Clinical Outcomes of Rose Bengal Photodynamic Therapy for Severe Infectious Keratitis

Abstract

Purpose:

To report the clinical outcomes, microbiological profile, and efficacy of Rose Bengal photodynamic therapy (RB-PDT) as adjuvant treatment in patients with severe infectious keratitis (IK).

Methods:

Prospective, noncontrolled, interventional study. Severe IK was defined as meeting ≥2 of the following: central ulcer, infiltrate ≥3 mm, presence of corneal melting or hypopyon, and lack of clinical improvement with conventional medical treatment. Clinical success was defined as avoiding a therapeutic keratoplasty (TPK).

Results:

A total of 21 eyes from 21 patients underwent RB-PDT. Clinical success was achieved in 61.9%. The mean follow-up was 9 ± 3.33 months (range: 4–15). Cultured samples were positive for 27 total microorganisms, including Fusarium spp (n = 12, 44.44%), S. epidermidis (n = 4, 14.81%), Acanthamoeba (n = 2, 7.40%), Pseudomonas spp (n = 3, 11.11%), and others. Cultured samples were grouped into bacterial (44.44%), fungal (48.14%), and polymicrobial (if cultured for multiple microorganisms, 28.57%) categories for analysis. Corneal melting (57.14%), hypopyon (57.14%), and fung-ball (38.1%) were present at the initial visit. The mean time from onset to clinical success was 89.69 ± 56.59 days. In cases of clinical success, the mean time from RB-PDT to clinical success was 46.69 ± 19.47 days. There was a total of 16 additional interventions, including evisceration (2), optical penetrating keratoplasty (5), conjunctival flap (3), amniotic membrane transplant (1), and topical losartan therapy (5). Clinical variables associated with clinical failure included diabetes and fung-ball at presentation.

Conclusions:

RB-PDT shows potential as an adjuvant therapy for managing severe infectious keratitis. TPK was avoided in 61.9% of cases.

Infectious keratitis (IK) can be an ophthalmic emergency when it presents with rapid progression and severe complications, as it is a leading cause of corneal blindness in both developing and developed countries.¹ Standard treatments typically involve the frequent application of topical antimicrobials and antifungals, and in severe cases, therapeutic keratoplasty(TPK).² The increasing prevalence of antibiotic resistance, the limited access to compounded and fortified antibiotics, and the lack of access to eye banking in low- and middle-income areas represent challenges in treating IK.³ Even in patients who receive adequate treatment, severe IK may lead to corneal perforation and the need for a TPK. However, the outcomes of TPK in an inflamed eye are often suboptimal compared with a quiet eye.⁴

Rose Bengal photodynamic therapy (RB-PDT) has emerged as a novel therapeutic modality that has gained significant interest over recent years. Driven by scientific efforts by Amescua et al^5–12 at the Bascom Palmer Eye Institute, RB-PDT has demonstrated in vivo, ex vivo, and clinical safety and efficacy in treating IK. However, there is a paucity of clinical studies on the real-world effectiveness of RB-PDT. This study aims to report on the clinical outcomes, microbiological profile, and efficacy of RB-PDT as an adjuvant therapy in patients with severe IK.

METHODS AND SUBJECTS

This prospective, noncontrolled, interventional study was conducted at the Instituto de Oftalmologia Fundacion Conde de Valenciana IAP in Mexico City, Mexico, from March 2023 to March 2024. All patients enrolled in the study participated voluntarily after reading and signing a written informed consent, previously reviewed and approved by our institution's committees of ethics, research, and biohazards, following the tenets of the Declaration of Helsinki and the Health Insurance Portability and Accountability Act.

Participants Selection

Patients aged 18 years or older with infectious keratitis caused by bacteria, fungi, or protozoa were included in the study. Those considered for RB-PDT had severe infectious keratitis, defined as meeting 2 or more of the following clinical criteria: (1) a central ulcer involving the central 3 mm of the visual axis, (2) an infiltrate larger than 3 mm, (3) corneal melting, and (4) presence of hypopyon. In addition, all included patients showed persistent clinical deterioration, defined as an increase in at least 1 of the following parameters: size of the epithelial defect, hypopyon height, or extent of corneal infiltrates, despite receiving a minimum of 2 weeks of conventional medical treatment. The same investigator assessed all visits to ensure consistency (REB).

