Comparative efficacy of intrastromal amphotericin B versus voriconazole in treating deep-stromal refractory fungal keratitis

Abstract

Purpose:

To compare the outcome of adjuvant intrastromal Amphotericin B (ISAMPB) and Voriconazole (ISVCZ) in managing refractory deep stromal fungal keratitis.

Methods:

We reviewed the records of patients treated between January 2020 and December 2022 for microbiologically confirmed refractory fungal keratitis that had worsened despite treatment with topical Natamycin, with or without oral Ketoconazole. These patients received adjuvant therapy with either intrastromal Amphotericin B (ISAMPB, 5–10 µg/0.1 ml) or intrastromal Voriconazole (ISVCZ, 50 µg/0.1 ml). The demographics, clinical profiles, and outcomes of the ISAMPB and ISVCZ groups were compared, with key outcomes being the proportion of patients achieving clinical resolution and the number of therapeutic keratoplasty procedures performed. A logistic regression model was constructed for multivariate analysis to adjust for potential confounders.

Results:

Medical records of 49 patients with refractory fungal keratitis were analyzed: 34 (69.3%) received ISAMPB, and 15 (30.6%) received ISVCZ. The overall mean LogMAR visual acuity was 1.90 ± 0.6, and the infiltrate size averaged 4.28 ± 1.30 mm vertically and 4.3 ± 1.4 mm horizontally. Aspergillus and Fusarium each accounted for 32.3% of ISAMPB patients, and 40% and 20% of ISVCZ patients, respectively. The median time to intrastromal injection was 12 days for ISAMPB and 11 days for ISVCZ. ISAMPB achieved resolution in 53% of cases, while 41.1% required TPK; ISVCZ resolved in 10%, with 86% requiring TPK (P < 0.01). The adjusted odds ratio for ISAMPB response was 21.98 (P = 0.013).

Conclusion:

Intrastromal Amphotericin B has been shown to provide better outcomes than Voriconazole in the treatment of deep stromal refractory fungal keratitis.

Fungal keratitis is one of the most important causes of microbial keratitis,[1] especially in tropical countries,[2] where it accounts for nearly 50 percent of the cases. In its mild stage, while the disease is still limited to the anterior stroma, it can be managed in 60% of cases[3] with commercially available topical Natamycin[4,5] alone, which is the first line of antifungals. But due to missed diagnosis,[6] use of topical corticosteroids,[7] associated systemic diseases such as uncontrolled diabetes mellitus,[8] along with the presence of virulent fungal species,[3,9] nearly 1/3^rd of the cases present with an advanced form of keratitis along with extensive posterior stromal involvement. At this stage, combined topical Natamycin and systemic Azoles[5,10,11] are recommended due to the limited penetration of Natamycin in the deeper corneal layers. Various medical and surgical modalities have been described to handle deep stromal infection, such as newer formulations of Natamycin (water-soluble Natasol 1%),[12] early lamellar keratoplasty,[13] epithelial debridement, or intrastromal injections of antifungals.[14,15,16] But despite these urgent actions, many of these cases still need urgent therapeutic penetrating keratoplasty (TPK) to eradicate infection with overall poor outcomes.[17] To avoid TPK, timely intrastromal antifungal injections have been explored as targeted drug delivery. While studies exist on intrastromal Voriconazole[18,19,20,21] and Amphotericin B,[22,23,24,25,26] their outcomes have not been compared. This paper compares the effectiveness of Amphotericin B and Voriconazole intrastromal injections in managing deep stromal fungal keratitis unresponsive to medical therapy.

Methods

Medical records of microbiologically confirmed fungal keratitis cases from January 2020 to September 2022 were reviewed, with approval obtained from the Institutional Review Board (LEC-BHR-R-09-24-1240). The study followed the principles of the Declaration of Helsinki. We have included only cases of microbiologically confirmed fungal keratitis and worsening on topical and oral antifungals with a minimum of 1-week duration. We have excluded cases with severe corneal involvement and extensive limbal/scleral and posterior segment involvement involvement. All cases initially received topical Natamycin 5% ophthalmic suspension, with or without oral ketoconazole 200 mg twice daily. Patients whose fungal keratitis did not improve or worsen despite medical therapy for a week and subsequently received intrastromal Amphotericin B (ISAMPB, 5–10 µg/0.1 ml) or Voriconazole (ISVCZ, 50 µg/0.1 ml) injections were included in the study. Although the decision to administer intrastromal injections was based on clinical signs of worsening, such as deeper stromal involvement on Natamycin, the choice of drug was not based on the the antifungal susceptibility analysis, but based on the individual physician’s preference, as per their experience of managing fungal keratitis (ranging 5–15 years). All the physicians involved in the decision making were experienced to manage fungal keratitis.

