Post-keratoplasty Infectious Keratitis: Epidemiology, Risk Factors, Management, and Outcomes

Post-keratoplasty infectious keratitis (PKIK) represents a unique clinical entity that often poses significant diagnostic and therapeutic challenges. It carries a high risk of serious complications such as graft rejection and failure, and less commonly endophthalmitis. Topical corticosteroids are often required to reduce the risk of graft rejection but their use in PKIK may act as a double-edged sword, particularly in fungal infection. The increased uptake in lamellar keratoplasty in the recent years has also led to complications such as graft-host interface infectious keratitis (IIK), which is particularly difficult to manage. The reported incidence of PKIK differs considerably across different countries, with a higher incidence observed in developing countries (9.2–11.9%) than developed countries (0.02–7.9%). Common risk factors for PKIK include the use of topical corticosteroids, suture-related problems, ocular surface diseases and previous corneal infection. PKIK after penetrating keratoplasty or (deep) anterior lamellar keratoplasty is most commonly caused by ocular surface commensals, particularly Gramme-positive bacteria, whereas PKIK after endothelial keratoplasty is usually caused by Candida spp. Empirical broad-spectrum antimicrobial treatment is the mainstay of treatment for both PKIK, though surgical interventions are required in medically refractory cases (during the acute phase) and those affected by visually significant scarring (during the late phase). In this paper, we aim to provide a comprehensive overview on PKIK, encompassing the epidemiology, risk factors, causes, management and outcomes, and to propose a treatment algorithm for systematically managing this challenging condition.

Post-keratoplasty infectious keratitis (PKIK) represents a unique clinical entity that often poses significant diagnostic and therapeutic challenges. It carries a high risk of serious complications such as graft rejection and failure, and less commonly endophthalmitis. Topical corticosteroids are often required to reduce the risk of graft rejection but their use in PKIK may act as a double-edged sword, particularly in fungal infection. The increased uptake in lamellar keratoplasty in the recent years has also led to complications such as graft-host interface infectious keratitis (IIK), which is particularly difficult to manage. The reported incidence of PKIK differs considerably across different countries, with a higher incidence observed in developing countries (9.2-11.9%) than developed countries (0.02-7.9%). Common risk factors for PKIK include the use of topical corticosteroids, suture-related problems, ocular surface diseases and previous corneal infection. PKIK after penetrating keratoplasty or (deep) anterior lamellar keratoplasty is most commonly caused by ocular surface commensals, particularly Gramme-positive bacteria, whereas PKIK after endothelial keratoplasty is usually caused by Candida spp. Empirical broad-spectrum antimicrobial treatment is the mainstay of treatment for both PKIK, though surgical interventions are required in medically refractory cases (during the acute phase) and those affected by visually significant scarring (during the late phase). In this paper, we aim to provide a comprehensive overview on PKIK, encompassing the epidemiology, risk factors, causes, management and outcomes, and to propose a treatment algorithm for systematically managing this challenging condition.

INTRODUCTION
Corneal opacity is the 5th leading cause of blindness globally, with around 6 million of the population being affected (1)(2)(3). Among all aetiologies, infectious keratitis (IK), also known as microbial keratitis or infectious corneal ulceration, consistently features as the most common culprit of corneal blindness, particularly in the developing countries. IK is a painful and potentially blinding condition that may require hospital admission for intensive medical treatment and/or surgical interventions (1,2). It can be caused by a wide array of organisms, including bacteria, fungi, viruses, parasites or mixed infection (1,4,5). Nonetheless, as the ocular surface is equipped with a multifaceted defence system (6,7), IK rarely occurs in the absence of any predisposing factor. Commonly reported risk factors include contact lens wear, trauma, ocular surface disease and ocular surgery, particularly keratoplasty (1,2,8,9).
Post-keratoplasty infectious keratitis (PKIK) represents a challenging clinical entity that often poses significant diagnostic and therapeutic challenges. It is uniquely different from a "standard" IK in several ways. Firstly, the occurrence of IK in a graft can result in potentially devastating complications such as graft rejection, failure and endophthalmitis (10)(11)(12). Secondly, topical corticosteroids are often required to reduce the risk of graft rejection. However, in the event of PKIK, the use of topical corticosteroids may act as a double-edged sword as it can worsen the infection during the acute phase, particularly in fungal infection. Furthermore, there has been a paradigm shift in keratoplasty in the past decade where lamellar keratoplasty such as deep anterior lamellar keratoplasty (DALK), Descemet stripping automated endothelial keratoplasty (DSAEK), pre-Descemet's endothelial keratoplasty (PDEK), and Descemet membrane endothelial keratoplasty (DMEK) have superseded penetrating keratoplasty (PKP) as the preferred choice of keratoplasty for anterior and posterior corneal pathologies (13)(14)(15)(16)(17)(18)(19). However, this has also resulted in a number of complications that are not usually observed following PKP, including graft detachment and interface infectious keratitis (IIK) (20)(21)(22)(23). IIK, a unique subtype of PKIK which can develop after ALK or EK, is a difficult-to-treat condition as the sequestration of the infective microorganisms at the graft-host interface hinders access for obtaining samples for microbiological culture and for topical antimicrobial treatment to effectively reach and treat the affected site.
As corneal transplant is the most commonly performed type of transplant worldwide, the occurrence of PKIK and its resultant complications have a significant impact. In this paper, we aim to provide a comprehensive overview on PKIK, encompassing the epidemiology, risk factors, causes, management and outcomes, and to propose a treatment algorithm for systematically managing this challenging condition.

