Infectious keratitis after corneal crosslinking: a systematic review

Abstract

Corneal crosslinking is an FDA approved therapy to stiffen the cornea and prevent progression of corneal ectasia in patients with keratoconus. The standard procedure involves removal of the corneal epithelium (epithelial-off) prior to treatment. Variations to the standard procedure include accelerated crosslinking and transepithelial procedures. This study reviewed what is known regarding the risk for infection following epithelial-off crosslinking, the spectrum of pathogens, and clinical outcomes. Twenty-six publications were identified. All eyes were fit with a bandage contact lens post-operatively. Available data indicate that the overall frequency of infectious keratitis after epithelium-off crosslinking is low. Bacterial infections are the most common, presenting an average of 4.8 days after surgery. The use of steroids and bandage contact lenses in the immediate postoperative period and/or a history of atopic or herpetic disease were associated with infection. These patients require intense post-operative care with prophylactic anti-viral therapy when appropriate.

Introduction

Corneal crosslinking was first proposed as a treatment for keratoconus in 2003 by Wollensak et al.¹ It has been shown to successfully stop the progression of corneal thinning by stiffening the cornea and potentially delaying the need for penetrating keratoplasty in these patients.^2–5 The standard procedure describes the removal of the corneal epithelium to facilitate permeation of the stroma with a riboflavin solution. The cornea is then exposed to UVA light at 370 nm and energy density of 3 mW/cm² for 30 minutes. Variations of the standard procedure such as accelerated corneal crosslinking and transepithelial crosslinking are also utilized.^6–10 Accelerated corneal crosslinking is performed with UVA intensity of 9 mW/cm² for 10 min. Both standard corneal crosslinking and accelerated corneal crosslinking have been shown to be effective in adult and pediatric populations.^{11, 12} While corneal crosslinking is a relatively safe and effective procedure, there have been reports of complications following the procedure.^13–16 These complications include severe pain, sterile infiltrates, infectious keratitis, persistent epithelial defect and corneal melt, corneal haze or scarring, corneal edema, endothelial damage, and in some cases, progression of keratoconus.

Corneal crosslinking has also been used to treat severe, multidrug-resistant corneal ulcers.^17–19 The benefit of the procedure may be due in part to the well-studied antimicrobial effects of riboflavin and UVA light.^20–22 In addition, the cornea becomes more resistant to enzymatic degradation, which could offer resistance to proteases that drive corneal melting.^{23, 24} Corneal crosslinking has not been shown to be effective in cases of viral keratitis and the procedure has exacerbated corneal melting in some cases.²⁵ With further studies and randomized controlled trials, corneal crosslinking may be a promising adjunct to therapy for treatment resistant infectious keratitis, especially in cases where there is antibiotic resistance. The focus of this work is to review what is known regarding the risk for infection following epithelium-off crosslinking, the spectrum of pathogens, and clinical outcomes.

Methods

A comprehensive literature search was performed in PubMed using combinations of the following search terms: “corneal cross linking” and “keratitis” or “infectious” or “microbial” or “complications.” In addition, references from the articles were manually searched for additional relevant studies. Fifty-seven studies were identified. Nineteen studies were excluded due to discussion of sterile keratitis post corneal crosslinking.^26–44 Four studies did not report new cases in the literature and were excluded.^45–48 Two studies were excluded for discussing treatment of infectious keratitis with corneal crosslinking.^{49, 50} Six studies mentioned cases of infectious keratitis but did not report specific clinical data and were excluded. ^{13–16, 51, 52} A total of 26 publications were included in this study.

Statistical analysis: Statistical analysis was performed using Sigma Plot (Systat Software, San Jose, CA). For categorical variables, data are represented as raw number and percent of total cases. For continuous variables, data are represented as mean ± standard deviation. A chi-square test or a t-test were used where appropriate. Statistical significance was set at p<0.05.

Results

Together, the studies reported on data from 49 eyes (detailed in Table 1). All studies were case reports or cases series of infectious keratitis after corneal crosslinking procedures except one, which was a comparative retrospective review and reported 9 cases of infectious keratitis.⁵³ Of the reported cases, 33 patients were male and 12 were female. Sex was not specified in one case. Three reported cases were bilateral. One bilateral case was female and two were male. The average age of patients was 22.6 years ± 9.4 years. Twenty-four eyes underwent standard epithelium-off corneal crosslinking. Twenty-two eyes underwent accelerated crosslinking. Three eyes underwent transepithelial crosslinking. Of the bilateral cases, each of the three patients had undergone different procedures: one with standard crosslinking, one accelerated crosslinking, and one transepithelial crosslinking.

