The arrows indicate when a sample for plating was taken; and the white boxes, 1-hour room temperature incubation; and the gray boxes, 4°C incubation.
The dotted line indicates the limit of detection at 10 colony-forming units/mL.
aSignificant difference from 0 hour by 1-way Kruskal-Wallis with the Dunn multiple comparison test.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Brothers KM, Shanks RMQ, Hurlbert S, Kowalski RP, Tu EY. Association Between Fungal Contamination and Eye Bank–Prepared Endothelial Keratoplasty Tissue: Temperature-Dependent Risk Factors and Antifungal Supplementation of Optisol–Gentamicin and Streptomycin. JAMA Ophthalmol. 2017;135(11):1184–1190. doi:https://doi.org/10.1001/jamaophthalmol.2017.3797
Does eyebank tissue processing increase the likelihood of fungal growth in donor rim cultures?
This study found that compared with uncut or unstripped tissue, eyebank preparation of donor corneal tissue for endothelial keratoplasty was a risk factor for fungal growth from donor rims cultures. Experimental findings in the absence of corneal tissue confirmed increased room temperature incubation time promotes growth of Candida species in optisol–gentamicin and streptomycin corneal storage media, and the addition of antifungals reduced growth of Candida in a species dependent manner.
These findings suggest reduced room temperature incubation and addition of antifungals to optisol–gentamicin and streptomycin should be considered when processing corneal tissues.
Fungal contamination and infection from donor tissues processed for endothelial keratoplasty is a growing concern, prompting analysis of donor tissues after processing.
To determine whether eyebank-processed endothelial keratoplasty tissue is at higher risk of contamination than unprocessed tissue and to model eyebank processing with regard to room temperature exposure on Candida growth in optisol–gentamicin and streptomycin (GS) with and without antifungal supplementation.
Design, Setting, and Participants
An examination of the 2013 Eversight Eyebank Study follow-up database for risk factors associated with post-keratoplasty infection identified an increased risk of positive fungal rim culture results in tissue processed for endothelial keratoplasty vs unprocessed tissue. Processing steps at room temperature were hypothesized as a potential risk factor for promotion of fungal growth between these 2 processes. Candida albicans, Candida glabrata, and Candida parapsilosis endophthalmitis isolates were each inoculated into optisol-GS and subjected to 2 different room temperature incubation regimens reflective of current corneal tissue handling protocols.
Main Outcomes and Measures
Eversight Eyebank Study outcomes and measures were follow-up inquiries from 6592 corneal transplants. Efficacy study outcomes and measures were fungal colony–forming units from inoculated vials of optisol-GS taken at 2 different processing temperatures.
Donor rim culture results were 3 times more likely to be positive for fungi in endothelial keratoplasty–processed eyes (1.14%) than for other uses (0.37%) (difference, 0.77%; 95% CI, 0.17-.1.37) (P = .009). In vitro, increased room temperature incubation of optisol-GS increased growth of Candida species over time. The addition of caspofungin and voriconazole decreased growth of Candida in a species-dependent manner.
Conclusions and Relevance
Detectable Candida growth in donor rim cultures, associated with a higher rate of post keratoplasty infection, is seen in endothelial keratoplasty tissue vs other uses at the time of transplantation, likely owing in part to eyebank preparation processes extending the time of tissue warming. Reduced room temperature incubation and the addition of antifungal agents decreased growth of Candida species in optisol-GS and should be further explored to reduce the risk of infection.