Intrastromal Amphotericin B is prepared by dissolving 50 mg of Amphotericin B powder (Bharat Serum Vaccine Limited, India) in 10 ml of 5% dextrose, resulting in a concentration of 5 mg/ml (or 500 µg per 0.1 ml). This solution is then further diluted with 9.9 ml of 5% dextrose in a syringe, producing a final concentration of 500 µg/10 ml, or 50 µg/ml, five µg per 0.1 ml. The dose of intrastromal Voriconazole is 50 µg per 0.1 ml. To prepare it, take 1 mg of available Voriconazole powder (Aurolab, India) and reconstitute with 2 ml of distilled water to achieve 50microgram/0.1 ml concentration. All drug reconstitution was performed by operating ophthalmologists in the operating theater (OT) following aseptic measures. All intrastromal injections were administered under topical or peribulbar anesthesia, using an operating microscope in a supine position under sterile conditions in OT. The decision of the type of anesthesia (local/peribulbar) was based upon the patient’s cooperation and stromal thinning (preferring topical in cooperative patients in the absence of stromal thinning and vice versa). A 29G needle was used to deposit the drug in the mid-stromal level near the infiltration area, creating a 360° barrage of the infected stroma, with corneal hydration (0.05 ml) guiding the coverage area.

The demographics, clinical features, microbiological profiles, treatment approaches (medical and surgical), and outcomes were analyzed and compared between the two groups. Ophthalmic examinations at each follow-up included corrected visual acuity and slit-lamp biomicroscopy to document ulcer characteristics such as infiltrate diameter, depth of involvement, and presence of hypopyon. Clinical signs were compared between the initial presentation and the point at which intrastromal injections were administered due to worsening infection. For microbiological evaluation, corneal scrapings underwent direct microscopic examination (for Gram’s stain, 10% potassium hydroxide, and 1% calcofluor white) and were plated on culture media (blood agar, chocolate agar, Sabouraud dextrose agar, potato dextrose agar, and non-nutrient agar with Escherichia coli overlay) according to the institutional protocol.[27]

Statistical methods

The features were compared using Fischer’s exact test and Chi-square test (for categorical variables), Wilcoxon Rank sum test, and t-test (for continuous variables). Univariate logistic regression was used to assess the association between each independent variable and the binary outcome of treatment response. Crude odds ratios (ORs), 95% confidence intervals (CIs), and P values were calculated using Wald’s test.

A logistic regression model was constructed for multivariate analysis to adjust for potential confounders. Model selection was performed using the stepwise Akaike Information Criterion (AIC) method via the stepAIC function from the MASS package in R. This approach selected the most appropriate variables for the final model by minimizing the AIC. Adjusted odds ratios (adj. ORs) and 95% confidence intervals were reported, and statistical significance was determined using Wald’s test with a P value threshold of < 0.05.

Results

Medical records of 49 patients with refractory fungal keratitis were analyzed. The median age of the patients was 47 years (IQR, 38–58) with a male-to-female ratio of 30 (56.7%) to 19 (35.8%). Clinical and demographical profiles of patients from both the groups, N = 34 (69.3%) in ISAMPB versus N = 15 (30.6%) in ISVCZ, at the time of presentation to the clinic are compared in Table 1. The median visual acuity (LogMAR) was 2.3 (IQR, 1.8, 2.3) versus 1.8 (IQR, 1.4, 2.3), and the size of corneal infiltrate in both vertical and horizontal diameters was 4.1 ± 1.2 and 4.3 ± 1.4 versus 4.8 ± 1.6 and 4.3 ± 1.4 mm. Among both the groups, posterior stroma was involved in 70.5% versus 60% of cases at the time of intrastromal injections.