METHOD OF LITERATURE SEARCH
We searched PubMed (January 1980-2021) for relevant articles related to IK after keratoplasty. Keywords such as "infectious keratitis, " "corneal ulcer, " "corneal infection,n, "microbial keratitis, " "keratoplasty, " "corneal transplantation, " and "corneal graft" were used. Only articles published in English were included for the review. The search was first performed on 10 September 2020 and was last updated on 05 January 2021. The literature search retrieved 328 articles, of which the abstracts and titles were screened for those that fulfilled the eligibility criteria. After excluding ineligible studies, 48 were included in the qualitative synthesis. A PRISMA flow chart is provided in Supplementary Figure 1. The demographic factors, clinical characteristics and outcomes of PKIK of large case series (>500 cases) are summarised in Table 1.

Incidence/Prevalence
The incidence of PKIK differs considerably across different countries, with a higher incidence observed in developing countries than developed countries. Depending on the study design, patient cohort and follow-up duration, the incidence of PKIK in developed countries ranges from 0.02 to 7.9% (Table 1) (26,28,29,31,33,(36)(37)(38)(39)(40)(41)(42)(43)(44). The incidence of PKIK in developing countries is less well documented within the literature, with a higher overall incidence of up to 9.2-11.9% (35,45). The higher incidence may be attributable to the reduced access to healthcare, poor follow-up compliance, lower level of education, increased risk due to trauma and poor hygiene, poverty and a higher proportion of primary keratoplasty performed as therapeutic keratoplasty for IK (35,45).
Age PKIK affects patients of all age groups, with the majority of cases reported in the literature being between the range of 17-95 years (12,26,28,33,34,36,(46)(47)(48)(49)(50). This reflects the varied indications for keratoplasty such as keratoconus, pseudophakic bullous keratopathy, Fuchs endothelial dystrophy and corneal ulceration, scarring, or perforation within the adult population (12,51). In comparison, the main indications for keratoplasty within the paediatric population include keratoconus, regraft, and herpes simplex keratitis (HSK), with a higher preponderance of congenital conditions such as anterior segment dysgenesis (including Peter's anomaly), congenital hereditary endothelial dystrophy and sclerocornea, amongst others (52)(53)(54). A study in Denmark evaluated keratoplasties performed in children under 16 over a 40-year period and found that infection was responsible for 20% of failed grafts (55). In a longitudinal retrospective study of 168 paediatric eyes in India, PKIK occurred in 29% of eyes and was responsible for 50% of failed grafts (45). However, the most common indication for keratoplasty in this study was infectious keratitis (43%), which was associated with a high recurrence rate (56).

Socioeconomic Status, Level of Education, and Occupation
Patients from rural regions with lower socioeconomic status and lower levels of education have reduced access to healthcare and are less likely to attend follow-up appointments following keratoplasty. This is reflected within a study conducted in India whereby 75% of patients were from rural communities, with a high rate (28%) of PKIK being observed (45). Additionally, a China study reported farmers to be a significant independent risk factor of PKIK due to higher risk of trauma, particularly from plants resulting in fungal keratitis (57). These populations also have a higher risk for non-compliance of post-operative medication administration and hygiene (58).