Descemet stripping automated endothelial keratoplasty (DSAEK) constitutes a significant majority of corneal transplants performed in the United States.1 Improved patient outcomes were a primary factor in the rise of DSAEK; provision of tissue precut by the eyebank has improved access of DSAEK tissue. However, stromal fungal interface infections, almost unique to DSAEK, have been increasingly reported raising concern because of their often poor outcomes.1-6
A nonstatistically significant trend toward greater fungal infection rates in DSAEK vs penetrating keratoplasty (PKP) has been seen, and persistent reports of these infections led to enhanced tracking of corneal rim cultures and infections starting in 2013.7 This surveillance identified an elevated risk of infection in endothelial keratoplasty (EK) vs other uses and a higher rate of infection in precut tissue.8 Unlike postkeratoplasty bacterial infections, significant correlation of positive fungal donor rim culture results with the development of subsequent infection is well known and suggests that the source of these infections is contamination from the donor tissue.9,10 A cluster of fungal interface infections at our eyebank prompted a subanalysis of our data as well as a laboratory investigation of a potential link between tissue-processing practices and increased fungal contamination of corneas prepared for EK in comparison with other uses, and the effect of anti-fungal additives on this phenomenon.
Data from standard corneal tissue transplant follow-up queries at Eversight Eyebanks, collected as part of an established Eye Bank Association of America quality assurance program, was analyzed for corneal transplant tissue placed in 2013. This included corneal rim cultures performed, microbiologic isolate information analyzed, and resultant infections. De-identified data collected as part of a quality assurance program for corneal transplantation is, by definition, exempted from institutional review board at our institution. Patient consent was waived because the donor rim cultures were from donor tissue. The total number and individual types of corneal transplant tissues provided by Eversight, as well as individual surgeon or surgical center follow-up forms returned, corneal rim cultures reported as being performed, and, when positive, organism(s) isolated, were extracted. Additional queries were sent to surgeons to maximize the total number of corneal rim cultures, as well as both positive and negative rim cultures. Comparisons between groups of surgeons receiving tissue processed by the eyebank for endothelial transplant vs full-thickness tissue with regard to infection rate, culture rim positivity, and type of organism isolated were performed with a Pearson χ2 test with a significant P value of less than .05.
The most common postkeratoplasty fungal pathogen being Candida species,1Candida albicans, Candida glabrata, and Candida parapsilosis endophthalmitis isolates were chosen for this study. Isolates were retrieved from a clinical bank of deidentified isolates saved for diagnostic and susceptibility testing. Stocks were streaked to single colonies on yeast extract peptone dextrose (YPD) agar (20 g/L of peptone, 10 g/L of yeast extract, 20 g/L of glucose, 20 g/L of agar) supplemented with 10 μg/mL of tetracycline to prevent growth of bacteria. Cultures were prepared by inoculating single colonies into YPD broth and were grown overnight at 30°C with shaking. Cells were counted with a hemocytometer to determine fungal concentration and the inocula in colony-forming units per milliliter were confirmed with standard colony count determinations.
Optisol–gentamicin and streptomycin (GS) corneal storage media containing the entire volume (20 mL) (n = 6) were inoculated with a mean of 4.2 × 103 CFU/mL of C albicans, C glabrata, and C parapsilosis. Two different temperature incubation schedules were used (regimens A and B) to mimic different presurgical cornea processing protocols (Figure 1). For regimen A, incubation at room temperature for 1 hour followed by storage at 4°C was conducted on day 1. Incubation at room temperature for 2 hours followed by storage at 4°C was conducted on day 2. Incubation at room temperature for 3 hours followed by storage at 4°C was conducted on day 3. Incubation at room temperature for 2 hours followed by storage at 4°C was conducted on day 4. And incubation at room temperature for 1.5 hours was conducted on day 5. Regimen B was similar to A, except that samples were warmed to room temperature only on days 1, 2, and 5 (Figure 1). Three independent cultures per treatment group were grown for regimen A and regimen B. The experiment was performed on 3 separate occasions with 2 samples for each regimen A treatment group and 1 sample for each regimen B treatment group.