Table 1:

Reports of infectious keratitis following corneal cross-linking

	Year	Design*	CXL method	Number of eyes	Patient age(s)	PK**	Final BCVA
Schear et al.	2020	CR	standard	1	17	1	20/20–2
McGirr et al.	2020	CR	standard	1	57	1	20/25
Kodavoor et al.	2020	CS	aCXL	3	15–22		20/40, 20/30, 20/40
Tzamalis et al.	2019	RR	aCXL	9	12–28		20/30, 20/400, 20/30, 20/80, 20/200, 20/30, 20/40, 20/120, 20/60
Ting et al.	2019	CR	aCXL	1	23		20/30
Sitaula et al.	2019	CR	standard	2***	18		20/30
Maharana et al.	2018	CS	aCXL	7	11–17	1	6/18, 6/36, 6/18, 1/60, 2/60, 3/60, CF
Oakey et al.	2017	CR	transepi	2***	33	2	20/20, 20/20
Al-Amry et al.	2017	CR	standard	1	20		20/30
Rana et al.	2015	CS	Transepi aCXL & aCXL	2	19, 18		CF, 20/80
Kodavoor et al.	2015	CR	aCXL	1	15	1	20/40
Fasciani et al.	2015	CR	standard	1	22		HM
Al-Qarni et al.	2015	CS	standard	2	18, 21	1	20/25, 20/30
Shetty et al.	2014	CS	standard	4	16–27		20/20, 20/30, 20/120, 20/200
Gautam et al.	2013	CR	standard	1	36		20/60
Hafezi et al.	2012	CR	-	1	21		20/30
Bodemann and Kohnen	2012	CR	standard	1	40	1	-
Yuksel et al.	2011	CR	standard	1	31		20/25
Rama et al.	2011	CR	standard	1	35		20/400
Sharma et al.	2010	CR	standard	1	19		20/200
Garcia-Delpech et al.	2010	CR	standard	1	23		HM
Zamora et al.	2009	CR	standard	1	32	1	20/50
Rama et al.	2009	CR	standard	1	32		20/200
Pollhammer and Cursiefen	2009	CR	standard	1	42		20/63
Perez et al.	2009	CR	standard	1	29		20/22
Santonja et al.	2007	CR	standard	1	21		20/25

T*Design: case report (CR), case series (CS), retrospective review (RR)

^**PK: eyes requiring a penetrating keratoplasty

^***bilateral case

As described in Table 2, bacteria were the most common etiological agent (P<0.001). Thirty one eyes were culture positive.^{53–59,60–68} Twenty-eight eyes were infected with gram-positive bacteria, with the most common organism being Staphylococcus aureus. Other gram-positive organisms identified included coagulase-negative Staphylococcus species ^{54, 56, 59} and viridans streptococci, such as S. salivarius, S. oralis, and S. sanguinis.^{56, 58} Only two eyes were infected with gram-negative bacteria. These eyes were culture positive for Pseudomonas aeruginosa and Escherichia coli. ^{55, 57} Six eyes grew fungal organisms, with three of the cases diagnosed as polymicrobial with a fungal component. There were two reported cases of Alternaria keratitis.^{59, 69} One of the Alternaria cases was co-infected with S. haemolyticus. ⁶⁹ In separate studies, keratitis due to microsporidia keratitis and Fusarium solani were also reported.^{70, 71} One polymicrobial fungal keratitis case was culture positive for S. epidermidis and Aspergillus, and the other was culture positive for S. aureus and Mucor.⁵⁹ Six eyes were diagnosed with viral keratitis.^72–75 All cases were proven or assumed herpes simplex virus (HSV) keratitis. One case was bilateral. Two studies reported cases of Acanthamoeba keratitis following corneal crosslinking.⁷⁶ One of which was polymicrobial with Acanthamoeba and S. aureus. ⁵⁹ Two studies reported microbiological results as inconclusive.^{77, 78} The distribution of the causative organisms is detailed in Table 2.