Influence of the Types of Keratoplasty
The types of keratoplasty, including PKP, DALK and EK, have also been shown to greatly influence the incidence, risk and types of PKIK (i.e. ocular surface-related infection or IIK). A large retrospective cohort study of 2,098 keratoplasty performed between 2007 and 2018 in the US observed a PKIK incidence of 5.9% and 1.3% following PKP and EK, respectively (24). The higher proportion of PKIK occurring in PKP (93%) compared to DALK (6%) and EK (0%) was similarly depicted in a UK study of 1,508 grafts (25). The higher proportion of PKIK after PKP hinges on a combination of factors, including the indication for surgery, the use of sutures, and the prolonged use of topical corticosteroids. The indications for EK tend to be non-infective causes such as endothelial dystrophy, whereas a wider range of ocular comorbidities indicated for PKP may include IK and repeat (high-risk) grafts (24). Additionally, the requirement for corneal sutures in PKP, compared to EK, poses substantial risk of IK. This however does not completely explain the difference between PKP and DALK, with both procedures requiring the same number of sutures, though DALK usually does not require long-term topical corticosteroids due to zero-risk of endothelial graft rejection (59).
Conversely, a retrospective study using data from the Eye Bank Association of America analysing all adverse events of corneal grafts found a higher proportion of PKIK in EK (67%) compared to PKP (29%) and ALK (3%) (29). However, it is important to note that this study only included cases of PKIK that were caused by graft-transmitted infection. In addition, when taking into account the total number of each procedure performed (PKP/ALK/EK), the incidence of graft-transmitted infection was similar between EK and ALK (both 2.6 cases per 10,000 grafts) but higher than PKP (0.9 per 10,000 grafts). Interestingly, a higher rate of fungal infection was observed when compared to non-US studies, possibly related to the lack of antifungal agent in the corneal storage medium in the US. (11,29). The authors also noted a 1.5-3 times higher risk of fungal infection following EK (compared to ALK and PKP), potentially related to the increased warming time associated with the preparation of EK tissues in the eye bank (29).

RISK FACTORS Topical Corticosteroids
Topical corticosteroids are usually administered following keratoplasty to reduce the risk of graft rejection (60). As such, the majority of studies have found topical corticosteroids to be the main contributing factor (72.2-100%) for PKIK, primarily attributed to its local immunosuppressive effect (12, 25, 32-34, 41, 42, 47, 50) (Figure 1A). A US study observed that 82.6% eyes that developed PKIK were on topical corticosteroid therapy, of which the rates between PKP and EK were comparative at 81.3 and 90.9%, respectively (24). Constantinou et al. (30) performed a retrospective study evaluating non-suture-related PKIK after PKP between 1998 and 2008 in Australia. Long-term topical corticosteroids use was noted in 88% of eyes with PKIK, with 61% eyes developed infection more than 2 years after PKP. Similarly, a UK study (25) observed 89% of their patients developed PKIK (after PKP or DALK) whilst on topical corticosteroids, with a median time of developing IK at 25 months post-keratoplasty (25). This is an interesting observation as one would expect PKIK to develop sooner if the use of topical corticosteroids is directly implicated in the pathogenesis of PKIK since it is often used at a higher frequency and dose during the early postoperative period. Plausible explanations for late occurrence of PKIK include the occurrence of loose or broken sutures, the development or exacerbation of ocular surface diseases such as dry eyes and neurotrophic keratopathy (with persistent epithelial defect), and graft failure with resultant bullous keratopathy (25,30). In addition, while many studies reported the association of PKIK and use of topical corticosteroids, they did not examine the proportion of grafts that did not develop PKIK while on topical corticosteroids. Future studies examining the incidence of PKIK in all corneal grafts while on topical corticosteroids (including those that did not develop PKIK) would be of clinical interest.