For both treatment schedules, 300-μL aliquots were taken at time 0 and within 5 minutes of each room temperature incubation (Figure 1). Samples were serially diluted and 100 μL was plated on to YPD agar to determine viable fungal CFU per milliliter. The limit of detection for this assay was 10 CFU/mL. Room temperature in the laboratory ranged from 20 to 22°C. The pH indicator color of optisol-GS remained unchanged throughout the duration of the experiment. No contamination was detected in the control vials (n = 3) containing no Candida species that were processed according to regimen A.
An additional condition to the experiment was completed using regimen A in which vials (n = 6) were inoculated with Candida species here, but supplemented with either 50 µg/mL of caspofungin or 50 µg/mL of voriconazole.11 As a negative control, an equal volume of sterile water (200 µL) was added to the untreated and regimen B vials to keep all volumes at a constant 20 mL.
Colony count data were log10 transformed using Microsoft Excel and graphed using GraphPad Prism. Kruskal-Wallis with the Dunn multiple comparison test and 2-way analysis of variance were used to determine statistical significance, which was set to P < .05. Significance from the 0 hour point is denoted in Figure 2.
A total of 6592 corneal transplants from the Eversight Eyebank system in 2013 were used with 2550 prepared for EK, either DSAEK or stripped for Descemet membrane EK (DMEK), and 4042 distributed for either anterior or full-thickness (PKP). A total of 12 infections (0.18%) were reported, with 7 occurring in EK (5 fungal and 2 no growth but presumed fungal) and 5 occurring in PKP procedures (3 bacterial, 1 fungal, and 1 unknown). The total number of follow-up forms returned was 4084 (62%), with tissue cultures reported as performed in 2623 cases (40%).
Culture-positive rims were reported in 46 of 2550 tissues (1.8%) prepared for EK and 43 of 4042 (1.1%) of anterior lamellar keratoplasty or PKP tissues, which was statistically significant (P = .006) (Table 1). Because only fungal rim contamination has been associated with perioperative infection, a focus on yeast contamination found 21 of 2550 EK rim cultures (0.82%) were positive, which was statistically higher than the 7 of 4042 corneal rims (0.17%) for other uses (P<.001).
Because of the potential for positive reporting bias for those cultures that were either positive or resulted in an infection from surgeons who did not routinely report culture results, a sensitivity analysis was performed where the analysis was further restricted to those centers or surgeons that reported culturing at least 50% of their donor rims to reduce the effect of those centers that either did not culture or did not report (Table 2). This group represented 3010 total corneal transplants for which 2732 forms (90%) were returned and 2398 cultures (80%) reported. A difference remained for 19 of 1664 EK tissues (1.14%) vs 5 of 1346 (0.37%) of other keratoplasty tissues that were culture positive for yeast (difference, 0.77%; 95% CI, 0.17-1.37) (P = .009).
Donors prepared for EK demonstrated a higher rate of contamination in comparison with other corneal tissue prompting an examination of differences in handling and processing between these 2 groups. Because previous work had identified room temperature exposure during tissue processing as promoting the bactericidal effect of optisol-GS, but also promoting growth of organisms not inhibited by either gentamicin or streptomycin (GS),12 a laboratory study to simulate the handling of processed endothelial tissue (DMEK or DSAEK) in comparison with minimally processed (deep anterior lamellar keratoplasty, PKP, or keratolimbal allograft) tissue based on an internal survey of Eversight protocols and practices was performed.
Temperature cycling regimen A stored the vials at 4°C, except for daily incubations at room temperature. This totaled 9.5 hours of room temperature incubation over 5 days (Figure 1). The second procedure, regimen B, was similar, but with room temperature incubations only on days 1, 2, and 5, for a total of 4.5 hours of room temperature incubation (Figure 1).