Table 1.

Demographics, clinical characteristics, and outcomes of two groups with refractory fungal keratitis treated with intrastromal Amphotericin B and Voriconazole

Clinical Parameters		Group A IS AmpB (n=34)		Group B IS VCZ (n=15)		P
Age (Mean±Std)		49 (± 11.3)		47.7 (± 15.6)
Male: Female ratio		22:12		10:5		0.60
Duration of symptoms (Days)		14.5 (7-20)		20 (7-25)		0.82
Vision at presentation (LogMAR±Std)		2.3 (1.8,2.3)		1.8 (1.4,2.3)		0.34
Duration of Treatment (Days)		4 (3-16.5)		4 (2-9)		0.33
Type of medications at presentation
1. Topical Natamycin 5%		23 (67.6%)		6 (40%)		0.07
2. Topical Natamycin 5% + Oral Ketoconazole 200 mg BD		3 (9%)		3 (20%)		0.35
3. Topical Voriconazole 1%		6 (17.6%)		4 (26%)		0.47
5. Steroids		4 (26%)		0		0.30
6. Topical Moxifloxacin or Topical Gatifloxacin		22 (64.7%)		10/15 (66.7%)		0.89
Clinical features at the presentation
• Size of infiltrate (mm, ± SD)
1. Vertical, mm, ± SD		4.1 (1.2)		4.8 (1.6)		0.09
2. Horizontal, mm, ± SD		4.3 (1.4)		4.3 (1.4)		0.92
• Location of the ulcer
a. Central (<3 mm)		28 (82.3%)		12/15 (80%)		1
b. 3<6 mm		6 (17.6%)		2 (13.3%)		1
c. >6 mm		1 (2%)		1 (6.4%)		0.5
• Depth of the infiltrate (based on slit-lamp examination)
a. Anterior stromal		17 (50%)		9 (60%)		0.52
b. Mid Stromal		13 (38.2%)		8 (53.3%)		0.325
c. Posterior stromal		14 (41.1%)		8 (60%)		0.5
• Presence of plaque		11 (32.3%)		8 (27.5%)		0.29
• Hypopyon (mm)		31 (91.1%)		13 (86.6%)		1
• Satellite lesions		8 (23.4%)		1		0.65
• Endoexudates		9 (26.4%)		6 (40%)		0.74
Clinical features before intrastromal injections
Median size of the infiltrate
1. Vertical (mm)		4.46 (3.5-5)		4 (3.1-5.0)		0.42
2. Horizontal (mm)		4 (3.9-5.25)		4 (3.9-6.0)		0.06
Depth of involvement
a. Anterior stromal		6 (17.6%)		9 (60%)		0.01
b. Mid-stromal		15 (44.1%)		8 (53.3%)		0.55
c. Posterior stromal		24 (70.5%)		8 (60%)		1
Presence of Endoexudates n (%)		9 (26.5)		6 (40)		0.73
Microbiological profile
1. Aspergillus		11 (32.3%)		6 (40%)		0.60
2. Fusarium		11 (32.3%)		3 (20%)		0.50
3. Other fungus		3		3		1
4. No growth		7 (23.5%)		2 (7%)		0.5
5. Bacteria		5 (14.7%)		2 (13.3%)		1
Duration of injection (Days)		12 (8-19.75)		11 (8-19.5)		0.87
Procedures along with Intrastromal Injections
1. Tissue adhesives		8 (23.5%)		6 (40%)		0.99
2. Intracameral		7 (20.5%)		12 (8%)		0.053
3. Intravitreal		1 (2%)		1 (6.6%)		1
4. Anterior chamber wash		4 (11.7%)		7 (46.6%)		0.37
Outcome
Resolved without TPK		18 (53%)		2 (10%)		0.009
Final visual acuity		1.85 (0.9-1.5)		1.8 (1-2.3)		0.25
Therapeutic penetrating keratoplasty		14 (41.1%)		13 (86%)		<0.01
Lost to Follow-up		2		0

On microbiological evaluation, as described in Table 2, while all cases had fungal filaments on smear examination by direct microscopy (grams and/or CFW + KOH stain), fungus grew in culture in 37 (75.5%) cases. Bacterial growth was observed in 7 patients, 5/34 (14.7%) and 2/15 (13.3%) in ISAMPB and ISVCZ groups, respectively. Among the fungi grown, Aspergillus and Fusarium species were found in 11 equal patients (32.3%) in the ISAMPB group, compared to 6 out of 15 patients (40%) and 3 patients (20%) in the ISVCZ group.