Table 2:

Causative organism

Organism	Number of Eyes (n=49)
Bacteria	31 (63.3%)
Fungus	3 (6.1%)
Virus	6 (12.3%)
Parasite	1 (2.0%)
Polymicrobial - Bacteria and Fungus	3 (6.1%)
Polymicrobial - Bacteria and Parasite	1 (2.0%)
Inconclusive Results	4 (8.2%)

All cases of infectious keratitis utilized a bandage contact lens post operatively for patient comfort and to promote re-epithelialization of the cornea. The most common class of antibiotics used for prophylaxis post-operatively was fluoroquinolones. Twenty-eight eyes were also prescribed topical corticosteroids. Post-operative topical NSAIDs were utilized in eight eyes. The most common day of presentation was on post-operative day 3. The average time of presentation overall was 4.8 days ± 5.4 days. The average time of presentation for cases with fungal keratitis or a polymicrobial keratitis with a fungal component was 8.8 days ± 10.1 days. This difference was not significant (P=0.250, t-test).

Data on final visual acuity after infection was available for 47 eyes out of the total 49 eyes. The average final BCVA (logMar) was 0.64 ± 0.70. Nine eyes required penetrating keratoplasty (PK).^{59, 60, 62, 63, 66, 69, 76, 78} Five eyes of those patients required an emergent therapeutic PK due to corneal perforation.^{59, 66, 76, 78} Twenty patients (21 eyes) had a history of atopy.^{53, 59, 62, 63, 65, 67, 68} Fourteen eyes infected with S. aureus had significant drug resistance, including fourth generation fluoroquinolones.^{59, 62, 63, 66, 68}

Discussion

The majority of post-operative infections presented within five days following the procedure. With standard epithelium-off and accelerated corneal crosslinking, this was attributed to the duration required for full re-epithelialization. Given this time course for infection, patients should have intense follow up in the immediate post-operative period. Other contributing factors for post-operative complications that were associated with infection included the use of bandage contact lenses and steroids.⁵³ In their six year study, Tzamalis et al. compared the effects of two treatment regimens post crosslinking in 1273 eyes. While the overall frequency of infection was low (0.71%), all nine cases of infectious keratitis occurred in the group that utilized a bandage contact lens and steroids immediately post operatively. Five of the nine cases ultimately required penetrating keratoplasty. They further went on to show that withholding a bandage contact lens and steroids until the epithelium had healed produced no significant difference in corneal haze. Thus, the authors concluded that waiting to initiate steroids and foregoing a bandage contact lens may reduce the incidence of infectious keratitis in epithelium-off crosslinking without impacting visual outcome.

Atopic diseases such as allergic eye disease, asthma, and eczema are common co-existing conditions in patients with keratoconus. Up to 35% of keratoconus patients have been reported to have a history of atopy.⁷⁹ However, in a recent comprehensive meta-analysis evaluating keratoconus prevalence worldwide and associated risk factors, atopy did not significantly increase the risk of developing keratoconus.⁸⁰ This finding conflicts with a prior report that suggested atopy was a significant predictor for keratoconus.⁸¹ Evidence exists to support that atopic conditions can worsen the course of infectious keratitis due to an abnormal immune response, colonization with MRSA, and the likelihood of patients utilizing chronic steroids. In support of this view, Nakata and colleagues found that 67% of patients with atopic dermatitis were colonized with MRSA versus 6% of non-atopic patients.⁸² Other studies have shown decreased tear IgA levels and decreased IFN-gamma levels in patients with atopic dermatitis.^{83, 84}

Shetty et al. reported a case series of four patients with moxifloxacin-resistant S. aureus keratitis following standard crosslinking.⁶² All patients had a positive history for atopic conditions and were using either steroids or immunosuppressants, which likely contributed to susceptibility to infection. Rana, et al. also reported two cases of young patients with keratoconus and allergic eye disease that experienced unusually fulminant courses of infectious keratitis following crosslinking.⁶⁵ In both cases, the patients were admitted for treatment. The cornea perforated on day 7 in the first case and within 48 hours in the second case. Both were treated with cyanoacrylate glue. The authors concluded that the increased inflammatory activity in vernal keratoconjunctivitis may have contributed to the extreme keratolysis in these cases.

Ting et al. later reported a case of infectious keratitis with S. aureus following accelerated corneal crosslinking in a 23 year old patient with keratoconus and eczema. After resolution of infection, the patient was diagnosed with acute corneal hydrops, which later resolved with antibiotics and corticosteroids.⁶⁷ Fasciani et al. reported a case of a MRSA keratitis post standard corneal crosslinking in a 22 year old male patient with atopic dermatitis and keratoconus.⁶³ The patient required a penetrating keratoplasty along with subtotal vitrectomy, anterior synechiolysis, iridectomy, and a sclerally fixed IOL implant. Visual acuity was not improved after multiple surgeries, resulting in hand motion vision. These data indicate that a detailed medical history is vital to identifying these higher risk conditions that may precipitate infection following crosslinking and be taken into consideration for appropriate patient selection.