The initial inoculum of C albicans was a mean (SD) of 3.4 (1.7 × 103) CFU/mL. With regimen A, the mean (SD) number of viable C albicans increased to 2.1 (2.7 × 105 CFU/mL) at the end of the experiment on day 5 (97.5 hours) (P <.01) (Figure 2A). To test the efficacy of antifungal agents, vials of optisol-GS were supplemented with 50 μg/mL of caspofungin and 50 μg/mL of voriconazole and processed according to regimen A. Caspofungin reduced viable C albicans to a mean (SD) of 48 (56) CFU/mL by 97.5 hours (P < .05) (Figure 2A). Voriconazole reduced viable fungal growth at 48 hours to the limit of detection of this assay (10 CFU/mL) (P < .01), growth increased to a mean (SD) of 0.95 (1.3 × 103) CFU/mL at 51 hours, but decreased to 25 (34) CFU/mL by 97.5 hours (Figure 2A). This decrease in CFU per milliliter at 48 hours was interesting to note because a 3-hour room temperature incubation followed, resulting in an increase in CFU per milliliter, suggesting there was some growth after room temperature exposure even in the presence of an antifungal agent. Although voriconazole resulted in a small increase of viable C albicans at 51 hours, the use of both antifungals in optisol-GS corneal storage media reduced growth of C albicans compared with the untreated (C albicans mock [untreated] mean [SD] 2.06 [2.74 × 105] CFU/mL compared with caspofungin treated mean [SD] 48.3  CFU/mL and Voriconazole treated at mean [SD] 25  CFU/mL [P < .05]). Regimen B, with only 4.5 hours of room temperature incubation compared with 9.5 hours with regimen A, supported a mean (SD) growth of only 4.9 (5.0 × 103) CFU/mL by 97.5 hours and was not different compared with the initial inoculum (P > .05) (Figure 3).
For C glabrata, the initial mean (SD) inoculum was 5.3 (2.1 × 103) CFU/mL. With regimen A, the viable yeast cells increased to mean (SD) of 1.5 (1.4 × 105) CFU/mL at the end of the experiment (P < .001) (Figure 2B). Caspofungin reduced C glabrata below the limit of detection at 25 hours C. glabrata mock (untreated) at 8.73 (4.45 × 103) CFU/mL compared with caspofungin treated at mean (SD) of 10 (0) CFU/mL (limit of detection) (P <.001), which lasted throughout the remainder of the experiment (Figure 2B). Voriconazole prevented growth of C glabrata with a final mean (SD) concentration of 3.4 (1.1 × 103) CFU/mL at 97.5 hours (P < .05) (Figure 2B). The use of caspofungin in optisol-GS reduced growth of C glabrata compared with the untreated (P < .05). Regimen B resulted in a small but insignificant increase in CFU per milliliter to a mean (SD) of 1.3 (0.86 × 104) by 97.5 hours (P > .05) (Figure 3).
The mean (SD) initial inoculum for C parapsilosis was 4.0 (1.2 × 103) CFU/mL. Following regimen A, viable C parapsilosis increased to a mean (SD) of 4.0 (4.4 × 105) CFU/mL by 97.5 hours (P < .001) (Figure 2C). Caspofungin inhibited growth of C parapsilosis to a final mean (SD) of 2.8 (1.4 × 103) CFU/mL by 97.5 hours (P > .05) (Figure 2C). Voriconazole reduced viable C parapsilosis to a mean (SD) of 70 (95) CFU/mL at 24 hours, but growth increased to 1.5 (2.0 × 103) CFU/mL by 97.5 hours (P < .001) (Figure 2C). Despite the static growth of C parapsilosis with the addition of caspofungin, both antifungals resulted in less growth of C parapsilosis compared with the untreated (P < .05). With regimen B, the viable yeast cells had an insignificant increase from a mean (SD) of 4.0 (1.2 × 103) CFU/mL to 1.9 (1.6 × 104) CFU/mL by 97.5 hours (P > .05) (Figure 3).