Table 2.

Microbiological profile of patients with fungal keratitis

Species		Fungus on culture
Aspergillus		17
Fusarium		14
Acremonium		2
Verticillium		1
Exserohilum Rostratum		1
Scedospermium Apiospermum		1
Chaetomiun		1
Total fungus		37
Species		Bacteria on culture
Rothia dentocariosa, Pseudomonas Burkholderia Cepacia E. coli Staph epidermidis Stenotrophomonas Staph haemolyticus		7

The median interval between the time of presentation and intrastromal injection was 12 (8–19.75) versus 11 (8–19.5) days. While in 34 patients of the ISAMPB group, 18 (53%) resolved, 14 (41.1%) needed TPK, and 2 were lost to follow-up; in 15 patients of ISVCZ group, 2 (13.3%) resolved, and 13 (86%) needed TPK, P = 0.009. We have done logistic regression [Table 3] to assess the factors responsible for worsening. A similar observation was found while adjusting with other factors where intrastromal Voriconazole is associated with poor outcome, adjusted OR = 21.98 (1.94,249.61). The presence of Aspergillus microbiologically, adjusted OR = 5.5 (CI, 0.69,44.14), and hypopyon, 12.63 (CI, 0.84,189.83), was found to be associated with poorer response. However, as mentioned in Table 1, both factors are present in both groups in similar proportions.

Table 3.

Univariate and multivariate logistic regression analysis, highlighting factors associated with the resolution of fungal corneal ulcers, with intrastromal Voriconazole showing a poorer response compared to intrastromal Amphotericin B

Factors for Logistic regression predicting response: Worsened vs Resolved infection
		Univariate Analysis				Multivariate Analysis
		Crude OR (95%CI)		Crude P		Adj. OR (95%CI)		P (Wald’s test)
Duration of Treatment before presentation		1.0055 (0.9602,1.053)		0.815
Use of prior Steroids		0.2 (0.02,2.11)		0.181
Visual Acuity (logMAR)		0.87 (0.33,2.28)		0.779
Vertical diameter of infiltrate		0.9914 (0.6525,1.5063)		0.968
Horizontal diameter of infiltrate		1.03 (0.68,1.56)		0.896
Hypopyon		3.37 (0.55,20.55)		0.187		12.63 (0.84,189.83)		0.067
Endoexudates		0.39 (0.11,1.39)		0.147		0.1 (0.01,1.03)		0.053
Deep Stromal involvement		0.29 (0.07,1.22)		0.091		0.15 (0.02,1.1)		0.061
Aspergillus keratitis		2.12 (0.6,7.41)		0.241		5.5 (0.69,44.14)		0.108
Fusarium keratitis		1.35 (0.37,4.86)		0.646		7.84 (0.92,66.53)		0.059
Intracameral along with intrastromal injections		4.6 (1.1,19.22)		0.036		7.32 (1.08,49.66)		0.042
ISvcZ vs ISampB		7.31 (1.43,37.47)		0.017		21.98 (1.94,249.61)		0.013

Additional procedures (tissue adhesives, intracameral antibiotics, intravitreal, or AC wash) were also performed along with intrastromal injections in 8 (23.5%), 7 (20.5%), 1 (2%), and 4 (11.7%) patients in Ampho B group versus 6 (40%), 12 (80), 1 (6.6%), and 7 (46.6%) patients in voriconazole group. The indications of TPK were either corneal perforation or extensive thinning or extension of infiltrate posteriorly. Among the patients (n = 9, 26.4%) with repeat ISAMPB, 5 cases responded, while the remaining 4 required therapeutic penetrating keratoplasty (TPK). Only one patient received repeat ISVCZ, but it worsened. In the ISVCZ group, out of 13 patients who underwent TPK, 10 (76.9%) buttons had fungus growth from the excised corneal button and evidence of fungal infection on histopathological examination. In 16 cases of the ISAMPB group, 7 (43.7%) fungi grew from the button, and 9 (56.2%) had histopathological evidence. None of the cases had recurrence after resolution during the follow-up period. Fig. 1 shows the clinical picture of patients at the time of worsening, before intrastromal injection (A-E), and at the time of resolution (F-I).

Figure 1.

The figure depicts the clinical profile of fungal keratitis. Images a to e represent patients with active keratitis (a and d-Aspergillus flavus, b- Cladosporium species, and c, e- No growth on culture), while images f to j show the clinical appearance at the final visit following intrastromal amphotericin B injections, demonstrating complete resolution

Discussion

This study compared the outcomes of patients with refractory fungal keratitis treated with either intrastromal Amphotericin B (ISAMPB) or Voriconazole (ISVCZ). Both groups were similar in their clinical and demographic profiles at presentation. However, a significant difference in outcomes was observed between the groups. Urgent therapeutic penetrating keratoplasty (TPK) was avoided in 50% of ISAMPB cases, compared to only 13.3% in the ISVCZ group. Despite concerns about potential corneal toxicity with ISAMPB, our findings suggest better results with Amphotericin B. A study by Qu et al.[26] examined the pharmacokinetics of ISAMPB in a rabbit model and found that concentrations of 5–10 µg/0.1 ml were nontoxic, while 20 µg/0.1 ml caused corneal edema, epithelial erosion, and severe neovascularization. As reported in the comparative trial by Saluja et al.[28] as well, there was more incidence of deep stromal vasculrization with the use of intrastromal amphotericin B. As per standard practice, we used 5–10 µg/0.1 ml ISAMPB, and none of our patients showed corneal toxicity, except for deep vascularization, which regressed over time. This can be due to lesser number of injections as compared to the previous studies.

ISAMPB’s efficacy was further supported by a lower proportion of fungal growth in the corneal button, with 43.7% in the ISAMPB group compared to 76.9% in the ISVCZ group. Similarly, Narayana et al.[29] found no additional benefit with intrastromal Voriconazole and a 2.85-fold increase in corneal perforation. We also accounted for other confounding factors, such as species variation and clinical features, but found no significant impact on the outcomes. Even the time of injections in both the groups in our study was found to be similar, 12 (8-19.75) versus 11 (8-19.5) days.

In 10 cases (9 ISAMPB, 1 ISVCZ), intrastromal injections were repeated, with 6 responding and 4 requiring TPK. This suggests that repeated injections can improve outcomes in cases with a suboptimal initial response. Sharma et al.[10,19] recommended intrastromal injections 3–4 times every 72 hours for patients not responding to topical therapy. Guber et al.[30] reported success with repeated intrastromal Voriconazole (100 mg/1 ml) every 2–3 days, while Aydin et al.[25] showed an 87% resolution rate in 32 fungal keratitis patients after repeated injections.

In addition to intrastromal injections, the intracameral route has been explored. While Hu et al.[24] and Kaushik et al.[31] showed improved outcomes with intracameral Amphotericin B. Sharma et al.[27] reported no significant difference. In our study, only 3 out of 7 patients who received intracameral Amphotericin B responded, while the rest required TPK. Similarly, intracameral Voriconazole did not prevent TPK in any of the 8 patients treated.

Limitations of this study include its retrospective design and the absence of antifungal susceptibility data. Although the sample size was small, we observed a significant difference in outcomes between the two groups, and based upon the observations, we plan for a prospective study with a larger sample size. The superior effect of amphotericin B in comparison to voriconazole can be due to various facts including longer stay in the corneal stroma,[26] more susceptibility for Aspergillus flavus,[32] and probably more inflammation as evidenced by more stromal vascularization.[28]

Conclusion

Intrastomal antifungals, particularly Amphotericin B, should be considered as adjuvant therapy in cases of deep fungal keratitis that do not respond to topical treatment, potentially preventing the need for TPK.

Conflicts of interest:

There are no conflicts of interest.

Funding Statement

Hyderabad Eye Research Foundation, India.