The timing of symptom onset to first RB-PDT, days from first RB-PDT to resolution or TPK, and days from symptom onset to resolution or TPK were recorded. The RB-PDT was considered successful if patients avoided a TPK, and clinical failure was defined as progression to corneal perforation, the need for TPK, or evisceration because of inadequate response to RB-PDT. Patients classified as having achieved clinical resolution had no epithelial defects, pain, signs of clinical inflammation, or corneal infiltrates. Exclusion criteria included patients with active ocular surface disorders unrelated to infectious keratitis (eg, autoimmune keratopathy or herpetic keratitis) and those unable to provide informed consent. Herpetic keratitis was specifically excluded because of the known risk of viral reactivation after photodynamic therapy. To reflect real-world applicability, patients previously treated with antimicrobials at other centers were not excluded.

Microbiological Diagnosis

Cultures were performed either in the outpatient clinic or the emergency room. Corneal scrapings were obtained after applying topical tetracaine hydrochloride 0.15% using a calcium alginate swab. Samples were inoculated onto 3 solid media (blood agar, chocolate agar, and Sabouraud agar) and 1 liquid medium (brain–heart infusion broth), and incubated at 37°C with 5% CO₂. This approach was intended to facilitate the recovery of a wide range of potential pathogens in infectious keratitis. A culture was considered positive if the growth of the same organism was observed on more than 1 solid medium, or if confluent growth occurred at the site of inoculation on a single solid medium. Positive growth in the liquid medium was also considered significant if it was corroborated by consistent clinical signs and microscopic findings. Gram stains were performed on all samples; however, positive microscopy without bacterial growth on solid culture was considered insufficient for a positive result. The VITEK automated system (bioMérieux, France) was used for bacterial identification, and antimicrobial susceptibility was determined using the disk diffusion method according to CLSI (Clinical and Laboratory Standards Institute) guidelines. Fourteen antibiotics were tested, including polymyxin B, oxacillin, neomycin, sulfamethoxazole, vancomycin, gentamicin, ciprofloxacin, ofloxacin, ceftriaxone, cefazolin, ceftazidime, tobramycin, tetracyclines, and norfloxacin. Antibiograms for microorganisms not included in the automated system were performed using the Kirby–Bauer method. Patients who had a negative culture result were recultured up to 3 times.

Ophthalmological Examination

All participants underwent a comprehensive medical history, symptom assessment, pain evaluation, and ophthalmological examination at diagnosis. The ophthalmological evaluation included uncorrected visual acuity (UCVA) and best-corrected visual acuity (BCVA) using a Snellen chart, slit-lamp examination with and without fluorescein, and assessment of eyelid edema and erythema, conjunctival edema and hyperemia, discharge, epithelial defect area, infiltrate area and depth, ulcer location, presence of corneal necrosis or stromal degradation (“corneal lysis”), hypopyon, presence of a fung-ball (organized, immobile, whitish mass in the anterior chamber, distinct from a hypopyon, which typically appears as a layered, gravity-dependent accumulation of inflammatory cells), corneal vascularization, anterior chamber cell and flare, fundus examination, slit-lamp photography, and tonometry. Visual acuity values of count fingers, hand movement, light perception, and no light perception were taken as 2.1, 2.4, 2.7, and 3.0 LogMAR, respectively. All patients with fungal keratitis received topical antifungals, including natamycin, voriconazole, and itraconazole. These were continued after RB-PDT in cases of clinical success, and in patients who required TPK, antifungal therapy was also maintained postoperatively.

RB-PDT Surgical Technique

Rose Bengal powder was obtained sterilely in an amber vial containing 5 grams (198250-5G; Sigma Aldrich; St. Louis, MO). Ten milligrams of Rose Bengal were weighed into Eppendorf tubes and stored covered with aluminum foil in a dark room. On the day of the procedure, the Rose Bengal powder was diluted with 10 mL of balanced saline solution to achieve a concentration of 0.1%. The procedure was performed as previously described by Naranjo et al.⁹ After topical anesthesia, a lid speculum was placed in the eye while the procedure was performed. If the epithelial defect of the ulcer was smaller than 8 mm in diameter, the adjacent epithelium was removed to create an 8-mm deepithelialized area to enhance Rose Bengal absorption. An 8-mm sponge was placed on the cornea, and 3 drops of the Rose Bengal solution were applied to the sponge, followed by 1 drop every 3 minutes for 30 minutes to maintain saturation. The sponge was removed, and the ocular surface was irrigated with a balanced saline solution (BSS). Irradiation with a custom-made green LED light source at 6 mW/cm² was applied for 15 minutes to deliver a total energy dose of 5.4 J/cm². During irradiation, BSS was used every minute to prevent corneal stromal dehydration. Finally, a drop of 0.5% moxifloxacin was applied at the end of the procedure.

