Flow of patients throughout the trial. Asterisks indicate corneas that were not clear enough to allow specular microscopy; dagger, corneas that were a secondary graft failure; and double dagger, cornea pairs with one unclear mate were excluded.
The mean values at each time point are presented. The error bars represent SDs.
Naor J, Slomovic AR, Chipman M, Rootman DS. A Randomized, Double-Masked Clinical Trial of Optisol-GS vs Chen Medium for Human Corneal Storage. Arch Ophthalmol. 2002;120(10):1280–1285. doi:10.1001/archopht.120.10.1280
To determine whether the clinical outcome of corneal transplantation performed with tissue stored in Chen Medium (CM; Chen Laboratories, Phoenix, Md) is superior to that of Optisol-GS (Chiron Ophthalmics, Irvine, Calif).
Randomized, double-masked trial.
Participants and Setting
Ninety patients undergoing corneal transplantation at a tertiary center.
Donor cornea pairs were stored, one in CM and the other in Optisol-GS, transplanted, and followed up for 3 months.
Main Outcome Measures
Endothelial cell loss (ECL) at 3 months. Secondary outcomes were thickness and epithelialization at days 1, 7, 30, and 90, as well as primary and secondary graft failure.
Mean percentages of ECL 3 months after surgery were similar for corneas stored in Optisol-GS and corneas stored in CM. The 95% confidence interval(CI) for the difference was −3.4% to 13.2%. Covariate-adjusted 95% CI was −6.1% to 9.1%. Pachymetry values were similar in the 2 treatment groups. Eighty-eight percent of the corneas stored in each medium achieved epithelialization of 75% to 100% of the surface area within 1 week after transplantation. No primary donor failures occurred in either group. One graft that was preserved in CM failed during the study period.
The clinical outcomes of corneal transplantation with tissue that was preserved in CM were similar to those of grafts preserved in Optisol-GS.
INTERMEDIATE-TERM CORNEAL storage allows storage of the tissue at 4°C for as long as 2 weeks. It offers flexibility in scheduling the keratoplasty, a longer period to observe and test the tissue, and it involves no complex technical procedures. Consequently, it is the most widespread preservation method in the world.
Chen Medium (CM; Chen Laboratories, Phoenix, Md) was recently introduced as an intermediate-term corneal storage medium and received permission for marketing by the US Food and Drug Administration. It is isotonic (approximately 305 mOsm) and contains the metabolic substrate β-hydroxybutyrate, streptomycin sulfate, gentamicin sulfate, and N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid buffers to stabilize the pH at approximately 7.4 on phenol red indicator, and approximately 7% dextran in Tc199 medium modified by reducing sodium chloride to 85mM and omitting sodium bicarbonate.1 Chen medium enables the cornea to maintain a high level of metabolic activity and to generate adenosine triphosphate with concurrent suppression of acid formation and accumulation and uses a built-in bicarbonate-generating system for pH stability and for prevention of exogenous sodium bicarbonate–elicited adverse effects.2
Whenever a new therapeutic modality becomes available, it is important to compare its safety and efficacy to the existing alternatives. The end user can then make an informed choice based primarily on the medical considerations. To that end, to determine whether there is an advantage to using CM, we conducted a randomized, double-masked clinical trial comparing it with Optisol-GS (Chiron Ophthalmics, Irvine, Calif), the most commonly used medium.3,4
The study was conducted at a tertiary eye-care service; namely, the Cornea and External Disease Service at the Department of Ophthalmology, Toronto Western Hospital, University Heath Network, Toronto, Ontario. Two surgeons participated in the study (A.R.S. and D.S.R.). The protocol was approved by the Human Subjects Review Committee of the University of Toronto and the University Health Network Research Ethics Board. Informed consent was obtained from all study participants. Patients were enrolled from September 1999 through June 2000, when the target sample size was attained.
The study population consisted of patients undergoing corneal transplantation either alone or in conjunction with the following procedures: cataract extraction with or without intraocular lens insertion, intraocular lens exchange, and intraocular lens insertion. We excluded patients with the following conditions, which carry a poor prognosis for graft clarity, and would consequently make the main outcome assessment difficult: severe chemical burns, radiation burns, ocular pemphigoid, Stevens-Johnson syndrome, and neuroparalysis. Patients who could not make themselves available for preadmission tests and the informed consent process 1 week prior to surgery, or who would be unavailable for follow-up were also excluded. Patients were approached consecutively.
Donor corneas for the study were processed by the Eye Bank of Canada(Ontario Division, Toronto) in the usual manner and according to the standards are set by the Eye Bank Association of America.5 After obtaining consent from the donor's family, corneal tissue was retrieved by whole-globe enucleation. Donor blood was collected and tested for human immunodeficiency virus, hepatitis B, and hepatitis C. Whole globes arrived at the eye bank laboratory and underwent a screening slitlamp examination. Corneoscleral rims were excised, and based on the results of the randomization, they were transferred to either CM or Optisol-GS. Specular microscopy of the central endothelium was performed for each cornea, and endothelial cell counts were recorded (Video Enhancement Digitization System, Bioptics, Arlington, Mass). Endothelial cell counts were the mean of 3 measurements of at least 50 endothelial cells per analysis. Corneas with endothelial cell densities less than 2000 cells/mm2 were considered not suitable for use. Tissue was released to the surgeon at the day of surgery after a second screening slitlamp examination.
During surgery, the Hanna trephine system and viscoelastics were used in combination with either interrupted and running 10-0 nylon, or a double-running suture (10-0 and 11-0 nylon). Anterior vitrectomy was performed in eyes with displacement of vitreous into the anterior chamber. Subconjunctival gentamicin sulfate, cefazolin sodium, and methylprednisolone were administered at the end of each procedure. Patients received a tobramycin-dexamethasone eye drop 4 times daily for the first few weeks followed by topical corticosteroids as needed thereafter.
Randomization was performed at the eye bank and recorded in a log. The instrument for randomization was a table of random digits. Single corneas were the unit of randomization. Each cornea was randomly assigned to one of the 2 storage media, one of the 2 surgeons who participated in the study, and one of the patients scheduled to be operated on by the assigned surgeon. A pair of corneas from a single donor was preserved with one in CM and the other in Optisol-GS, hence matching for donor characteristics.
Recipients were masked to the type of storage media used for the tissue that they had received. The labels on the vials containing the corneas, which were distributed to the surgeons, did not specify the type of solution used. During tissue manipulation, the color of the storage solution was visible to the surgeons. Optisol-GS is a lighter hue of red than CM; however, to distinguish between the two, one would need to place them one beside the other. It was expected that the lack of a frame of reference would impose masking. A masked observer analyzed the endothelial images.
Patients underwent clinical assessments on days 1, 7, 30, and 90. In addition to a slitlamp examination, an ultrasonic pachymetry of the central cornea (Ophthasonic A-scan/Pachymeter; Teknar Inc, St Louis, Mo) and an assessment of the extent of epithelial defect with fluorescein staining (0%-25%, 26%-50%, 51%-75%, and 76%-100%) were performed. During the last visit (at day 90) specular microscopy of the central cornea was performed. Endothelial cell counts were carried out in a manner similar to the one described for the donor tissue at the Eye Bank.
The sample size was determined by evaluating data from a pilot study at the same settings and conducted by the same investigators. Thirteen pairs of patients received corneal buttons preserved in either CM or Optisol-GS. Two months after transplantation, the pooled mean endothelial cell loss was 16.4% (SD, 11.6%). For the present study it was decided that a clinically relevant reduction in mean endothelial cell loss would be 10% or more (eg, 15% to 5% loss). A sample size of 34 pairs of donor eyes (68 recipients) was calculated for a type I error P value of .05, a type II error P value of .20, a 2-sided alternative hypothesis, and a paired test. Ninety patients were enrolled to allow for loss to follow-up.
The primary efficacy outcome was endothelial cell loss (ECL) 3 months following surgery. It was defined as the percentage difference between the baseline cellular density (measured at the eye bank in vitro) and the cellular density 3 months following surgery (measured at the ophthalmology clinic in vivo). Secondary efficacy outcomes were corneal thickness, measured at 1, 7, 30, and 90 days following surgery; the extent of epithelialization of the graft (measured at similar time points); as well as primary and secondary graft failures. Primary graft failure was defined as an irreversible opacification of the graft occurring within 2 weeks after transplantation.
Demographics and baseline characteristics were summarized and tested for treatment group comparability using the t test or the χ2 test. The treatment effect on the primary outcome(ECL) was calculated using a paired t test. Subsequently, to increase the precision of the estimate of the treatment effect on ECL, multiple linear regression methods were used. A set of explanatory variables(recipient characteristics and operative and postoperative factors) was evaluated and included in an initial model if they had a significant association with ECL or a larger than 10% influence on the estimate of the medium effect compared with the paired t test. Variables were then eliminated in a backwards fashion based on the probability of their association with the outcome and the effect of their elimination on the standard error of the estimate of the storage medium effect. The final model yielded the smallest SE for the estimated medium's effect. For average corneal thickness, it was decided a priori that a statistical comparison be made only for the interval where, based on clinical judgment, a significant disparity exists. Comparisons were completed using the repeated-measures analysis of variance. We used the SAS 7.0 (SAS Institute, Cary, NC) GLM (generalized linear models) and MIXED procedures. A P value of .05 or less was considered statistically significant. Other secondary outcomes, namely, the distribution of epithelial defects by size at different time points, as well as primary and secondary graft failure rates, were described without employing formal statistical tests.
Donor information was kept at the eye bank as part of its regular record system, in the form of a Microsoft Access (Microsoft Inc, Redmond, Wash) electronic database. Recipient characteristics and the results of the follow-up examinations were collected in a case report form and were later entered into a Microsoft Access electronic database that was created for this purpose. For the purpose of analysis, the donor and recipient databases were merged and imported to SAS. No prospectively defined stopping rules were used.
A total 45 pairs of donor corneas were assigned individually and randomly to CM (45 corneas) or Optisol-GS (45 corneas) and subsequently distributed to 90 eyes of 90 patients. The baseline characteristics of the pairs of donor corneas are presented in Table 1. Corneas from a single donor were not necessarily transplanted on the same day; nevertheless, the difference in storage time between the 2 treatment groups was not statistically significant. The baseline characteristics of the 90 recipients were statistically balanced (Table 2).
Eighty-seven patients (96.7%) completed the 90-day examination (44 [97.8%] in CM and 43 [95.6%] in Optisol-GS) (Figure 1). The corneas of 7 patients failed to achieve sufficient clarity to allow specular microscopy at 90 days (3 [6.7%] in CM and 4 [8.9%] in Optisol-GS). Endothelial cell counts were available for 35 pairs of corneas (77.8%) at 90 days.
The average percentage of ECL (observed 90 days after transplantation) for the 2 treatment groups combined was 21.5% (SD, 19.3%) (Table 3). This corresponded to a reduction in cell density of 567 cells per mm2 (SD, 505 cells per mm2). The average percentage of ECL in the Optisol-GS group was 4.9% more than in the CM group (24.0% [SD, 20.5%] and 19.1% [SD, 18.0%], respectively). This difference was not statistically significant (P = .25; 95% confidence interval [CI],−3.4% to 13.2%).
Constructing a multivariable model increased the precision of the estimate of the treatment effect on ECL. The model that yielded the smallest SE for the estimated medium's effect on ECL included variables to adjust for donor characteristics, the presence of glaucoma prior to surgery, and the occurrence of a rejection episode. We achieved with this model an estimate of 1.52 with an SE of 3.87 (t = 0.39, P =.70; F1,37 = 2.04, P = .02; the F test applied to the entire model). Hence, the adjusted average percentage of ECL difference between the Optisol-GS group and the CM group was reduced to 1.5%. When compared with the paired t test, the calculated reduction in the variance of the estimate was 16%, and its 95% CI had tightened to between −6.1% and 9.1%.
Figure 2 graphically displays corneal thickness values, by medium, 1, 7, 30, and 90 days after the operation. Corneal thickness is at its highest immediately following the operation and decreases over time to reach a plateau. The corresponding numerical data is presented in Table 4. At days 1, 7, and 30, Optisol-GS–stored corneas were, on average, thinner than corneas stored in CM (30 µm, 5 µm, and 1 µm, respectively), while at day 90, the CM-stored corneas were thinner, on average, by 14 µm than the corneas stored in Optisol-GS. For both storage media, corneal thickness values at days 30 and 90 were in the vicinity of the 512 µm (SD, 35µm) reported for normal corneas.6 Consequently, the repeated-measures analyses of variance were performed only for the first 30 postoperative days. The results for the model, which include variables for time and treatment, are presented in Table 5. The estimate for the time effect was significant, showing a 45-µm decrease in thickness (SE, 0.5 µm) for every unit of time(in days). The estimate of the treatment effect was not significant; thus, there was no difference in thickness between the corneas that were stored in CM and in Optisol-GS during the first 30 postoperative days. Other models were fitted with time in several categories (to relax the assumption of a linear association between thickness and time) and with interaction terms for treatment with time (to allow for different treatment effects at different points in time). The estimate for the effect of the storage medium remained nonsignificant and there was no evidence of interaction.
Table 6 presents the distribution of epithelial defects by size that was observed at the first and seventh postoperative days. On the first postoperative day, 27 patients (77%) and 28 patients (80%) in the CM and Optisol-GS groups, respectively, had epithelial defects extending throughout more than 25% of their graft surface. On the seventh postoperative day, the numbers decreased to 4 patients (11%) in each group. At postoperative days 30 and 90, none of the patients had an epithelial defect larger than 25% of the graft area surface.
There were no primary graft failure events. Secondary failure of one graft that was preserved in CM was attributed to a touch of the intraocular lens with the posterior corneal surface.
The primary outcome of this study was ECL following transplantation. At 90 days after surgery, the cornea had usually cleared, and specular microscopy was possible. During most of this time, patients received topical antibiotics and steroids to prevent infection and an inflammatory reaction. Ninety days is close enough to surgery to assume that the endothelial cell counts predominantly reflect the effects of storage as opposed to recipient-related factors.
On average, Optisol-GS–stored corneas lost 4.9% more endothelial cells than CM-stored corneas 90 days after transplantation. This difference was not statistically significant (P = .25; 95% CI,−3.4-13.2). It includes the 10% judged clinically relevant, which was factored into the sample size calculation. A multivariable model that adjusted for covariates enhanced precision and the 95% CI had tightened to between −6.1% and 9.1%, which does not include 10%. This increases the confidence that the 2 media are similar with respect to the primary end point of the study. It concurs with the results of a study comparing the first 15 patients who received corneas preserved in CM with 17 patients who received corneas preserved in Optisol-GS.7 The latter group's average ECL 2 months after transplantation was a nonsignificant 3% higher than the ECL of the former (95% CI was not reported).
The mean (SD) percentage of ECL that we observed 90 days after surgery for Optisol-GS–stored corneas (24.0% [20.5%]) was higher than values reported in the past by Lass et al,8 Lindstrom et al,4 and Bourne et al7 (3.6%[15.7%], 5.0% [18.4%], and 11% [9%], respectively). However, the largest diagnostic group in the first 2 studies was pseudophakic bullous keratopathy (PBK). It constituted 42% in the first study and 53% in the second. In the third study, Fuchs dystrophy constituted 52% of the reported diagnoses. In contrast, the largest diagnostic group in the present study was regrafted, accounting for 22.2%. In a univariable analysis, preoperative diagnosis was significantly associated with ECL (P<.01); specifically, regrafting was associated with 8.8% and 13.3% higher rates of ECL 90 days after transplantation as compared with PBK and Fuchs dystrophy, respectively. Hence, the discrepancy in ECL rates during the early postoperative period between the above studies and the present one could be related to differences in determinants of ECL between the 2 study populations; namely, the preoperative diagnosis leading to the corneal transplantation. Another contributing factor to the observed variability in ECL rates after corneal transplantation could be different eye bank practices; specifically, in situ corneal excision vs whole globe removal (in the present study, the latter method was used exclusively).
Corneal thickness, a secondary outcome in this study, is known to be associated with the function and number of corneal endothelial cells. Therefore, it serves as a surrogate measure for the health and number of endothelial cells during the early postoperative period when endothelial cell density cannot be measured. Dextran, a constituent of both Optisol-GS and CM, an osmotically active agent, enters the cornea during storage.9,10 This results in swelling of the cornea once it is exposed to the aqueous humor(an isotonic solution), with subsequent thinning as the endothelium resumes its function. Indeed, for both storage media, corneal thickness was highest immediately following transplantation, and it gradually decreased throughout the follow-up.
Previous studies8,11- 14 compared postoperative corneal thickness independently at different time points during the postoperative period, thereby ignoring the fact that the values at each time point are from the same individuals.15 In the present study, a more exhaustive method was used (repeated-measures analysis of variance). Additionally, models were fitted allowing for a nonlinear rate of thinning (treating time as a categorical variable) and for different treatment effects at different time points (treatment × time interaction). Corneal thickness throughout the first 30 postoperative days was similar for the 2 treatment groups. The only statistically significant finding was confirmation of the observed corneal thinning during the postoperative period in the 2 treatment groups. Specifically, the pooled (for all the corneas in the study) average rate of thinning during the first 30 postoperative days was 4.5 µm(SE, 0.5 µm) per day (P<.01).
Reportedly, corneas stored in CM are thicker during storage than those stored in Optisol-GS.1 This study does not suggest that this affects postoperative corneal thickness as measured from day 1 onward. Admittedly, corneal thickness was not measured during the operation. Intraoperative corneal thickness may influence tissue manipulation, but whether it affects ECL or graft survival remains to be determined. The time points at which thickness measurements were made were imposed by the follow-up schedule of patients who have undergone penetrating keratoplasty. More frequent measurements may increase the likelihood of detecting a difference. Yet, the clinical significance of such a small true difference is questionable, as it is unlikely to be perceived by the patient.
Postoperative pachymetry values observed in our study for Optisol-GS compare favorably with those found in the literature.4,7,8 The values reported by Chen et al13 for CM are somewhat higher; however, the sample contained only 5 pairs of patients.
The donor's epithelium is gradually replaced by that of the recipient, and corneas with damaged epithelium are still considered suitable for use. The degree of corneal epithelialization the first day after surgery is to a great extent the result of insults that occurred during tissue manipulation from the time of death of the donor, through processing at the eye bank, and finally, during surgery. These factors are hard to measure. In the present study, at the first postoperative day, 23% and 20% patients in the CM and Optisol-GS groups, respectively, had completed epithelialization of 75% to 100% of their cornea. By the seventh day, the number increased to 88% of the patients in each group. By day 30, all the patients had reached this stage. As argued for corneal thickness, more frequent measurements may increase the likelihood of detecting a difference. Another issue is the categorical scale that was used to measure the deepithelialized area. Photography and image analysis could further increase precision in future studies. However, it is debatable as to whether the clinical significance of such small differences at the early postoperative period justifies the effort.
From a patient's perspective, graft failure may be the most relevant outcome for a clinical study that compares 2 corneal storage media. Yet, the rarity of these events makes their use as outcome measures impractical since it would require large sample sizes and long follow-up periods. The incidence of secondary graft failure during the first few months after surgery is further reduced by the application of topical steroids to prevent rejection.
Generalization of the study results to the population of patients who require corneal transplantation seems adequate. Corneal transplantation is usually performed in tertiary centers similar to the one where the present study took place. Eligible for the study were patients undergoing all types of corneal transplantations, whether alone or supplemented by another surgical procedure. Although exclusion criteria included a few ocular conditions that carried a very poor prognosis for graft clarity, in reality, no such patients were encountered during the 9-month study period, testifying to the rarity of these conditions. As to internal validity, the balance of baseline characteristics between the treatment groups confirms effective randomization with respect to known risk factors, and suggests balance with respect to unknown risk factors. Nearly all patients (96.7%) completed the 90-day examination, and primary outcome measurements were possible for 78%. The only statistically significant difference in baseline characteristics between the patients who completed the follow-up and those who did not, was a mean storage time 12 hours longer for the latter group. Average (SD) storage time for all the patients in the study was 99 (22) hours. Bourne et al7 demonstrated a significant correlation between ECL at 2 months and storage time; however, in the present study, no such association was detected.
In summary, this study found CM and Optisol-GS to be similar with respect to corneal endothelial survival 3 months after transplantation, as well as corneal thickness and reepithelialization. Thus, there seemed to be no safety or efficacy differences that are clinically relevant between the 2 storage media. It is reasonable that eye banks, in collaboration with the corneal surgeons, determine which storage medium to use based on considerations like availability, cost, and ease of work.
Submitted for publication January 4, 2002; final revision received May 15, 2002; accepted May 29, 2002.
The authors would like to thank The Toronto Eye Foundation (Toronto) and the Ontario Division of the Eye Bank of Canada (Toronto).
Corresponding author: Joel Naor, MD, MSc, 3841 W 23rd Ave, Vancouver, British Columbia, Canada V6S 1K8 (e-mail: firstname.lastname@example.org).