This study identified an increased risk of detectable donor tissue contamination in corneas processed in advance for EK in contrast to other uses. The probable mechanism for this increased risk is increased room temperature exposure during processing. Supplementation with antifungal agents reduces this risk in an established in vitro time-temperature model of optisol-GS.12 While EK offers functional and safety advantages over PKP for isolated diseases of the corneal endothelium, postkeratoplasty infection remains one of its most feared complications, sometimes leading to loss of the eye.2-6 A comprehensive review of the Eye Bank Association of America adverse reporting system during 2007 to 2010 revealed a trend toward increased infection in eyes undergoing DSAEK (0.0022%) compared with PKP (0.0012%).7 Many factors specific to EK are suggested as causative factors including the unique lamellar interface sheltered from the eye’s normal aqueous circulation and immune defense. To our knowledge, increases in fungal culture positivity in tissues precut for EK have not been previously identified.
Of the 6592 corneas transplanted from Eversight in 2013, 2623 (40%) were reported as cultured, 89 of 2623 (3.4%) were reported as culture positive with 32 of 2623 (1.2%) positive for fungus, consistent with prior literature,9,10,13 but statistically higher in precut tissues (0.82%) in comparison with unprocessed tissues (0.17%) (P> .001). The rate for overall infections was higher in EK (0.27%) vs other procedures (0.12%), with all EK infections proven or presumed to be fungi in contrast to only 1 of 5 PKP infections. Our findings showed that only 40% of transplanted tissues were reported as cultured, but only 60% of follow-up forms were returned, leaving a potentially large pool of unreported data. For this reason, we performed a subanalysis of those surgeons who reported cultures from at least 50% of tissues distributed to them to control for any positive reporting bias. In this group of 3010 corneas, 2732 forms (90%) were returned and 2398 cultures (80%) reported, which likely reflects that those who culture do so consistently. In this group, positive cultures for yeast were reported in 1.14% of tissues processed for EK and only 0.37% in unprocessed tissues (difference, 0.77%; 95% CI 0.17-1.37) (P = .009). Regardless, detectable fungal growth is more common in EK tissue than in tissue for other uses.
Positive fungal donor rim cultures are associated with a higher risk of keratoplasty-related fungal infection as well as a concordance of cultured pathogens between donor rim and host in penetrating keratoplasty.9,10 Keyhani et al13 reported that 28 of 2466 donor rims were positive for fungus (all Candida species) from the New York Eye and Ear Infirmary resulting in 4 yeast infections (14%) all in fungus culture–positive donors. A meta-analysis by Hassan and Wilhelmus10 found 2459 of 17 614 donor rims positive, with a total of 31 cases of endophthalmitis demonstrating a 12-fold greater risk of infection if a donor rim culture was positive. A total of 21 cases had donor rim/endophthalmitis cultures concordant, with 10 of these being a Candida species for a predictive value of 1% for bacterial isolates and 3% for fungal isolates. Aldave et al7 reported that, of the eyes developing infection, 16 of 22 (73%) of the mate corneas had culture-positive rims, with 10 of 15 (67%) of those developing infection. These studies and the 3 times greater detectable tissue contamination at the time of transplantation suggest that the increased rate of fungal infection after DSAEK is, at least in part, due to the higher rate of detectable tissue contamination, leading to a proportional increase in postkeratoplasty fungal infection.
Eyebank procedures and facilities are strictly designed to avoid contamination and these processes are confirmed through rigorous testing on a regular basis. Positive rim cultures, disparate species, and infections were sourced from many different facilities, making it unlikely that breakdown of these processes occurred at a particular facility. For these reasons, we tested the possibility that source contamination could occur through greater room temperature exposure of EK processing compared with tissues not undergoing this processing. This method was based on an established time-temperature model for optisol-GS, modified to reflect current tissue-handling processes.12
In this model, the number of viable C albicans, C glabrata, and C parapsilosis in optisol-GS corneal storage media increased 1.8-, 1.4-, and 2-log10, respectively (P < .05), with daily room temperature incubations (regimen A). Less frequent room temperature incubations (regimen B) did not result in significant increases in CFU per millilter (0.25-, 0.48-, and 0.74-log10 increases, respectively). These data support that limiting room temperature incubation is important for preventing growth of pathogenic yeast during presurgical corneal processing.