Antibiotic resistant bacteria have also been recovered from cases of infectious keratitis post crosslinking. As mentioned, Shetty et al. first reported moxifloxacin-resistance in four cases.⁶² The authors hypothesized that the UVA light itself could have induced a mutation in these organisms. In a case series by Maharana et al., five out of six cases that were culture positive for S. aureus were resistant to moxifloxacin.⁵⁹ In the report by Fasciani et al.,⁶³ the strain of MRSA was found to be resistant to penicillin, moxifloxacin, gentamycin, ciprofloxacin, tobramycin, and oxacillin with reduced sensitivity to vancomycin and linezolid. Infection resolved with vancomycin 5% ophthalmic solution and IV linezolid and daptomycin. The authors suggested prophylaxis with 1.25% povidone-iodine solution in atopic patients. Interestingly, Oakey et al reported bilateral MRSA keratitis after epithelium-on corneal crosslinking in a 33 year old male patients with no history of atopy. ⁶⁶ Both eyes perforated and required emergent penetrating keratoplasty. Cultures grew pan-resistant MRSA. The patient was treated with vancomycin, trimethoprim/polymyxin B, oral doxycycline, ascorbate and artificial tears following both penetrating keratoplasties. Three S. aureus keratitis cases were also reported by Kodavoor et al following accelerated corneal crosslinking.⁶⁸ Two of the three cases were resistant to fourth generation fluoroquinolones. One case resolved with tobramycin in addition to moxifloxacin and the other two cases resolved with fortified cefazolin. Taken together, these data suggest that physicians must have a high suspicion for antibiotic resistance in infectious keratitis following corneal crosslinking and tailor antibiotics quickly and appropriately if no improvement is seen.

Reactivation of herpetic keratitis following corneal collagen cross-linking has also been reported in the literature. In one systematic review assessing effectiveness of cross-linking in the management of infectious keratitis, the two cases of HSV keratitis treated with cross-linking resulted in enhanced corneal melting and poor outcomes.²⁵ In 2007, Kymionis et al. reported a case of PCR-positive HSV keratitis in a 21 year old female following standard corneal cross-linking.⁷² The authors hypothesized that UVA light could be a stimulus to trigger reactivation of latent HSV infection. In this particular case, the patient had no previous history of ocular infection with HSV. While HSV reactivation due to UVA light has been reported in animal and human studies, the exact mechanism is poorly defined.^{85, 86}

Yuksel et al. also described a similar case following standard cross-linking in a 31 year old female with no known history of ocular herpetic disease or previous surgery.⁷³ No viral prophylaxis was given in either case as there was no history of infection. Al-Qarni and AlHarbi reported two additional cases in young male patients, both of which had good outcomes and recovery of BSCVA to pre-operative levels.⁷⁴ The authors observed that delayed healing is typically the first sign of a viral keratitis following cross-linking. They proposed that systemic antivirals should be added to the prophylactic regimen following surgery in patients with history of herpetic infection to decrease risk of reactivation. Recently, Sitaula et al. reported a bilateral case of viral keratitis diagnosed clinically in an 18 year old male following a bilateral standard cross-linking procedure.⁷⁵ The patient’s dendritic ulcer healed in 10 days with oral acyclovir and acyclovir ointment as well as ofloxacin. The authors cautioned against performing a bilateral cross-linking procedure and consistently with other reports, they proposed that prophylactic antivirals could be useful. Physicians should be aware of the risks of ocular herpes reactivation with UVA light and educate patients on this risk before the procedure.

Conclusion

Overall, the frequency of infectious keratitis after epithelium-off crosslinking is low. It should be noted however that a key limitation to the current study is the dependence on accurate reporting in the literature. Since the majority of corneal crosslinking procedures are performed in a private practice setting, the actual incidence of complications such as infectious keratitis may be underreported. The use of steroids and bandage contact lenses along with a history of herpetic or atopic disease have been associated with risk for post-operative infection in multiple studies. These patients require intense post-operative care with prophylactic anti-viral therapy when appropriate.