Green Light–Emitting Device

A custom device was developed to provide a controlled irradiance of green light (525 nm) for therapeutic applications. The LED array was characterized based on peak wavelength and emission power, with 3 models tested under specific power and thermal conditions. A printed circuit board (PCB) and 3D-printed mounts were designed to house the LEDs. Key components of the system include a mobile arm for positioning, an optical distance indicator, and a microcontroller (Atmega328P) for power and timing control. To characterize the LED array, we measured the power density at multiple distances and conducted a spectrographic analysis to confirm emission within the desired wavelength band (84% at 525 ± 5 nm) while ensuring no harmful light emissions. Temperature tests confirmed adequate heat dissipation, and luminous power-distance curves were plotted to determine optimal placement. Temperature measurements were taken using an optical thermometer, and the maximum temperature recorded on the back of the phenolic board was 45°C. To mitigate this, aluminum heat sinks were installed, effectively reducing the operational temperature. Testing demonstrated that a single LED (BL_01), positioned 6 cm from the target and operated at 2.2V, delivered a constant power density of 6 mW/cm², resulting in a total energy dose of 5.4 J/cm² after 15 minutes of continuous exposure. Although an array was used in the main device configuration, this measurement confirmed that a single LED was capable of reaching the target irradiance under controlled conditions. The output remained stable over the entire 15-minute period.⁶ In addition, the geometry of the device was calculated to ensure the transverse diameter of the green light beam was 9 mm at a distance of 6 cm from the base of the device. The device was assembled and tested for functionality. The optical distance indicator successfully guided positioning, and LED current consumption and light intensity were validated against initial characterizations. The built-in timer ensured accurate operation, and heat dissipation systems maintained safe operating temperatures.

Postoperative Evaluation

Patients were evaluated on day 1, day 3, week 1, week 2, month 1, month 3, and month 6 postprocedure to assess symptoms, pain scale, and ophthalmological parameters. Additional evaluations were performed as necessary based on clinical severity. Data recorded at initial consultation, day 1, week 1, month 1, month 3, and month 6 after RB-PDT were considered for the study. Slit-lamp examination with and without fluorescein and clinical photography was performed at each visit. Patients continued receiving targeted medical therapy according to the presumed or confirmed etiology of the infectious keratitis. When microbiological results were available, antimicrobial regimens were adjusted accordingly. At 2 weeks, the need for a second procedure was evaluated. If significant clinical improvement was observed, defined as a reduction in epithelial defect, infiltrate size, and hypopyon without signs of worsening, patients were maintained on specific medical treatment until complete resolution of the infection. If no definite improvement was observed but some clinical progress was noted, a second RB-PDT session was performed. In cases of complications such as corneal perforation, TPK was indicated. Other procedures, such as a conjunctival flap, were performed to avoid perforation. Patients were subsequently monitored weekly, bi-weekly, or monthly, depending on their progress. Upon resolution, a comprehensive ophthalmological examination was conducted, including visual acuity assessment, slit-lamp examination with fluorescein, clinical photography, tonometry, and, when feasible, fundoscopy. Topical corticosteroids were used selectively after RB-PDT to control inflammation, primarily in bacterial cases or after clinical resolution in other etiologies. They were not used in active fungal cases to avoid exacerbating fungal growth.

In addition, in selected patients with clinical success and corneal scarring, the use of off-label topical losartan 0.1% 6 times a day was prescribed.¹³ This treatment was specifically for patients with corneal scarring affecting the central visual axis but not extending beyond the mid-stromal layer. It was prescribed for individuals who no longer needed other treatments and were deemed cured. This treatment approach did not include patients with deeper stromal involvement or those requiring more aggressive interventions. The reason for starting topical losartan therapy was to improve corneal transparency, especially in cases where the scarring was localized but affected visual function. Also, the selected patients were not prescribed concomitant steroids during this period.

Statistical Analysis

All data were captured in an Excel spreadsheet for subsequent analysis. The statistical analysis was performed using GraphPad Prism version 10.1.1 for macOS (GraphPad Software, Boston, MA). The Kolmogorov–Smirnov test was used to assess data distribution. Descriptive statistics are presented as means and standard deviations for data with a symmetric distribution and medians and interquartile ranges for skewed data. Categorical data are presented as proportions. In cases of bilateral disease, information from the most severely affected eye was included. A paired, two-tailed Student t test was performed to compare continuous numerical variables before and after treatment. An unpaired, two-tailed Student t test was used for comparisons between groups. Fisher exact test was employed to analyze categorical variables. The Baptista–Pike test was used to calculate the odds ratio. A P-value of <0.05 was considered statistically significant.

