Infectious Disease Risk Factors of Corneal Graft Donors

Objective  To determine how donor health status affects the risk of infection after corneal transplant.

Methods  An adverse reaction surveillance registry was used to conduct a matched case-control study among transplanted donor corneas from January 1, 1994, to December 31, 2003. Cases comprised 162 reports of endophthalmitis after penetrating keratoplasty including 121 with microbial recovery, of which 59 had concordant donor and recipient microbial isolates. Two controls were matched to each case by surgery date. Conditional logistic regression estimated adjusted odds ratios with 95% confidence intervals according to the premortem status of decedent donors.

Results  Postkeratoplasty endophthalmitis was associated with recent hospitalization (odds ratio, 2.84; 95% confidence interval, 1.61-4.98) and fatal cancer (odds ratio, 2.46; 95% confidence interval, 1.53-3.97) among donors. Endophthalmitis appeared more likely with tissues transplanted longer than 5 days after donation (odds ratio, 1.55; 95% confidence interval, 1.02-2.35). The prevalence of concordant microbial isolates from donors and recipients was greater among fungal endophthalmitis than among bacterial endophthalmitis (P < .001).

Conclusions  Corneal grafts with eye tissue obtained from donors dying in the hospital or with cancer may have an increased risk of postsurgical endophthalmitis, possibly due to donor-to-host microbial transmission. Together with donor screening and processing, improvements in microbiological control may reduce infection associated with corneal transplant.

Infection is an uncommon but serious complication of corneal transplant.^1-3Most infected eyes lose vision or become blind.^1,2,4The preemption and deterrence of transmissible diseases are leading concerns in eye and tissue banking.^5,6

Judicious donor screening and good tissue practices aim to avert infections associated with transplant. Preventive strategies include deferral of donors with septicemia or endocarditis,^7-9antiseptic preparation and aseptic retrieval of donor tissue,^9-11and anti-infective preservation.^12,13Microbiological assessment may also be done before distribution¹⁴or at keratoplasty.¹⁵

Opportunities for further measures are excluding donors with prolonged hospitalization or malignant neoplasm^16,17and shortening storage of refrigerated tissue before surgery.¹⁸To examine how the donor's health status and the donation process influence postsurgical infection, we undertook a matched case-control study of postsurgical infections associated with corneal transplant that were reported to a national surveillance registry in the United States.

The Eye Bank Association of America systematically monitors adverse reactions of corneal transplant that are potentially attributable to donor ocular tissue.¹⁹Distributing eye banks seek postoperative outcomes from corneal surgeons and, if notified of a reportable infection, track the mated donor cornea, review donor corneoscleral rim cultures submitted at the surgeon's discretion during keratoplasty, and gather available microbiological results of intraocular fluids obtained from the corneal graft recipient. Adverse reaction information is forwarded to the source eye bank and submitted to the Eye Bank Association of America on a case report form that records the date of corneal transplant, donor age, cause of donor death, donation facility, interval between death and corneoscleral excision or enucleation, duration from recovery to transplant, and microbial species isolated from donor and recipient eyes. Data are entered into the adverse reaction registry, which is masked to personal identifiers of donors, surgeons, and recipients.

This matched case-control study nested in the surveillance registry was approved by the institutional review board at Baylor College of Medicine and endorsed by the Eye Bank Association of America Medical Advisory Board. Cases included all postsurgical endophthalmitis reported by source eye banks in the United States that followed corneal transplant performed between January 1, 1994, and December 31, 2003. The study period occurred during the use of Optisol-GS corneal storage medium (Bausch & Lomb Surgical, Irvine, California) and ended 2 years before data set construction to allow for complete reporting. While the sample size was constrained to 162 available cases, the power of this 1:2 matched case-control study was estimated to be 90% for detecting an odds ratio (OR) of 2.00 at a significance level of .05.²⁰

We aimed to have a control group that was representative of donor corneas distributed for penetrating keratoplasty in the United States during the surveillance decade. One eye bank each in the northeastern, southeastern, north central, south central, and western regions of the United States was randomly assigned case-surgery dates, identified controls by selecting the first 2 donor corneas distributed on or after these days, and electronically transferred data collated from retained records. Missing, illegible, illogical, and inconsistent responses for cases and controls were corrected by communication with the reporting eye bank. The interval between death and surgery was recorded for each control and was calculated for cases as the combined death-to-preservation and preservation-to-surgery intervals rounded to the next day. The total number of donor corneas distributed for transplant by US eye banks was obtained from the Eye Bank Association of America statistical report.²¹

Microorganisms recovered from corneal graft recipients diagnosed with endophthalmitis were compared with isolates from corresponding donor rims. The prevalence of concordant recipient and donor isolates was compared between bacterial and fungal endophthalmitis with a χ²test. Contingency tables and mean values of donor characteristics were initially examined among cases and controls using Fisher exact test for categorical variables and a 2-sided ttest for continuous data. Conditional logistic regression estimated ORs with 95% confidence intervals (CIs) for exposure variables using Intercooled Stata version 9 statistical software (Stata Corp, College Station, Texas). Categorization of continuous variables was guided by their distribution among controls and by linear trends of unadjusted ORs across quartiles compared with the referent category, with break points determined by the strength of association and clinical relevance. Multiplicative effect modification was tested in hierarchical models if Wald P < .15. Regression models were compared with the likelihood ratio test to achieve a parsimonious model that was examined with influence diagnostics.²²Subgroup analyses were performed after restricting cases to culture-positive corneal graft recipients, culture-positive recipients having the same microorganism recovered from the residual donor rim, culture-positive recipients with bacterial endophthalmitis, culture-positive recipients with fungal endophthalmitis, and recipients having endophthalmitis considered potentially attributable to donor corneal tissue. All of the statistical tests were 2-sided.

Eye banks distributed 340 174 donor corneas for surgery in the United States and 109 009 internationally during the 10-year study period.²¹The adverse reaction surveillance registry received 162 reports of endophthalmitis that followed penetrating keratoplasty with donor corneas distributed in Optisol-GS from 49 US eye banks in 29 states of the northeastern (21 reports), southeastern (49 reports), north central (41 reports), south central (24 reports), and western (27 reports) United States. The median duration between transplant and diagnosis of endophthalmitis was 7 days (25% and 75% quartiles, 3 and 32 days, respectively).

Among 146 recipient eyes that had intraocular fluid submitted for microbiological assessment, 18 yielded enterococci, 16 pneumococci, 26 other streptococci, 12 staphylococci or unspeciated gram-positive cocci, 3 clostridia, 7 gram-negative rods, 38 yeasts, and 1 filamentous fungus (Table 1). Of 121 cases of endophthalmitis that had both donor and recipient cultures done, 59 (48.8%) had concordant microbial isolates, including 28 of 81 (34.6%) with bacterial endophthalmitis and 31 of 40 (77.5%) with fungal endophthalmitis (P < .001). Endophthalmitis occurred after transplant of the contralateral cornea from the same donor among 24 cases, including 6 mated pairs that yielded similar microbial isolates from both recipient eyes (4 mated pairs with Candidaspecies, 1 pair with Clostridium perfringens, and 1 pair with Streptococcus agalactiae). Four cases with donor mate infection did not submit a separate form for the paired tissue recipient. No tissue mate among the controls was reported with endophthalmitis, although 3 mated transplants had primary graft failure. The contralateral cornea was not distributed for 12 cases and 15 controls.

Table 2compares donor and recipient characteristics of 162 cases of endophthalmitis and 324 matched controls. The donor corneas of 7 cases and 6 controls were imported from another US eye bank. The mean (SD) donor age was 53 (17) years for cases and 53 (17) years for controls. The mean (SD) recipient age was 69 (19) years for cases and 64 (21) years for controls. In a final multivariable model in which effect modification was not found (Table 3), the odds of endophthalmitis rose by 17% (95% CI, 6%-30%) for every 10 years of recipient age (P = .003).

The odds of hospital retrieval among reported cases of postkeratoplasty endophthalmitis were 2.84 (95% CI, 1.61-4.98) times that of other donation sites (P < .001). In comparison with other causes of death, cancer occurred significantly more often among cases (OR, 2.46; 95% CI, 1.53-3.97; P < .001). Malignant neoplasms among 53 cases and 54 controls with cancer-related deaths were breast (7 and 6, respectively), gastrointestinal (10 and 7, respectively), genitourinary (11 and 7, respectively), head and neck (6 and 7, respectively), pulmonary (12 and 19, respectively), skin or bone (2 and 3, respectively), and not reported (5 and 5, respectively). A donation-to-surgery duration exceeding 5 days occurred more often among cases than controls (OR, 1.55; 95% CI, 1.02-2.35; P = .04). Restricting the case group to 121 eyes with culture-confirmed endophthalmitis or to 59 culture-confirmed recipient infections having the same microorganism recovered from the donor corneal rim produced comparable results (Table 3). Similar ORs were also found for bacterial and fungal endophthalmitis subgroups and after excluding 56 endophthalmitis cases judged by medical directors to be probably not due to donor eye tissue, including 5 with concordant microbial isolation (data not shown).

Transplant agencies and surgeons take several precautions to safeguard against contamination and infection.²³Some strategies such as culturing donor eyes after cold storage have an uncertain preventive role.¹⁵We observed that recipient culture results correlated with donor rim isolates for a third of reported bacterial endophthalmitis and among three-fourths of fungal endophthalmitis. Better detection and control of contaminants, especially fungi, may improve the safety of eye banking.

A related opportunity for preventing donor-to-host transmission is to preclude potential donors who have an infectious disease risk.²⁴Decedents with untreated septicemia are proactively deferred, but the importance of other medical conditions is less clear.²⁵Although a previous study²⁶found that decedent illness was not associated with microbial recovery from residual donor tissue, a small series²⁷suggested a possible link between postkeratoplasty endophthalmitis and donor malignancy. Like sepsis, other donor conditions may affect the incidence of microbial colonization or contamination of the ocular surface and cornea.^7,9,16,17

This study examined the medical status of corneal donors before donation. We found that hospitalization or malignant neoplasm at the time of death more than doubled the odds of infection after keratoplasty. The causal pathway is unclear, but a chain of contamination is biologically tenable. Inpatients and patients with cancer are exposed to nosocomial pathogens such as enterococci and yeasts and may harbor microorganisms in the bloodstream and tissues.^9,17,28As this study implicates an infectious risk with corneas from donors dying in the hospital or with cancer, ensuring vigilance during medical record review to identify signs of active infection could potentially enhance the safety of corneal transplant.

Another prospect for prevention that is amenable to action and further research is to optimize tissue allocation activities. Refrigerated corneal preservation does not necessarily foster microbial persistence,^17,26,29but prolonging storage beyond 5 days has been associated with increased bacterial recovery.¹⁸Eye tissues obtained and distributed within 5 days of donation appear less likely to be associated with postsurgical infection in this study. Eye banks should continue endeavors to place tissue for transplant expeditiously.³⁰

The findings of this study are plausible, reasonably precise, consistent across subgroups, and likely to be representative of corneal transplants in the United States. We tried to ensure trustworthy results by considering possible challenges to validity in this case-control study. Sampling bias was minimized by choosing controls from geographically diverse eye banks and by matching on surgery dates of cases to incorporate regional and temporal trends. Selective registration of endophthalmitis cases, however, could have misrepresented the incidence and spectrum of infection after transplant. Despite requested notification from corneal surgeons and requisite reporting by eye banks, postkeratoplasty endophthalmitis was reported less frequently than in a systematic literature review.³

Another limitation may have occurred because information was accrued contemporaneously for cases but extracted retrospectively for controls. Donor records of infected cases could have been more closely scrutinized than successful transplants, but record review bias seems unlikely because the prevalence of cancer among controls equaled that among all corneal donors in the United States during the same time span²¹and was similar to the proportion in a prospective cohort.³¹We also doubt that computation bias affected results even though the death-to-surgery duration was recorded by control banks but was calculated for cases from cadaveric and storage intervals.

Eye banks and surgeons cope with potential sources of infectious agents other than the donor,^32-34and additional factors could have confounded the analysis. Adjunctive lens and vitreous procedures increase the risk of postkeratoplasty infection,³⁵and ocular surface flora can enter the eye during or after surgery.^36,37While our study was largely focused on donor characteristics and eye-banking practices, we corroborated that older patients have an increased risk of endophthalmitis after anterior segment surgery.^38,39

A national surveillance registry offers a unique resource to undertake etiologic investigations on the safety of eye banking and the outcomes of corneal transplant. Our results provide evidence that premortem conditions may affect donor eye tissue but do not warrant excluding broad categories of otherwise eligible donors.^40,41Blanket deferral from donation by hospitalized patients or those with cancer would be unreasonable as most corneas from these donors do not result in complications. Rather, efforts are needed to determine what illnesses, interventions, or other reasons might explain the pathway linking certain donors with recipient infection. We advocate judicious evaluation of decedents and encourage efficient recovery and delivery of donated tissues, but we also recognize an opportunity for better methods of microbiological assessment and control to reduce infections associated with corneal transplant.

Correspondence:Kirk R. Wilhelmus, MD, PhD, Sid W. Richardson Ocular Microbiology Laboratory, Department of Ophthalmology, Cullen Eye Institute, Baylor College of Medicine, 6565 Fannin St, NC-205, Houston, TX 77030 (kirkw@bcm.tmc.edu).

Submitted for Publication:December 19, 2006; final revision received February 2, 2007; accepted February 20, 2007.

Members of the Medical Review Subcommittee of the Medical Advisory Board for the Eye Bank Association of America:Patricia Dahl, CEBT; Chris Hanna, CEBT; Ben Jenkins, CEBT; Marian S. Macsai, MD (co-chair); Michael L. Nordlund, MD, PhD; Kevin Ross, MS, MPH; Bradley Tennant, CEBT (co-chair).

Financial Disclosure:None reported.

Funding/Support:This study was supported under a Ruth L. Kirschstein National Research Service Award (EY016631) from the National Eye Institute, National Institutes of Health, Bethesda, Maryland, and by unrestricted grants from the Eye Bank Association of America, Washington, DC, Research to Prevent Blindness, Inc, New York, New York, and the Sid W. Richardson Foundation, Fort Worth, Texas.

Additional Contributions:Tara Fischer, Donor Network of Arizona, and Trisha Watkins, CEBT, Georgia Eye Bank, Inc, collected data from control eye banks; Kim Harano, Baylor College of Medicine, maintained the surveillance database; and Rusty Kelly, CAE, Eye Bank Association of America, administered the electronic report system.