Suture-Related Problems
Suture-related problems are a major risk factor for IK and has been implicated in 20-50% cases of PKIK, mainly after PKP and DALK (12,34,42,61,62) (Figure 1B). Furthermore, suture complications increase the risk of graft rejection and failure (42,63). The occurrence rate of PKIK caused by loose or broken sutures is reported to be as high as 71.6% (10,12,25,26,28,36,39,43,49). Four main causes of suture loosening have been described, which include corneal deturgescence, incomplete epithelialisation over the suture material, suture degradation (exacerbated by corneal vascularisation around the sutures), and cheese-wiring (39). The broken or loose suture is implicated in PKIK by causing a resultant epithelial defect that can be contaminated by environmental and ocular surface commensals (39). Cheese-wiring is seen particularly in corneas with keratoconus whereby little support is offered by the thin host cornea (39). In addition, patients with keratoconus are often affected by atopic disease, which increases the postoperative risk of corneal vascularisation around the graft sutures, loose/broken sutures, and graft failure (64). A greater propensity for suturerelated infections seems to occur within the interpalpebral zone, likely due to the increased risk of exposure and reduced protection of the eyelids (34,42).
Suture-related problems generally occur either within 1 year or around 30 months following keratoplasty (37,39,42,63,65,66). The reason for the bimodal peak noted in these studies may be due in part to the process of suture-loosening and surgeons' preference as some may remove all corneal graft sutures at 12-18 months post-keratoplasty. Corneal deturgescence and incomplete epithelialization over the suture may result in an earlier onset of infection, whereas suture degradation and cheesewiring of the corneal tissue contribute to a later onset (39,42). Christo et al. advocated the removal of sutures as soon as the graft-wound interface is healed at 1 year for vascularised recipients and 18 months for all other cases to minimise the risk of suture-related PKIK, with earlier removal in children (39,67). However, individualised care is necessary due to variable speed in wound healing (e.g., slower in elderly patients) whereby wound dehiscence or large changes in keratometry can occur upon premature suture-removal (39,68).

Previous History of IK
Keratoplasty serves as a useful therapeutic modality in managing patients with IK. It can be performed in the form of optical keratoplasty for visual rehabilitation (by removing the corneal scar) or in the form of therapeutic keratoplasty to manage active, medically refractory IK (15,69). However, the occurrence of PKIK following therapeutic keratoplasty is high (6-41%) and the risk may be influenced by the type of previous infection (56,(70)(71)(72). Wagoner et al. (12) observed 18.6% of those that developed bacterial PKIK were associated with a history of previous bacterial keratitis. Fungal recurrence rates are variable with a range of 7.4-32.7%, with most recurrences presenting within 2 weeks of surgery (73,74). Due to the propensity of fungi for deepseated infections with corneal penetration and anterior chamber invasion, the final outcomes of graft clarity (51-84%) and final cure rate requiring no further regrafts (69-90%) are reduced in comparison to a recurrence of bacterial keratitis (69-90% and 90-100%, respectively) (73)(74)(75)(76). Therapeutic keratoplasty performed for refractory Acanthamoeba keratitis (AK) is often unsuccessful in elimination of the infection, necessitating repeat grafts with resultant guarded outcome (71). In addition, recurrence of HSK post-keratoplasty is common, with ∼50% seen within the first 2 years following PKP and in 33% at 3 years for DALK, and is usually associated with a high risk of graft rejection and failure (77)(78)(79). However, it is noteworthy to mention that most of these studies were conducted more than 1-2 decades ago. Recognition of the high recurrent risk of HSK had led to increased use of prophylactic oral aciclovir postkeratoplasty, which could reduce the risk of HSK recurrence and resultant graft rejection/failure (80).

Cornea Preservation Method
Currently, two main methods are used to store and preserve donor corneas in the eye banks, including organ culture and hypothermia (83). Organ culture involves suspending corneal tissues in cell culture medium (most commonly Eagle's minimum essential medium) with foetal bovine serum. Antibiotics and antifungals are added to prevent growth of microorganisms (83). Additionally, frequent testing of the suspension medium for microbial growth is conducted to ensure the sterility before transplantation. This method is able to preserve corneas for up to 4 weeks at 28-37 • C (83). On the other hand, the hypothermia method utilises storage medium such as Optisol-GS (which contains dextran and chondroitin sulphate) to prevent corneal oedema and is able to preserve corneas for 7-14 days at a temperature of 2-8 • C (83). While presenting a clear advantage of its technical simplicity, the lower storage temperature and shorter storage duration may reduce the chance of microbial detection before transplantation, thereby increasing the risk of PKIK (84,85).
Hypothermic storage has been shown to have a higher positive rim culture rate (9.8%) compared to organ culture (1.3%) (84). Similarly, a Spanish study observed a 3.2% positive microbiological culture among 1,369 donor corneoscleral rims and found that 61.8% were related to corneas stored in hypothermia (86). Notably, the preferred hypothermic storage medium used in Europe and US contains only antibiotics (e.g., gentamicin) but without an antifungal agent as seen in the organ culture medium (29,85). This may explain the higher rates of PKIK due to Candida spp. following EK utilising hypothermicstored corneas at European and US centres (29,85). It is also worth noting that subsequent postoperative fungal infection is seen in 7% of the corneas with positive donor rim fungal culture (87). In view of these issues, addition of antifungal agent into the hypothermic storage medium has been proposed (88,89). However, further investigations into the efficacy, safety and choice of antifungal agent are required as the microbiological profiles may be highly varied across different regions and antifungal agent may cause significant toxicity to endothelial cells (88).