RESULTS

A total of 21 eyes from 21 patients underwent RB-PDT for severe IK. Clinical success was achieved in (n = 13) 61.9% of cases. Table 1 and Figure 1 summarize the demographic and clinical characteristics. Fourteen patients were female (66.6%) and 7 were male (33.3%). The mean age at the time of RB-PDT was 47.71 ± 19.8 years (range: 18–89). The mean follow-up was 9 ± 3.33 months (range: 4–15). The median initial BCVA was 2.4 (IQR: 2.4–2.7) LogMAR. All patients underwent corneal cultures before RB-PDT. A total of 27 microorganisms were isolated, including Fusarium spp (44.44%), S. epidermidis (14.81%), Acanthamoeba (7.4%), Pseudomonas spp (11.11%), and others, and they were grouped in bacterial (44.44%), fungal (48.14%), and polymicrobial (if cultured for multiple microorganisms, 28.57%) categories for analysis. Tables 2 and 3 show a detailed breakdown of the cultured microorganisms and the antimicrobial resistance profile. Identified risk factors for corneal ulcer were ocular trauma (28.57%), diabetes (28.57%), and contact lens use (33.33%). On presentation, the mean largest diameter of the epithelial defect was 5.92 ± 2.31 mm, and the mean infiltrate was 5.66 ± 1.75 mm. Corneal melting (57.14%), hypopyon (57.14%), and fung-ball (38.1%) were present at the initial visit. Regarding the number of RB-PDTs performed, 17 patients (81%) underwent 1 session, whereas 4 patients underwent 2 sessions (19%). In these 4 cases, the RB-PDT was repeated because of partial improvement after the first session. The mean time from onset to the first RB-PDT was 57.25 ± 53.63 days (range: 15–133). For the second RB-PDT, the mean time after the first session was 21.00 ± 8.04 days (range: 15–32). The median time from symptom onset and RB-PDT was 28 days (IQR: 22–35). The mean time from onset to clinical success was 89.69 ± 56.59 days. In cases of clinical success, the mean time from RB-PDT to clinical success was 46.69 ± 19.47 days. In cases of clinical failure, the mean time from RB-PDT to perforation was 44.25 ± 42.11 days. There was no significant difference between the group with clinical success and the group with clinical failure regarding mean time from onset to RB-PDT (P = 0.76). There was a total of 24 additional interventions, including TPK (n = 8, clinical failure), evisceration (n = 2), optical penetrating keratoplasty (n = 5), conjunctival flap (n = 3), amniotic membrane transplant (n = 1), and losartan eye drops (n = 5). The 2 cases of evisceration had undergone a prior TPK and had grueling pain with no visual potential and were, therefore, eviscerated. In all cases of clinical success, the anterior segment anatomy was retained, without perforation or structural loss. Of the 13 patients who achieved clinical success, 5 (38.5%) subsequently underwent optical PKP for visual rehabilitation. In contrast, the remaining 8 (61.5%) either maintained functional vision with spectacles or declined surgery because of personal or socioeconomic reasons. The 5 cases of optical penetrating keratoplasty were performed 6 months after the resolution of infectious keratitis in cases of clinical success, as an intervention for visual rehabilitation. The mean BCVA at the last follow-up was 1.64 ± 1.11 LogMAR. Clinical variables associated with clinical failure were diabetes (OR: 20.0, 95%CI: 1.93–249.1) and fung-ball at the time of presentation (OR: 84.0, 95% CI: 5.13–978.7). Table 4 shows variables associated with clinical failure. Among the 5 patients who received topical losartan 0.1% after achieving clinical success, the mean BCVA improved from 0.47 ± 0.23 to 0.25 ± 0.27 LogMAR, suggesting a possible visual benefit temporally associated with losartan use. Likewise, corneal densitometry analysis showed a reduction, from an initial value of 33.42 ± 4.55 GSU to 27.95 ± 7.99 GSU (Δ: −5.47, P = 0.07) at the final follow-up.

TABLE 1.