Supplementation of optisol-GS with antifungal agents is a potential way to reduce contamination and inadvertent infections.11 Caspofungin was effective at reducing CFU per millilter of C albicans, C glabrata, and C parapsilosis at 1.8-, 2.7-, and 0.14-log10 decreases, respectively, when compared with the untreated. Voriconazole reduced CFU per milliliter at 2.1-, 0.19-, and 0.42-log10 decreases, respectively. Although neither antifungal resulted in a 3-log decrease in CFU per milliliter indicating it was fungicidal, these results indicate the addition of antifungals could reduce outgrowth of Candida species in optisol-GS corneal storage media. The observation that caspofungin and voriconazole were not equally effective against the 3 Candida species endophthalmitis isolates used in this study suggests that the use of a combination of antifungal agents or other antimicrobial agents may be required to eliminate contaminating pathogenic yeast.
The limitations of this study are both its in vitro nature and the use of noncorneal tissue–containing media. This is in part mitigated by the survey evidence of increased contamination of EK-processed tissues over minimally processed tissues as well as our previous study demonstrating that the inclusion of corneal tissue did not alter the results of this time-temperature model of microbial growth.12 Also, we did not exclude the possibility of contamination during processing, but the protocols and procedures are rigorously tested frequently through internal and external examination at each individual facility. No mechanisms existed to accurately track the room temperature exposure of individual tissues. Further, individual culture methods at the different surgical facilities would also be expected to yield different rates of detection, which is why we did not take this into account for interpretation of our studies.
This study provides evidence that donor rim processing for EK increases fungal contamination that translates to a higher risk of post-DSAEK infection independent of the unique presence of a lamellar interface. We also demonstrate that increased room temperature exposure, integral to eyebank EK tissue processing, amplifies contaminant Candida CFUs in optisol-GS and that antifungal drug supplementation of the solution can reduce this rise.
These findings may, in part or in whole, explain the increased rate of infection and contaminations observed in EK, precut tissues, and mate corneas processed for EK. The findings provide a rationale for tightly tracking and reducing room temperature exposure of transplant tissue as well as exploring antifungal drug augmentation of corneal storage media to reduce the risk of fungal infection with EK for both DSAEK and DMEK.
Corresponding Author: Elmer Y. Tu, MD, Department of Ophthalmology and Visual Sciences, University of Illinois at Chicago, 1855 W Taylor St (M/C 648), Chicago, IL 60612 (email@example.com).
Accepted for Publication: August 6, 2017.
Published Online: September 28, 2017. doi:10.1001/jamaophthalmol.2017.3797
Author Contributions: Drs Brothers and Tu had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Brothers, Shanks, Kowalski, Tu.
Acquisition, analysis, or interpretation of data: Brothers, Shanks, Hurlbert, Tu.
Drafting of the manuscript: Brothers, Shanks, Tu.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Brothers, Shanks, Tu.
Obtained funding: Shanks, Tu.
Administrative, technical, or material support: All authors.
Study Supervision: Brothers, Shanks, Tu.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.
Funding/Support: This work was supported by a 2015 Lindstrom/Eye Bank Association of America grant and logistical/material support from Eversight and National Institutes of Health grant EY024785 (Dr Brothers) and National Eye Institute core grants EY08098 and EY027331 (Dr Shanks).
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Meeting Presentation: This study was presented at the Eye Bank Association of America 2015 Fall Educational Symposium; November 12, 2015; Las Vegas, Nevada.
Additional Contributions: We thank Eric G. Romanowski, MS (The Charles T. Campbell Ophthalmic Microbiology Laboratory, UPMC Eye Center, Department of Ophthalmology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania) for critical reading of the manuscript.
Create a personal account or sign in to: