repair, negative fluorescein staining, resolution of hypopyon and corneal edema, with no fungal hyphae detected on IVCM. Combined medical-surgical cure required surgical intervention (conjunctival flap coverage or corneal transplantation) following ineffective medical therapy, with subsequent lesion healing and complete fungal clearance confirmed by negative IVCM. Treatment failure was defined as absence of improvement or worsening of ulceration with perforation or intraocular infection, persistent or increased fungal hyphae on IVCM, ultimately necessitating enucleation. Epithelial healing time was defined as the duration required for corneal epithelium to completely cover the ulcer in medically responsive patients. Clinical cure time indicated complete fungal eradication from the lesion. In eyes requiring conjunctival flap coverage or corneal transplantation, in vivo confocal microscopy findings were interpreted in conjunction with clinical indicators, including epithelial healing, resolution of inflammation, and absence of progressive stromal infiltration, rather than being used as the sole criterion for fungal eradication.

Statistical analysis

SPSS 27.0 was used for statistical analysis. Continuous variables were expressed as mean ± standard deviation or median (interquartile range), and categorical variables as frequencies and percentages. Independent samples t-test was used for two-group comparisons of continuous variables, and one-way ANOVA for multiple group comparisons with LSD-t test for post-hoc analysis. Chi-square test was applied for categorical variables. P < 0.05 was considered statistically significant.

Results

Baseline characteristics comparison

The original study planned to enroll 20 patients per group; however, due to loss to follow-up (n = 12), non-compliance with inclusion/exclusion criteria (n = 7), poor image quality (n = 3), and missing data (n = 5), the final analysis included 29 cases in Group A (A1: 15, A2: 14) and 24 cases in Group B (B1: 12, B2: 12). Baseline characteristics including age, gender, affected eye, causative factors, systemic diseases, disease duration, previous antifungal treatment history, and presence of hypopyon showed no significant differences between the A1 and A2 groups or between the B1 and B2 groups (P > 0.05) (Table 1). Baseline corneal epithelial staining area also showed no significant difference between A1 vs. A2 groups and B1 vs. B2 groups (P > 0.05).

Table 1.

Comparison of baseline characteristics among four groups

Characteristics	Group A1 (n = 15)	Group A2 (n = 14)	P₁*	Group B1 (n = 12)	Group B2 (n = 12)	P₂*
Gender, No.			0.860			0.105
Male	7	7		5	9
Female	8	7		7	3
Age (years)	55.07 ± 10.21	59.14 ± 12.34	0.340	57.58 ± 9.28	53.58 ± 6.91	0.244
Affected eye, No.			0.579			0.105
Left	8	6		3	7
Right	7	8		9	5
Hypopyon, No.			0.682			0.351
None	11	11		8	5
< 0.5 mm	1	0		0	1
≥ 0.5 mm	3	3		4	6
Disease duration(days)	11.0(10.0,16.0)	28.0(10.3,31.5)	0.105	22.0(10.0,33.0)	29.5(15.0,51.3)	0.154
Systemic diseases, No.	2	1	0.591	0	2	0.148
Previous antifungal therapy, No.	5	3	0.481	4	6	0.408

*P¹, P²: Statistical results comparing baseline characteristics between A1 vs. A2 groups and B1 vs. B2 groups, respectively; categorical variables analyzed using Fisher’s exact test, continuous variables using independent samples t-test or Mann-Whitney U test as appropriate

ᵃIncludes diabetes mellitus, hypertension, and chronic gastritis

Causative factors comparison

Plant trauma represented the primary causative factor across all four groups, including grass debris, tree branches, sugarcane leaves, rice grains, wood chips, chestnut shells, and bean sprouts. Other factors were secondary, including insect bites, eye rubbing, sand particles, and iron rod injuries (Table 2).

Table 2.

Comparison of causative factors among four groups

Causative factors	A1 Group(n = 15)	A2 Group(n = 14)	B1 Group(n = 12)	B2 Group(n = 12)
Plant traumab	6(40.00%)	7(50.00%)	4(33.33%)	8(66.67%)
Other causesc	5(33.33%)	6(42.86%)	2(16.67%)	2(16.67%)
No clear cause	4(26.67%)	1(7.14%)	6(50.00%)	2(16.67%)

ᵇIncludes grass debris, tree branches, sugarcane leaves, rice grains, wood chips, chestnut shells, bean sprouts, and bamboo chips

ᶜIncludes insects, eye rubbing, sand, and iron rod injury

Fungal culture results

IVCM examination was positive in 52 patients (98.11%) and negative in 1 patient (1.89%). Among fungal culture results, 42 samples (79.25%) were positive, including 30 samples (56.66%) positive for Fusarium species, 5 samples (9.43%) positive for Aspergillus species (including A. flavus and A. fumigatus), 2 samples (3.77%) positive for Paecilomyces lilacinus, 2 samples (3.77%) positive for Curvularia species, and 3 samples (5.66%) positive for other fungal types, including Dematiaceae, Bipolaris, and crescent-shaped fungi, one case each (Table 3).

Table 3.

Fungal culture results among four groups

Fungal species	A1 Group (n = 15)	A2 Group (n = 14)	B1 Group (n = 12)	B2 Group (n = 12)	Total (n = 53)
Fusarium species	9(60.00%)	10(71.43%)	6(50.00%)	5(41.67%)	30(56.60%)
Aspergillus species	3(20.00%)	1(7.14%)	1(8.33%)	0(0.00%)	5(9.43%)
A.flavus	2(13.33%)	1(7.14%)	1(8.33%)	0(0.00%)	4(7.55%)
A.fumigatus	1(6.67%)	0(0.00%)	0(0.00%)	0(0.00%)	1(1.89%)
Dematiaceae	1(6.67%)	0(0.00%)	0(0.00%)	0(0.00%)	1(1.89%)
Paecilomyces lilacinus	0(0.00%)	1(7.14%)	1(8.33%)	0(0.00%)	2(3.77%)
Bipolaris	1(6.67%)	0(0.00%)	0(0.00%)	0(0.00%)	1(1.89%)
Crescent-shaped fungi	0(0.00%)	0(0.00%)	1(8.33%)	0(0.00%)	1(1.89%)
Curvularia species	0(0.00%)	0(0.00%)	0(0.00%)	2(16.67%)	2(3.77%)
Negative fungal culture	1(6.67%)	2(14.29%)	3(25.00%)	5(41.67%)	11(20.75%)

Corneal epithelial staining area and improvement ratio

All groups showed significant reduction in corneal epithelial staining area after 7 days of treatment (t/Z = -3.408, 9.576, 6.682, 5.606, all P < 0.05). The improvement ratios were: A1 group 37.19%, A2 group 36.11%, B1 group 14.95%, and B2 group 41.30%. In superficial stromal groups, no significant difference existed between A1 (37.19%) and A2 (36.11%) groups (Z = -0.228, P = 0.838). In deep stromal groups, B2 group (41.30%) showed significantly greater improvement than B1 group (14.95%) (t = 6.249, P < 0.0001) (Table 4; Fig. 3).

Table 4.

Comparison of corneal epithelial staining area before and after 7 days of treatment (Median, M (P₂₅, P₇₅))

Groups	Pre-treatment(mm²)	7-day post-treatment(mm²)	t/Z	P
A1(n = 15)	17.02(9.27,22.55)	9.87(5.68,13.95)	-3.408	< 0.0001
A2(n = 14)	24.69(17.85,30.40)	16.03(11.29,18.91)	9.576	< 0.0001
B1(n = 12)	22.52(17.54,28.23)	19.37(14.38,23.72)	6.682	< 0.0001
B2(n = 12)	24.69(12.80,33.18)	14.07(6.70,16.51)	5.606	0.0002

Visual acuity outcomes

Pre-treatment visual acuity showed no significant differences between groups (P > 0.05). At final follow-up, all groups demonstrated significant BCVA improvement compared to baseline (P < 0.05). After adjusting for pre-treatment BCVA, final BCVA in superficial stromal groups was A2: 1.1 (0.8–1.7) LogMAR vs. A1: 0.7 (0.3–1.6) LogMAR, with no significant difference (t = -0.810, P₁ = 0.425). In deep stromal groups, final BCVA was B2: 1.0 (0.7–1.9) LogMAR vs. B1: 1.0 (0.7–1.4) LogMAR, also showing no significant difference (t = -0.131, P₂ = 0.897) (Table 5).

Table 5.

Comparison of pre-treatment and final BCVA among four groups (Median, M(P25,P75), LogMAR)

Items	A1 Group (n = 15)	A2 Group (n = 14)	t/Z	P₁*	B1 Group (n = 12)	B2 Group (n = 12)	/Z	Pz*
Pre-treatment visual acuity	2.0(0.5,2.4)	2.0(0.8,2.1)	-0.403	0.687	2.0(1.1,2.4)	2.4(2.0,2.4)	-1.148	0.251
Pre-treatment BCVA	2.0(0.5,2.4)	2.0(0.8,2.1)	-0.403	0.687	2.0(1.1,2.4)	2.4(2.0,2.4)	-1.148	0.251
Final BCVA	0.7(0.3,1.6)	1.1(0.8,1.7)	-0.810	0.425	1.0(0.7,1.4)	1.0(0.7,1.9)	-0.131	0.897

*P¹, P₂: Statistical results comparing pre-treatment visual acuity, pre-treatment BCVA, and final BCVA between A1 vs. A2 groups and B1 vs. B2 groups, respectively

Epithelial healing time and clinical cure time

One patient in group B1 underwent enucleation due to treatment failure and was excluded from analysis, leaving 11 cases in the B1 group. Epithelial healing time in superficial stromal groups was A1: 33.0 (16.0–42.0) days vs. A2: 34.0 (26.8–61.3) days, showing no significant difference (Z = -0.655, P₁ = 0.512). In deep stromal groups, healing time was B1: 47.0 (32.8–69.3) days vs. B2: 46.0 (28.5–65.8) days, also without significant difference (t = 0.379, P₂ = 0.708). Clinical cure time in superficial stromal groups was A1: 42.0 (25.0–49.0) days vs. A2: 41.5 (32.8–69.3) days (Z = -0.480, P₁ = 0.631). In deep stromal groups, cure time was B1: 58.5 (43.3–70.5) days vs. B2: 51.5 (35.0-70.3) days (t = 0.699, P₂ = 0.493) (Table 6).

Table 6.

Comparison of epithelial healing time and clinical cure time among four groups(Median, M(P₂5,P75),days)

Items	A1 Group (n = 15)	A2 Group (n = 14)	Z	P₁*	B1 Group (n = 11)	B2 Group (n = 12)	t	P₂*
Epithelial healing time (days)	33.0 (16.0, 42.0)	34.0 (26.8, 61.3)	−0.655	0.512	47.0 (32.8, 69.3)	46.0 (28.5, 65.8)	0.379	0.708
Clinical cure time (days)	42.0 (25.0, 49.0)	41.5 (32.8, 69.3)	−0.480	0.631	58.5 (43.3, 70.5)	51.5 (35.0, 70.3)	0.699	0.493

Note: One case with enucleation in the B1 group was excluded from Table 6 statistics; final B1 group n = 11

P¹, P₂: Statistical results comparing epithelial healing time and clinical cure time between A1 vs. A2 groups and B1 vs. B2 groups, respectively

Final clinical outcomes

Clinical outcomes were classified as clinical cure or treatment failure, with clinical cure including medical cure and combined medical-surgical cure. Final follow-up showed clinical cure in 52 cases (98.1%), including medical cure in 42 cases (79.2%) and combined medical-surgical cure in 10 cases (18.9%); treatment failure occurred in 1 case (1.9%). No serious complications such as drug-induced toxic keratitis occurred in any group. No significant differences in final clinical outcomes existed among groups (χ² = 5.569, P = 0.555). Between superficial stromal groups A1 and A2, and between deep stromal groups B1 and B2, no significant differences were observed (χ² = 0.306, 2.386; P₁ = 0.580, P₂ = 1.000) (Table 7).

Table 7.

Treatment outcomes and surgical intervention rates among different treatment groups

Groups	Medical cure	Medical treatment failure requiring surgical intervention		Enucleation	Medical cure rate %
		Conjunctival flap coverage	Corneal transplantation
A1 Group (n = 15)	13(86.67%)	2(13.33%)	0	0	13(86.67%)
A2 Group (n = 14)	10(71.43%)	4(28.57%)	0	0	10(71.43%)
B1 Group (n = 12)	10(83.33%)	1(8.33%)	0	1(8.33%)	10(83.33%)
B2 Group (n = 12)	9(75.00%)	2(16.67%)	1(8.33%)	0	9(75.00%)
x²					5.569

Disccussion

Fungal keratitis (FK) represents one of the leading causes of corneal blindness worldwide, particularly prevalent in tropical and subtropical regions. It is characterized by fungal invasion of corneal tissue, which triggers severe inflammatory responses and may progress to corneal perforation, scar formation, and ultimately blindness. The management of FK remains a significant challenge in ophthalmology. Lesions frequently infiltrate deep corneal layers; however, the cornea’s unique anatomical barrier properties and lipophilic-hydrophilic characteristics hinder the attainment of effective drug concentrations at lesion sites when antifungals are administered topically or systemically [16]. Against this backdrop, intrastromal injection of antifungal agents has emerged as a novel therapeutic modality that enables direct drug delivery to deep lesion sites, representing a promising breakthrough. This approach achieves rapid attainment of high local drug concentrations through direct intrastromal injection, thereby effectively inhibiting fungal proliferation or inducing fungal cell death. It offers clinicians novel therapeutic options and renewed hope for preserving visual function in affected patients [17].

Intrastromal voriconazole injection has recently emerged as a research focus for fungal keratitis (FK) treatment, though its therapeutic efficacy remains controversial. Jain et al. [16] reported a case of fungal infection at a phacoemulsification incision site following cataract surgery that was successfully managed with intrastromal voriconazole injection. Similarly, Sharma et al. [18] demonstrated the efficacy of intrastromal injection in treating deep FK, showing no significant tissue toxicity in both experimental and clinical settings. The primary advantage of this approach is the accumulation of high drug concentrations at infection sites, reflecting pharmacokinetic profiles that are challenging to attain via other local or systemic routes [17]. However, some scholars hold divergent views, suggesting that even voriconazole—despite its superior corneal penetration [19]—may exhibit limited efficacy when used exclusively via intrastromal injection [20]. Ganapathy Kalaiselvi et al. [21] confirmed in multiple studies that intrastromal voriconazole significantly elevates local corneal drug concentrations. However, clinical outcomes vary across studies, indicating that factors influencing efficacy are likely multifactorial, and no consensus has been established.

Potential reasons for the observed efficacy variations in intrastromal voriconazole injection for FK may encompass multiple factors. First, disease stage and infection severity vary among FK cases: superficial infections tend to respond better to conventional antifungal therapy, whereas deep infections often necessitate localized high-concentration drug delivery. Second, fungal species and their susceptibility profiles differ, with variable and inconsistent voriconazole susceptibility across pathogens. Third, drug distribution and metabolic rates within the corneal stroma may be influenced by factors such as corneal thickness, local tissue responses, inflammatory microenvironments, and other variables. Additionally, inter-study variations in intrastromal injection techniques, dosage regimens, drug concentrations, and follow-up durations may further complicate result interpretation. Previous literature has rarely employed stratified analyses by lesion depth, compromising the comparability of conclusions across studies.

Currently, no consensus exists on the optimal voriconazole concentration for intrastromal injection in fungal keratitis (FK) management. Li Suxia et al. [22] employed a 10 mg/mL concentration with a three-month follow-up period, reporting no significant reduction in corneal endothelial cell density or fungal recurrence, and noting visual improvement in 74.7% of patients. Lu Jiang et al. [12] utilized a 5 mg/mL concentration, finding no significant difference in adverse reaction rates between the treatment group (7.14%) and control group (2.38%), without compromising safety. Bai Liguang et al. [7] demonstrated that irrespective of voriconazole concentration (high or low), intrastromal injection achieved higher cure rates than conventional treatment, but higher concentrations conferred no additional efficacy and potentially increased the risk of adverse reactions. Therefore, to balance efficacy and safety, this study selected a 0.5% (5 mg/mL) concentration for intrastromal injection. Consistent with existing pharmacokinetic studies, voriconazole does not persist long-term in the corneal stroma but is rapidly absorbed and metabolized via osmotic diffusion [23]. Mechanistically, intrastromal injection enhances local corneal drug concentrations, optimizes drug utilization, and sustains effective antifungal activity within lesions over an extended period, thereby accelerating ulcer healing and reducing disease duration. However, this approach is not without risks. Some studies have reported that intrastromal injection may elevate the risk of corneal perforation or result in uneven drug distribution, compromising therapeutic efficacy [24]. These issues may stem from improper control of injection depth, inconsistent dosing, technical variability, and varying degrees of underlying corneal pathology. In the present study, no serious complications—including perforation or toxic keratitis—were observed, suggesting that with standardized procedures and appropriate dosing, intrastromal injection represents a relatively safe therapeutic modality. Regarding the safety profile of intrastromal voriconazole injection, the potential adverse effects reported in the literature include transient corneal edema at the injection site, needle track opacity, localized epithelial defects, and, in rare cases, corneal thinning or perforation associated with improper injection depth or excessive volume. In the present cohort, no cases of drug-induced toxic keratitis, corneal perforation, secondary infection at the injection site, or persistent corneal decompensation were observed. Transient corneal edema surrounding the injection site was noted in all patients who received intrastromal injection, which resolved spontaneously within 24–48 h without additional intervention. These findings suggest that intrastromal injection of 0.5% voriconazole at a volume of 0.1–0.3 mL, performed using the three-zone technique with standardized procedures, has an acceptable short-term safety profile. However, the long-term safety implications, including potential effects on corneal endothelial cell density and corneal biomechanics, remain to be evaluated in future prospective studies with extended follow-up.

Regarding efficacy, this study demonstrated a significant reduction in ulcer area across all four groups after 7 days of treatment (P < 0.05), indicating that conventional antifungal monotherapy already exhibits appreciable efficacy. In the superficial stromal groups (A1: monotherapy; A2: combination therapy with intrastromal injection), no significant difference in efficacy was observed between the two treatment arms (P > 0.05), suggesting that superficial infections do not require intrastromal injection to achieve satisfactory outcomes. Conversely, in deep stromal patients (B1: monotherapy; B2: combination therapy), the combination group (B2) showed significantly greater improvement in ulcer area at 7 days compared to the monotherapy group (B1) (P < 0.05). This suggests that intrastromal voriconazole injection may facilitate early lesion control in deep infections, although this early improvement should be interpreted cautiously as it reflects short-term disease suppression rather than definitive therapeutic superiority, consistent with the findings of Wu Siteng et al. [25]. However, in terms of epithelial healing time and final best-corrected visual acuity (BCVA) improvement, no significant differences were observed between treatment groups in either superficial or deep stromal cohorts (all P > 0.05). These results align with studies by Niki et al. [23], Sharma and Chacko [26], and Lukman Edwar and Yulia M. Janna [27], all of whom reported limited differences in final efficacy between single intrastromal voriconazole injection and topical therapy. Possible explanations for this phenomenon include the following: First, the conventional antifungal treatment protocol in this study was relatively standardized, and combination therapy with multiple agents already exhibited robust efficacy, thereby attenuating the advantages of intrastromal injection. Second, effective voriconazole concentrations in the corneal stroma may decrease more rapidly than expected, resulting in a shorter-than-anticipated duration of sustained action. Third, heterogeneous distribution of the drug solution from a single injection within the cornea—coupled with possible leakage to epithelial layers—may reduce local effective concentrations. Fourth, the relatively small sample size in this study may compromise statistical power. Notably, some studies indicate that increasing the frequency of injections may reduce treatment duration and enhance healing rates [11]. Therefore, future multicenter, large-scale, randomized controlled trials should further explore optimal combinations of injection frequency, dosing regimens, and drug concentrations. This study further suggests that the potential utility of intrastromal voriconazole injection in mitigating corneal scarring and improving final visual outcomes warrants further validation. Fungal-derived aggregation factors can inhibit host cell proliferation and compromise corneal epithelial barrier integrity, impairing wound healing [28]. Early elevation of antifungal drug concentrations in the corneal stroma may rapidly eradicate the fungal burden, thereby reducing inflammation and promoting tissue repair. However, this study did not observe a significant advantage of intrastromal injection on final best-corrected visual acuity (BCVA), suggesting that a single injection might be inadequate to alter long-term structural remodeling.

The innovation of this study lies in the utilization of anterior segment optical coherence tomography (AS-OCT) for quantitative lesion grading. By evaluating the therapeutic responses of FK patients with distinct lesion depths to intrastromal voriconazole injection, this work aims to establish a foundation for individualized treatment strategies. In this study, of the 53 enrolled patients, 29 (54.72%) had lesions confined to ≤ 50% of corneal thickness (categorized as superficial-intermediate FK), while 24 (45.28%) exhibited lesions involving > 50% of corneal thickness (categorized as deep FK). As a non-contact imaging modality, anterior segment optical coherence tomography (AS-OCT) enables precise measurement of total corneal thickness and the depth of the deepest lesion site [29]. For patient grouping, we first performed preliminary assessment of lesion depth and extent using slit-lamp microscopy, followed by confirmatory analysis with AS-OCT to ensure accurate classification. This dual-modality approach not only minimizes subjective bias inherent in conventional slit-lamp-based detection but also provides objective evidence for assessing disease severity and monitoring therapeutic efficacy.

Limitations

Several limitations should be acknowledged. First, this was a retrospective, non-randomized study. Because treatment allocation was based on patient preference rather than randomization, selection bias and residual confounding cannot be fully excluded, even though baseline characteristics were comparable between groups. Patients’ preferences may have been influenced by factors such as economic status, acceptance of invasive procedures, and individual risk perceptions, which could potentially affect causal inference. Second, the relatively small sample size in each subgroup, particularly in the deep stromal group (n = 12 per treatment arm), may limit statistical power and increase the risk of type II error; thus, non-significant findings should be interpreted cautiously. Third, as a single-center study conducted at the People’s Hospital of Guangxi Zhuang Autonomous Region, generalizability may be limited given regional variations in fungal epidemiology, patient presentation timing, and clinical practice patterns. Fourth, we did not perform species-specific efficacy analyses, precluding conclusions about differential treatment responses among fungal species with varying antifungal sensitivities. Fifth, we evaluated only a single intrastromal injection protocol; repeated dosing strategies, which may provide sustained drug concentrations for deep or refractory cases, warrant investigation in future prospective studies. Sixth, the 7-day post-treatment assessment was selected as an early efficacy evaluation time point because most patients were still hospitalized at this stage, enabling standardized clinical examination and data collection; however, after discharge, a substantial proportion of patients—particularly those from remote rural areas in Guangxi—were lost to follow-up, limiting the feasibility of establishing multiple intermediate follow-up time points. Future prospective studies should implement structured follow-up protocols with multiple time points (e.g., 1 month, 3 months, and 6 months) to comprehensively evaluate long-term outcomes including recurrence rate, corneal scar formation, and long-term visual prognosis.

Conclusion

In summary, our findings indicate the following: First, for superficial stromal fungal keratitis, conventional antifungal therapy yields favorable outcomes, and intrastromal voriconazole injection provides no additional therapeutic benefit; therefore, it is not recommended as a routine intervention to avoid unnecessary iatrogenic trauma and economic burden.

Second, for deep stromal fungal keratitis, intrastromal injection accelerates short-term ulcer area reduction (assessed by day-7 corneal epithelial staining/ulcer area as an early marker of lesion control); however, this early improvement did not translate into shorter cure time or better final visual outcomes, suggesting that early response reflects partial disease control rather than definitive long-term benefit.

Third, these results represent the incremental effect of intrastromal voriconazole as an adjunct to intensive combination antifungal therapy; the addition of intrastromal injection did not significantly improve overall treatment outcomes in this cohort, and these findings should not be extrapolated to monotherapy strategies or resource-limited settings where standard regimens cannot be fully implemented.

Abbreviations

AS-OCT

Anterior Segment Optical Coherence Tomography

BCVA

Best Corrected Visual Acuity

FK

Fungal Keratitis

IVCM

In Vivo Confocal Microscopy

ANOVA

Analysis of Variance

Author contributions

H.H, C.M. and Z.Z.wrote the main manuscript text.L.J, N.S. and F.L.data analysis.G.C, Y.L, B.C.and Y.l.acquisition data.F.T. and C.S.interpretation of data.M.L.substantively revised the work.L.C. and F.X.Research project design.Q.C, H.Z. and Q.L.made substantial contributions to the conception of the work.Hui Huang, Chaoting MO and Zhou Zhou have contributed equally to share first authorship.

Funding

The study was supported by Natural Science Foundation of Guangxi Zhuang Autonomous Region (No.2025GXNSFDA069056, No.2023GXNSFAA026074), National Natural Science Foundation of China (No.82560202), Scientific Research Project of Guangxi Health Commission (S2021038, S2020077, Z-A20250074,ZA20230133), and Guangxi Clinical Ophthalmic Research Center (No.AD19245193), The sponsor or funding organization had no role in the design or conduct of this research.

Data availability

The data sets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethical approval and consent to participate

This retrospective study was approved by the Institutional Review Board of the People’s Hospital of Guangxi Zhuang Autonomous Region (KY-IIT-202458), and written informed consent was obtained from all subjects participating in the study. The study was conducted in accordance with the Declaration of Helsinki and approved by the hospital ethics committee.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.