Viruses
Herpes simplex keratitis (HSK) represents an important cause for PKIK. Although accounting for <7% of microbial causes of PKIK (29,35), the incidence calculated by a Dutch study of 2,112 patients is 1.2 per 1,000 person-years (97). Comparatively, a Chinese study of 1,443 patients found the incidence to be 1.2% (98). Post-keratoplasty HSK can present as classic dendritic keratitis, geographic ulcer, or non-healing epithelial defect (97). Interestingly, HSK may develop following keratoplasty despite no previous diagnosis, with the majority presenting within 2 years of transplantation (97). This is likely to be attributable to the high seropositive rate within the population despite being asymptomatic and the suppression of local immunity with corticosteroids use (98,99). As such, HSV should be considered as a diagnosis in cases of non-resolving epithelial defects following keratoplasty. Cytomegalovirus (CMV) infection is another important cause of graft infection, rejection and failure following PKP and EK, though it is most commonly reported in East Asia (100,101). A recent UK study failed to identify the presence of CMV in any of the 92 cases of failed corneal graft tissues, suggesting that CMV may be a region-specific risk factor for graft infection and failure (102).

Others
Apart from recurrences following therapeutic keratoplasty, other types of IK, secondary to Acanthamoeba and acid-fast bacillus (e.g., non-tuberculous Mycobacterium), have rarely been reported after keratoplasty (12,33,103).

Microscopy, Culture, and Sensitivity
The diagnostic approach for PKIK is similar to "standard" IK, unless the PKIK is related to IIK. Corneal scraping for microscopy, culture and sensitivity testing represents the gold standard for diagnosing IK, though the culture yield varies between 24 and 77% (1,2,104,105). To maximise collection of microorganisms for culture, the corneal scrape should be taken at the ulcer base or leading edge. Microscopic examination with appropriate staining (e.g., Gramme stain, Giemsa stain, potassium hydroxide with calcofluor white) serves as a more rapid diagnostic method of IK (106). Various agars are used for culturing the causative microorganisms, including blood/chocolate agar (for bacteria), Sabouraud dextrose agar (for fungi), and non-nutrient agar with Escherichia coli overlay (for Acanthamoeba).
In suture-related PKIK cases, infected corneal sutures should also be sent for microbiological culture as they may provide additional information. Adler et al. (107) evaluated the presence of microbial growth and biofilm formation amongst corneal sutures removed following astigmatic correction (quiescent), loosening or breakage (exposed), or infection. Biofilms are composed of extracellular matrix secreted by microorganisms and are usually resistant to conventional antimicrobial treatment (108,109). They have been shown to form on biotic and abiotic surfaces, including sutures (107,110). In their study, the infection group demonstrated a culture yield of 60% based on corneal sutures, underlining corneal suture material (when infected) as a useful source for obtaining microbiological diagnosis. In addition, higher biofilm scores (on scanning electron microscopy) were observed in the infection and exposed groups, highlighting the importance of early removal of sutures to prevent suture-related PKIK.

In vivo Confocal Microscopy
In vivo confocal microscopy (IVCM) is a non-invasive diagnostic investigation allowing visualisation of the cornea at high resolutions of 1-2 µm by limiting scattered light and focusing the observation system to a single point (111). As such, it offers a useful tool for determining fungal keratitis and AK (but not bacterial keratitis). IVCM has the benefit of providing rapid diagnoses but the diagnostic accuracy is reliant on the operator's experience (112). Compared to culture, IVCM is able to detect fungal filaments with a sensitivity of 85.7-94% and a specificity of 78-81.4% (113,114). Similarly for AK, the sensitivity and specificity are 88.2-100 and 98.2-100%, respectively (113,115,116). Additionally, IVCM lends itself as an important investigation for determining the causes of IIK where access for corneal sampling is limited (22,117). Recently, artificial intelligence (AI)-assisted diagnosis based on IVCM images has been shown to reliably diagnose fungal keratitis (118,119).

Anterior Segment Optical Coherence Tomography
AS-OCT utilises low-coherence interferometry to provide highresolution, cross-sectional imaging of the cornea (120). As AS-OCT can provide a quantitative and objective measurement of the infection, it can be used to observe characteristic patterns of IK, determine the depth and extent of IK, and monitor the progression of IK and treatment response, especially in deep-seated infection or IIK (Figure 2) (121)(122)(123)(124). In addition, it is of particular importance in fungal keratitis as fungi have a propensity for deep-seated infection (which is more difficult to visualise on slit-lamp examination/photography) and a prolonged clinical course (125,126).

TREATMENT
The treatment strategy is guided by a number of factors, including the type of microorganism, severity, location, type of keratoplasty, and clinicians' experience and preference. A proposed systematic treatment algorithm of PKIK is illustrated in Figure 3.

Medical Treatment
Empirical broad-spectrum antimicrobial agents are the mainstay of treatment for PKIK following both PKP and LK. Commonly used topical antibiotics include fluoroquinolones and fortified cephalosporins and aminoglycosides. Depending on the type of fungal infection (filamentous vs. yeast), antifungal treatment such as natamycin, amphotericin B, and voriconazole are commonly administered. Early concurrent systemic antifungal treatment is often initiated in severe cases (127,128). Subsequent medical treatment is then tailored to the clinical progress and microbiological results. Akova et al. (36) reported successful medical management in 43% eyes of the 21 eyes with IK following PKP. Vajpayee et al. (62) observed a success rate of 74% in resolving PKIK with medical management alone. In the event where the use of topical steroids is contraindicated in grafted patients (e.g., infection or steroid responder), topical ciclosporin may serve as a useful substitute (129,130). IIK frequently poses a significant therapeutic challenge due to the deep location of the infective nidus and entrapment of organisms in the interface between host and donor tissues (Figure 4) (21,22,91). Topical medications often have limited penetration to the deep cornea and fail to reach a therapeutic concentration at the site of infection. Epithelial debridement may improve drug penetration. Although uncommon, the interface infection may be heralded by an ocular surface infection or may extend from the interface to the ocular surface, both of which allow for scraping for microbiological culture and better penetration of topical treatment (22,131). However, medical treatment alone has been shown to achieve successful eradication of infection in only 13.3-24.2% cases of IIK, with high proportion requiring surgical interventions (21,91,117,132).
Intrastromal injections or interface irrigation with antimicrobial agents may be used when the infection is not responding to topical treatment, particularly in deep-seated infection and IIK (Figure 4) (127,(132)(133)(134). Kalaiselvi et al. (135) demonstrated that intrastromal voriconazole injection was able to resolve 72% of deep recalcitrant fungal keratitis that did not respond to topical natamycin and natamycin drops. Tu and Hou (134) reported successful resolution of two cases of post-DSAEK fungal IIK with repeated intrastromal antifungal injection, obviating the need for PKP. However, it is important to bear in mind that excessive injection of treatment extending into the interface may weaken the graft-host attachment and risk graft detachment and endophthalmitis (91,127,134). In addition, a recent review showed that only 10% of the reported cases of post-DSAEK IIK resolved without any surgical intervention, highlighting the therapeutic challenge of this clinical entity (91). Interface irrigation with antibacterial agents such as vancomycin (5%) has reportedly been effective in clearing DALK-related IIK (132). Use of antifungal agents such as amphotericin B (0.15-0.5%), voriconazole (1.0-5.0%), and fluconazole (5%) have also been described (91,133,136,137). Apart from treatment, Wessel et al. (133) have suggested using the irrigation fluid obtained after interface wash for microbiological investigations. Although rare, risk of Descemet membrane (DM) perforation needs to be kept in mind in these cases (128,138,139).

Surgical Treatment
The choice of surgical treatment of PKIK is dependent on the extent of infection and types of primary keratoplasty. In cases of IK following PKP, therapeutic PKP is needed in large ulcers not responding to medical treatment whereas optical PKP can be performed at a later stage to remove significant IK scarring and improve vision (140). Studies have reported that an emergency TPK was required in ∼15-20% cases of severe PKIK and some (up to 10%) may even require evisceration if it progresses to endophthalmitis (36,62).
In cases with ALK, the choice of surgical treatment depends on the primary procedure. In cases where adequate host stroma is left behind in the primary procedure, such as in manual or automated ALK, the lamellar graft can be removed and the surface allowed to re-epithelialise whilst on antimicrobial treatment (141,142). This helps reduce the microbial load and facilitate the corneal healing (143). Once the infection has healed, secondary DALK may be considered (144)(145)(146). In cases where the primary procedure was DALK, the interface is more likely to provide a potential space for sequestration of infection resulting in recurrences later (147). Although clear grafts have been achieved in repeat DALK procedures following IIK post-DALK (136,148), recurrences of infections have been reported with cases then requiring PKP (147,149). Emergency therapeutic PKP may be required in cases with non-responding infiltrates and impending/actual perforations (139,147).
In cases of DSAEK, removal of donor lenticule may lead to intraocular spread of infection resulting in endophthalmitis (21). However, in the presence of concomitant endophthalmitis, removal of the donor lenticule helps by reducing the microbial load and aids faster resolution (150). As such, removal of donor lenticule is not recommended unless there is concomitant endophthalmitis. A repeat DSAEK can be considered once complete resolution of infection is achieved and if host cornea is clear (151). In medically refractory IIK post-DSAEK, an early excisional PKP (including the removal of the infected DSAEK) is advisable as it helps remove the interface infection and prevents intraocular spread of infection and subsequent endophthalmitis (152). Few cases of IIK have been reported after DMEK. Thompson et al. (153) reported a case of fungal keratitis and endophthalmitis post-DMEK. The authors removed the DMEK graft and administered intravitreal antifungal agents every alternate day until the infection resolved. DSAEK was then performed as a secondary procedure, which achieved a final best-corrected-visual-acuity (BCVA) of 6/18. Another case of post-DMEK interface fungal keratitis with endophthalmitis was reported by Doshi et al. (154). Initial conservative treatment with intracameral and intravitreal antifungals did not result in improvement. The patient was then subjected to pars plana vitrectomy and 3 mm of central plaque was removed from DMEK graft using a vitrector. Oral treatment with posaconazole was started and complete eradication of infection was observed at 2-month follow-up.
In the recent years, there has been an increasing popularity in the use of therapeutic corneal cross-linking (PACK-CXL) for treating bacterial and fungal keratitis, particularly in recalcitrant cases (155)(156)(157). Mikropoulos et al. (158) described an innovative use of PACK-CXL in managing a case of PKIK secondary to multidrug resistant fungal keratitis. PACK-CXL was applied to the infected graft and the affected corneoscleral rim intraoperatively followed by a same-day therapeutic keratoplasty. The graft remained free of infection during the 9-month followup. However, larger case series are required to examine the efficacy and safety of such approach.

CONCLUSION
PKIK is a clinical entity that often poses significant diagnostic and therapeutic challenges. It carries a high risk of serious complications such as graft rejection and failure, and less commonly endophthalmitis. PKIK after PKP and ALK is most commonly caused by ocular surface commensals, particularly Gramme-positive bacteria, whereas PKIK after EK is usually caused by Candida spp. Broken or loose sutures have been consistently shown to be main risk factor of PKIK and early suture removal is advocated whenever clinically possible. With the increased adoption of EK in the recent years, it is likely that the incidence of PKIK will reduce. Optimal management of ocular surface diseases such as dry eye, blepharitis, exposure keratopathy, and neurotrophic keratopathy will help reduce the risk of PKIK following PKP and DALK. Refinement in the preservation method (e.g., addition of antifungal agent in hypothermic method) may reduce the risk of graft-transmitted infection, particularly in EK. However, further investigations into the efficacy and the choice of antifungal agent are required as the microbiological profiles may be highly varied across different regions. A stepwise treatment strategy can often be used to successfully treat PKIK, though IIK often requires surgical interventions to achieve complete resolution of the infection.

AUTHOR CONTRIBUTIONS
DSJT: conceptualisation and supervision. AS, RD, and DSJT: data collection, curation, literature review, and manuscript drafting.