Demographic and Clinical Characteristics of Patients Who Underwent RB-PDT for Severe Infectious Keratitis

No.	Age/Sex	Eye	Risk Factor	Culture	Initial BCVA	Size of Epithelial Defect*	Size of Infiltrate*	Corneal Melt	Fung-Ball	Hypopyon	Follow-Up	No. of RB-PDT	Clinical Success	Additional Procedure	BCVA at the Last Follow-Up
1	29/M	OD	None	Fusarium spp	HM	6 mm	3 mm	Yes	No	No	12 mo	1	Yes	PK	20/80
2	43/M	OD	DM	Fusarium spp	HM	4 mm	4.8 mm	No	Yes	Yes	12	1	Yes	Losartan	20/30
3	43/F	OD	SLE, CL	Fusarium spp	LP	6.5 mm	4.7 mm	Yes	No	No	12	1	Yes	Conjunctival flap, PK	20/200
4	89/F	OD	None	P. aeruginosa	LP	4 mm	2.5 mm	No	No	No	6	1	Yes	None	HM
5	89/F	OD	HAS, CL, KPro, steroid	P. aeruginosa	HM	6 mm	6 mm	Yes	No	No	6	1	Yes	AMT	20/60
6	48/F	OS	HSK	S. epidermidis	LP	5 mm	6.3 mm	Yes	No	Yes	12	2	Yes	None	HM
7	24/F	OS	None	S. aureus	HM	4.3 mm	5.5 mm	No	No	No	12	1	Yes	Losartan	20/20
8	29/F	OS	CL	No growth	20/200	2.5 mm	5 mm	No	No	Yes	6	1	Yes	Losartan	20/20
9	26/F	OS	Trauma	Fusarium spp, S. warneri	CF	3.4 mm	4.2 mm	No	No	Yes	6	1	Yes	Losartan	20/60
10	60/F	OD	CL, steroid	No growth	HM	5 mm	5.9 mm	No	No	Yes	6	1	Yes	None	CF
11	18/M	OS	Trauma	Fusarium spp, S. warneri, S. epidermidis	CF	7 mm	4.5 mm	Yes	No	No	6	1	Yes	Losartan	20/80
12	30/M	OS	PK, steroid	Pseudomonas putida	HM	9 mm	8 mm	Yes	No	No	6	1	Yes	Conjunctival flap, PK	20/100
13	46/F	OS	CL, steroid	Fusarium spp, S. paucimobilis, Acanthamoeba	LP	12 mm	8 mm	Yes	No	No	6	1	Yes	Conjunctival flap	HM
14	66/F	OD	DM	Paecilomyces	HM	2.5 mm	3 mm	Yes	Yes	Yes	6	1	No	PK	LP
15	66/F	OD	DM, HAS, trauma, steroids	Fusarium spp	HM	9.5 mm	6 mm	Yes	Yes	Yes	6	1	No	PK	LP
16	50/F	OD	DM, trauma	Fusarium spp	LP	6.1 mm	7.5 mm	No	Yes	Yes	12	1	No	Evisceration	NLP
17	48/M	OD	CL	Fusarium spp, S. hominis	HM	6.8 mm	6.5 mm	No	Yes	Yes	12	2	No	None	NLP
18	65/F	OD	HAS, CL, HSK, steroids	Acanthamoeba, S. epidermidis	HM	7.2 mm	7 mm	No	No	No	12	2	No	Evisceration	NLP
19	47/M	OD	DM	Fusarium spp	LP	7 mm	9 mm	Yes	Yes	Yes	6	2	No	None	HM
20	31/F	OD	DM, trauma	Fusarium spp	HM	4.6 mm	4.6 mm	Yes	Yes	Yes	6	1	No	None	CF
21	55/M	OD	Trauma	Fusarium spp, S. epidermidis	HM	6 mm	7 mm	Yes	Yes	Yes	6	1	No	None	HM

CF, counting fingers; CL, contact lens; DM, diabetes mellitus; F, female; HAS, systemic arterial hypertension; HM, hand motion; HSK, herpes simplex keratitis; KPro, Keratoprosthesis; LP, light perception; M, male; NLP, no light perception; OD, right eye; OS, left eye; PK, penetrating keratoplasty; SLE, systemic lupus erythematosus.

^*Largest diameter.

FIGURE 1.

Clinical progression of patients undergoing RB-PDT.

TABLE 2.

Microbiological Profile of Patients Undergoing RB-PDT^*

Microorganism	N	%
Total	29	100
Positive cultures	27	93.10
Bacteria	12	44.44
Pseudomonas spp	3	11.11
S. epidermidis	4	14.81
S. warneri	2	7.40
S. aureus	1	3.70
S. hominis	1	3.70
Sphingomonas paucimobilis	1	3.70
Fungal	13	48.14
Fusarium spp.	12	44.44
Paecilomyces spp.	1	3.70
Protozoa	2	7.40
Acanthamoeba	2	7.40
Negative cultures	2	6.89

^*Total cultured microorganisms = 27 from 21 patients.

S, Staphylococcus.

TABLE 3.

Microbial Antibiogram Reported in Bacteria

Characteristic	Total	Sensitive		Resistant
	n	n	%	n	%
Gram −
Pseudomonas spp.
Meropenem	2	2	100.0	0	0.0
Piperacillin–tazobactam	2	2	100.0	0	0.0
Gentamicin	3	3	100.0	0	0.0
Ceftazidime	3	3	100.0	0	0.0
Ciprofloxacin	2	1	50.0	1	50.0
Ceftriaxone	2	0	0.0	2	100.0
Tigecycline	2	0	0.0	2	100.0
Sphingomonas paucimobilis
Cefotaxime	1	1	100.0	0	0.0
Netilmicin	1	1	100.0	0	0.0
Ciprofloxacin	1	1	100.0	0	0.0
Ampicillin	1	1	100.0	0	0.0
Gentamicin	1	0	0.0	1	100.0
Gram +
Staphylococcus epidermidis
Vancomycin	4	4	100.0	0	0.0
Levofloxacin	4	3	75.0	1	25.0
Erythromycin	4	3	75.0	1	25.0
Clindamycin	4	3	75.0	1	25.0
Gentamicin	3	2	66.7	1	33.3
Doxycycline	3	2	66.7	1	33.3
Moxifloxacin	3	2	66.7	1	33.3
Ciprofloxacin	4	2	50.0	2	50.0
Staphylococcus warneri
Vancomycin	2	2	100.0	0	0.0
Gentamicin	2	2	100.0	0	0.0
Doxycycline	2	2	100.0	0	0.0
Ciprofloxacin	2	2	100.0	0	0.0
Levofloxacin	2	2	100.0	0	0.0
Moxifloxacin	2	2	100.0	0	0.0
Clindamycin	2	2	100.0	0	0.0
Erythromycin	2	1	50.0	1	50.0
Staphylococcus aureus
Vancomycin	1	1	100.0	0	0.0
Ampicillin	1	1	100.0	0	0.0
Erythromycin	1	1	100.0	0	0.0
Clindamycin	1	1	100.0	0	0.0
Ciprofloxacin	1	1	100.0	0	0.0
Staphylococcus hominis
Vancomycin	1	1	100.0	0	0.0
Clindamycin	1	1	100.0	0	0.0
Ciprofloxacin	1	1	100.0	0	0.0
Ampicillin	1	0	0.0	1	100.0

TABLE 4.

Univariate Analysis of Clinical Variables Associated With Clinical Failure After RB-PDT

Variable	Clinical Success (n = 13, %)	Clinical Failure (n = 8, %)	OR (95% CI)	P
Fung-ball	1 (7.69)	7 (87.5)	84 (5.13–978.7)	<0.001
Diabetes	1 (7.69)	5 (62.5)	20 (1.93–249.1)	0.013
Hypertension	1 (7.69)	2 (25.0)	4.0 (0.37–61.94)	0.530
Fungal infection	6 (46.15)	7 (87.5)	8.16 (0.97–103.0)	0.085
Bacterial infection	8 (61.54)	3 (37.5)	0.37 (0.07–2.02)	0.387
Hypopyon at presentation	5 (38.46)	7 (87.5)	11.2 (1.24–138.9)	0.066
Steroid use at presentation	5 (38.46)	3 (37.5)	0.9 (0.18–5.13)	0.999
Use of contact lens	4 (30.77)	2 (25.0)	0.75 (0.11–5.17)	>0.999
Associated traumatism	2 (15.38)	4 (50.0)	5.50 (0.78–34.45)	0.146
Corneal melt at presentation	7 (53.85)	5 (62.5)	1.42 (0.28–7.11)	>0.999
Corneal neovascularization at presentation	5 (38.46)	3 (37.5)	0.96 (0.18–5.13)	>0.999
BCVA same or better than HM at presentation	9 (69.23)	2 (25.0)	0.14 (0.02–1.28)	0.080
Epithelial defect with largest diameter >6.0 mm	4 (30.77)	6 (75.0)	0.14 (0.02–1.28)	0.080
Infiltrate size with largest diameter >6.3 mm	2 (15.38)	5 (62.5)	9.16 (1.33–56.63)	0.055
Time from onset to RB-PDT >30 d	3 (23.08)	4 (50.0)	3.33 (0.54–17.32)	0.345

HM, hand motion; LP, light perception.

DISCUSSION

Similar to the chemical reaction seen in crosslinking, RB-PDT generates reactive oxygen species (ROS) that elicit apoptosis of infected cells by various microorganisms, including bacteria, fungi, and protozoa.^5,11,14 However, RB-PDT has demonstrated higher efficiency in vitro as a treatment for fungal and methicillin-resistant S. aureus.⁵ Figure 2 showcases the clinical progression of severe IK secondary to S. aureus after RB-PDT. In addition, a randomized controlled trial concluded that adjuvant riboflavin–mediated photodynamic therapy may lead to more perforations in deep fungal infections and had to stop enrollment early.¹⁵ However, the choice of crosslinking protocol and early treatment application might have influenced this outcome.¹⁶ Because we have more experience with RB-PDT than riboflavin-mediated photodynamic therapy, and in our geographic context we see an important proportion of Fusarium spp. keratitis, we opted for RB-PDT.

FIGURE 2.

Clinical progression of severe infectious keratitis after RB-PDT. A–D, A 24-year-old woman presented with severe infectious keratitis because of S. aureus, with an initial best-corrected visual acuity (BCVA) of hand motion. B, Postoperative day 1 after RB-PDT. C, At 3 weeks after RB-PDT, clinical resolution was achieved, and steroid and losartan eye drops were initiated to modulate corneal wound healing. D, At the 12-month follow-up, after RB-PDT and adjuvant losartan eye drops, the BCVA improved to 20/20. E–H, A 29-year-old man presented with Fusarium spp. keratitis, exhibiting signs of active corneal melting. F, Postoperative day 1 after RB-PDT. G, Four weeks after RB-PDT, clinical resolution was achieved. H, Eleven months after the initial presentation, the patient underwent optical keratoplasty and has a spectacle-corrected visual acuity of 20/80.

Previous studies have supported the clinical efficacy of RB-PDT. An initial study by Naranjo et al⁹ reported a success rate of 72%, which increased to 77.4% in a medium-term follow-up study.⁸ No adverse effects were associated with the treatment in this study. This study addressed infectious keratitis caused by various microorganisms, including Acanthamoeba (59%), Fusarium spp (24%), Curvularia spp (6%), and Pseudomonas spp (12%). In addition, in 2 cases (6.9%), there was no growth; however, a 93.1% culture positivity rate is consistent with the challenges of microbiological identification in severe infectious keratitis. The focus on severe keratitis with large infiltrates, higher microbial load, melting, or hypopyon likely contributed to the high culture positivity. In addition, our center has observed increased positivity rates after COVID-19, reflecting improved microbiological practices and a greater referral of severe cases, which may explain the high yield in this cohort.¹⁷ The time to resolution varied between the studies, with Naranjo et al⁹ reporting a resolution time of 46.9 ± 26.4 days after RB-PDT, and Sepulveda-Beltran et al⁸ indicated a resolution time of 2.72 ± 1.85 months. In our study, we observed a success rate of 61.9% with a mean resolution time of 46.69 ± 19.47 days, results that align with those of Naranjo and coworkers.⁹ Notably, 1 patient (#5) with a Lucia KPro in her left eye developed infectious keratitis unresponsive to standard antimicrobial therapy. A 91-year-old woman with a Lucia KPro developed severe Pseudomonas aeruginosa keratitis unresponsive to intensive antimicrobial therapy, with progressive infiltrate, epithelial defect, and corneal melting. Treatment with RB-PDT combined with an amniotic membrane graft led to complete resolution within 21 days, stable vision (20/60), and no recurrence at 6 months. To our knowledge, this is the first reported case of a Lucia KPro infection successfully managed with this approach.¹⁸ Although no pan-resistant isolates were encountered in this series, RB-PDAT has demonstrated potent anti-Pseudomonas activity and achieved rapid clinical resolution in a high-risk Lucia KPro case, supporting its potential role with multidrug-resistant microbes.¹⁹ These scenarios highlight the potential of RB-PDAT as an effective adjuvant therapy in complex cases, where conventional treatments may prove insufficient.

Addressing the potential reasons for our study's slightly lower success rate is crucial. Several factors may influence this outcome. First, we shared the same definition of clinical success, which was tied to preventing a TPK. We also reported a higher proportion of patients with vegetal ocular trauma, which is a known factor for fungal keratitis, especially Fusarium.²⁰ Likewise, deep fungal infections, seen in 8 of 21 patients (38.10%) with fung-balls, are often more aggressive and often end up with perforation despite traditional medical treatment.²¹ In this study, patients who experienced clinical failure (38.1%) after RB-PDT were of fungal (87.5%) and polymicrobial (12.5%) origin, and 87.5% of them presented with hypopyon and fung-ball. This observation may be explained by the fact that deep fungal infections may not receive the full benefit of Rose Bengal, as its penetration is limited, likely restricted to the first 100 μm of the anterior corneal stroma because of its strong interaction with collagen.^7,22 Likewise, diabetes was associated with clinical failure in this cohort and is a known independent risk factor that affects the severity of fungal keratitis.²³ Nonetheless, RB-PDT demonstrated fair clinical effectiveness in cases where fungal cultures were obtained (7 out of 13 cases). In addition, infections categorized as polymicrobial infections were predominantly fungal, which further underscores the complexity of managing such cases. Because of several socioeconomic limiting factors, personal/religious choices, eye banking availability, geographic restrictions, and, in some cases, with acceptable spectacle-corrected visual acuity, most patients with clinical success (76.9%) did not undergo an optical keratoplasty for visual rehabilitation. However, 5 patients with clinical success (38%) were prescribed off-label topical losartan 0.1%.¹³ In this group, we observed excellent outcomes and a reduction of the corneal scarring, with the mean final BCVA being 0.25 ± 0.27 LogMAR. Nonetheless, these findings should be taken with caution because it could be possible that the resolution of the corneal opacity was partly physiological corneal wound repair and the effect of topical steroids rather than a direct benefit from the topical losartan. Last, 2 patients (Fusarium spp and Acanthamoeba/Staphylococcus epidermidis infection) had to undergo evisceration because of deterioration despite medical treatment and RB-PDT. In the former, the patient experienced perforation and underwent a TPK. However, the patient developed corneal melt and ensuing endophthalmitis that did not improve with intravitreal treatment. In the case of combined infection of Acanthamoeba and S. epidermidis, the time from onset to the application of RB-PDT was 133 days, which may suggest a delay in diagnosis and a delay in starting treatment. Because this patient had grueling pain in the infected eye that did not respond to medical treatment, the patient opted for evisceration. S. epidermidis was identified in 14.81% of organisms, predominantly in cases of mixed infections with more virulent pathogens such as Fusarium or Acanthamoeba. Its role in these cases is likely complementary, rather than as a primary driver of severe IK.²⁴

Several limitations of this study must be addressed. First, a selection bias is present because we limited the application of RB-PDT to only severe IK using arbitrary clinical criteria. Also, we studied only Hispanic patients attending a tertiary eye-care hospital, many of whom were referred or had already started topical treatment elsewhere. Likewise, the microbiological profile and underlying risk factors may be distinctive of our geographical context and not generalized to other settings. This may limit the application of RB-PDT. Also, a limitation of this study is the inclusion of mixed microbiologic profiles, which may have distinct clinical courses, making direct comparisons challenging. Future studies may stratify outcomes by the pathogen to evaluate RB-PDT's efficacy more precisely. Another limitation of this study is the absence of antifungal susceptibility testing, as it was not available at our institution. Second, it was not possible to analyze the potentially toxic effects of RB-PDT on the corneal endothelium using specular microscopy because of the severity of the condition at presentation, and no anterior chamber washes were performed during the cases of TPK. Furthermore, the lack of depth assessment of infiltrates using AS-OCT may have impaired patient selection, particularly for those with deep stromal involvement. Future studies should incorporate standardized depth measurements, such as AS-OCT, to refine patient selection and improve the evaluation of RB-PDT efficacy. Last, to confirm the efficacy of RB-PDT as an adjuvant therapy for severe IK, a randomized trial comparing medical therapy only to combined medical therapy and RB-PDT must be performed. Despite these limitations, we replicate the success of RB-PDT reported in other institutions, particularly in cases of severe IK in a distinct geographic population. In addition, we also found fair results (but statistically nonsignificant) with adjuvant 0.1% losartan eye drops in selected cases of mid-superficial haze, which warrant further investigation. Future directions of RB-PDT could lead to improved green-light radiation protocols, enhanced Rose Bengal penetration in the corneal stroma, prognostic categories based on microbiological profile, and optimizing the timing of treatment application.

In conclusion, our findings suggest that RB-PDT may have potential utility as an adjuvant treatment for severe infectious keratitis. It could help shorten the time to clinical resolution and reduce the need for therapeutic keratoplasty in otherwise challenging cases; however, further studies are needed to confirm its effectiveness.