Flowchart explaining subsets of eyes analyzed.
Jampel HD, Moon JI, Quigley HA, Barron Y, Lam K. Aqueous Humor Uric Acid and Ascorbic Acid Concentrations and Outcome of Trabeculectomy. Arch Ophthalmol. 1998;116(3):281-285. doi:10.1001/archopht.116.3.281
To determine if there is an association between the surgical outcome of trabeculectomy and uric acid and ascorbic acid concentrations in the aqueous humor at the time of the procedure.
Patients, Materials, and Methods
Aqueous humor samples were collected from the eyes of 169 of 249 adult patients who underwent trabeculectomy alone for any type of glaucoma between April 1989 and July 1995. Postoperatively, all medical records were reviewed and outcomes were classified as successful, unsuccessful, or indeterminate. The ascorbic acid and uric acid concentrations were determined in masked fashion by high-pressure liquid chromatography. Factors associated with surgical outcome were determined.
Uric acid concentration was higher in unsuccessful eyes (mean±SD, 0.21±0.08 mmol/L, n=26) than in successful eyes (0.15±0.09 mmol/L, n=91, 95% confidence interval for difference, 0.02-0.10 mmol/L). Ascorbic acid levels were not significantly different in the eyes with unsuccessful (1129.9±601.9 µmol/L) and successful (1334.3±511.0 µmol/L) surgery (95% confidence interval for difference, −475.2 to 66.4 µmol/L, P=.13) surgery. Other factors associated with failure were previous surgery and surgery performed at the inferior limbus. A multiple polytomous logistic regression analysis was performed, after excluding the small number of operations performed at the inferior limbus. The odds ratio for failure increased by a factor of 1.68 for every 1-mmol/L increase in uric acid (95% confidence interval, 1.16-2.43, P=.006).
Uric acid levels were higher at the time of surgery in eyes that had unsuccessful outcomes than in those with successful outcomes. No significant difference in ascorbic acid levels was detectable. A higher uric acid level in the aqueous humor is a risk factor for trabeculectomy failure and might be tested as a prognostic indicator. Understanding the host characteristics that determine whether a trabeculectomy will be successful in a given eye is important. Data from the Fluorouracil Filtering Surgery Study Group suggest that in eyes with previous surgery the following are risk factors for surgical failure: time elapsed since the last procedure involving a conjunctival incision, the number of previous procedures involving conjunctival incisions, high preoperative intraocular pressure, and Hispanic ethnicity. In addition, previous intraocular surgery itself, age, race, and the diagnosis of uveitic or neovascular glaucoma are generally considered risk factors for trabeculectomy failure.
CERTAIN factors have been associated with poorer prognosis in trabeculectomy, namely, time elapsed since the last procedure involving a conjunctival incision, the number of previous procedures involving conjunctival incisions, high preoperative intraocular pressure (IOP), and Hispanic ethnicity.1 In addition, previous intraocular surgery itself,2- 4 age,5 race,6,7 and the diagnosis of uveitic or neovascular glaucoma8 are also generally considered risk factors for trabeculectomy failure. Why these factors are associated with a poorer surgical prognosis is unknown. Eyes with unsuccessful surgical outcomes exhibit a more vigorous wound healing response than their successful counterparts.9 There are 2 possible explanations for this more vigorous wound healing response. First, the conjunctiva may be more prone to scarring, a hypothesis supported by the observation that increased cellularity of the conjunctiva is associated with a decreased success rate.10 Second, the composition of the aqueous humor that bathes the surgical site in these unsuccessful trabulectomies may stimulate wound healing.11
We have previously shown that ascorbic acid, at concentrations present in human aqueous humor, is cytotoxic to human Tenon capsule fibroblasts.12 This observation led us to hypothesize that the ascorbic acid concentration in the aqueous humor might affect the outcome of surgery. Because there might be an important interaction between ascorbic acid and uric acid, we assayed uric acid as well.
All procedures were approved by the Joint Committee on Clinical Research of The Johns Hopkins University School of Medicine and The Johns Hopkins Hospital, Baltimore, Md.
From April 1989 to July 1995, all patients older than 18 years who underwent trabeculectomy by one surgeon (H.D.J.) at the Wilmer Institute, Baltimore, were asked if an aqueous humor specimen could be collected during their surgery and frozen for further analysis. Three hundred six trabeculectomies were performed during this period (Figure 1), and aqueous humor specimens were obtained in 169 of the procedures.
Aqueous humor specimens were obtained by a technique similar to that in previous studies.13 In preparation for surgery all patients received intravenous sedation, and local anesthesia was given in retrobulbar or peribulbar locations that included a 50:50 mixture of 2% lidocaine hydrochloride and 0.75% bupivacaine hydrochloride. In all eyes, aqueous humor was removed before the eye had been entered with any instrument. Approximately 100 µL of aqueous humor was removed using a 27-gauge needle on a tuberculin syringe over the course of 2 to 3 seconds. Care was taken not to contact the iris or the lens. The aqueous humor specimens were immediately placed in 600-µL microcentrifuge tubes (United Scientific, San Leandro, Calif) and kept on dry ice until storage at −85°C.
The clinical records of 249 patients were studied (Figure 1). Criteria for success and failure of surgery were established before clinical records were reviewed. The surgical outcomes were characterized as (1) successful: IOP equal to or less than 17 mm Hg and more than 25% lowering of IOP without IOP-lowering medications, and follow-up of 5 months or more; (2) indeterminate: IOP 17 mm Hg or lower with medications or 18 mm Hg or higher without medications, and follow-up of 5 months or more; (3) unsuccessful, IOP 21 mm Hg or lower, or more than or equal to preoperative levels, with at least 1 medication, and a follow-up of 5 months or more; or (4) incomplete: not enough information available from the medical record.
Aqueous humor from all 169 eyes with specimens was sent on dry ice to a laboratory for analysis of ascorbic and uric acid. Four batches of specimens were analyzed at different times. The first 3 batches (from 1992, 1993, and 1995) included only specimens from eyes in the successful or unsuccessful categories; the last batch (from 1996) included specimens from eyes with all 4 outcomes.
High-pressure liquid chromatography was performed using an o-Bondapax–NH2 (Waters Associates, Medford, Mass) column and a 5-mmol NH4H2PO4/0.05-mol EDTA buffer as previously described.14 The eluate from the column was monitored at 254 nm for ascorbic acid and 280 nm for uric acid. The operator of the high-pressure liquid chromatography was masked to the clinical outcome.
Differences in mean values for continuous variables, such as uric acid concentration, were analyzed using an unpaired t test, and the 95% confidence interval (CI) for the difference between 2 means was calculated. Comparisons of more than 2 means were made by analysis of variance (ANOVA). Proportions were analyzed using a χ2 test, and the odds ratio and 95% CI for the odds ratio were calculated.
Univariate polytomous logistic regression analyses were performed to determine which variables were statistically significantly associated with risk of failure. The statistically significant variables were entered into a multiple polytomous logistic regression model.
The patients from whose eyes specimens were obtained and those from whose eyes specimens were not obtained were similar for age, sex, race, preoperative diagnosis, preoperative medications, use of an antifibrosis agent, and the location of their surgery (Table 1). The eyes from those patients from whom specimens were not obtained had a higher likelihood of having had previous surgery, and a slightly higher preoperative IOP than those from whom specimens were obtained.
In those eyes with specimens, the concentrations of ascorbic acid and uric acid were analyzed by outcome and by year of analysis (Table 2). For the analysis of uric acid, there was no difference in the mean uric acid concentrations among the assay years (P=.12, ANOVA). Therefore, further analysis of uric acid concentration was done on pooled data from all 4 batches. However, for the ascorbic acid assay, the mean ascorbic acid concentration was much lower in the 1996 batch (P<.001, ANOVA) than in the preceding analyses, which did not differ from one another (P=.11, ANOVA). The lower values in the 1996 batch were not correlated with an increased duration of freezing (data not shown). Because of the discrepancy between the 1996 results and the earlier batches, only the ascorbic acid data from the 1992, 1993, and 1995 batches were used in the analysis.
The eyes in the unsuccessful group were more likely to have had a trabeculectomy performed at the inferior limbus and to have undergone previous conjunctival surgery than the eyes in the other groups (Table 3). There were no differences among the successful, unsuccessful, and indeterminate groups in age, preoperative IOP, sex, race, preoperative diagnosis, use of preoperative medications, or use of antifibrosis agents.
The uric acid concentration in the eyes in the unsuccessful group was higher than in the other 2 groups (P=.006, ANOVA), which did not differ from one another (P=.38, unpaired t test). The 95% CIs for the difference between the means were 0.35 to 1.6 for the unsuccessful vs successful eyes (P=.005), and 0.42 to 2.33 for the unsuccessful vs indeterminate eyes (P=.004). The uric acid concentration was lower in the indeterminate group than in the successful group, but the difference was not statistically significant (95% CI for the difference, −1.13 to 0.33, P=.38). All the above analyses were repeated with the incomplete eyes included, with similar results (data not shown). The ascorbic acid concentration in the unsuccessful group was lower than in the successful group, but the difference was not statistically significant (95% CI for the difference, −1.48 to 8.68, P=.77). The power of our study to detect a difference of this size, however, was only 0.30.
We performed univariate polytomous logistic regression analysis to determine if uric acid concentration was independently associated with outcome, and whether there were other variables that were significantly associated with outcome. Previous surgery (P=.02), location of surgery (P<.001), and uric acid concentration (P=.01) were significantly associated with outcome in this model. We next performed multiple polymotous logistic regression analysis, after including variables of age, race, sex, previous surgery, location of surgery, and uric acid level. In this model, there was only a trend for surgical failure with increasing uric acid concentration (the odds of failure increased by a factor of 1.33 for every 1-mmol/L increase in uric acid concentration [95% CI, 0.94-1.88, P=.1]).
We studied the data in an attempt to explain the discrepancy between the univariate analysis, in which uric acid concentration was strongly associated with surgical outcome, and the multivariate analysis, in which the association was weaker. This discrepancy seemed related to the inclusion of the 14 eyes that underwent trabeculectomy at the inferior limbus. This small group of eyes had a large variance in uric acid concentrations, and had an absolute association with previous surgery (all eyes that underwent surgery at the inferior limbus had undergone previous surgery). When those eyes that had undergone surgery at the inferior limbus were excluded from the multiple polytomous logistic regression analysis, so that only eyes undergoing trabeculectomy at the superior limbus were analyzed, uric acid concentration was more strongly associated with outcome (P=.01). The odds of being in the unsuccessful group compared with the successful group increased by a factor of 1.68 for every 1-mmol/L increase in uric acid concentration (95% CI, 1.16-2.43, P=.006). The odds of being in the indeterminate group compared to the successful group decreased by a factor of 0.73 for every 1-mmol/L increase in uric acid, although the difference was not statistically significant (95% CI, 0.4-1.33, P=.3).
We have previously postulated12 that the high concentration of ascorbic acid in the aqueous humor might explain the incomplete wound healing response that characterizes successful trabeculectomy surgery. To lend support to this hypothesis, we sought to discover whether aqueous humor ascorbic acid levels were associated with trabeculectomy success. Although the mean ascorbic acid concentration was higher in the successful group than in the unsuccessful group, the difference was not statistically significant. Unfortunately, the data from the 1996 batch of aqueous humor could not be used because of the markedly lower ascorbic acid values in that batch. The lower ascorbic acid values may have been due to oxidation of the ascorbic acid during transport for analysis. Perhaps if that assay had yielded credible data it might have confirmed an association.
We did find, however, that eyes that have had unsuccessful trabeculectomy surgery have higher aqueous humor uric acid concentrations than eyes that have had successful surgery. This association was seen in all 4 batches analyzed and was statistically significant in the pooled data. Thus we have shown for the first time, to our knowledge, that a feature of the aqueous humor is associated with the outcome of glaucoma surgery.
The observation that the indeterminate group had the lowest mean uric acid concentration complicates the interpretation of our finding concerning the association of uric acid with outcome. If there were an important relationship between uric acid and surgical outcome, one would expect the uric acid concentration in the indeterminate group to be intermediate between the successful and unsuccessful groups. The number of eyes in the indeterminate group was small, however, and the difference in uric acid concentration between it and the successful group was not statistically significant. Furthermore, the eyes in the indeterminate group may not have differed greatly in outcome from those in the successful group.
In addition to uric acid, a history of previous conjunctival surgery and the location of the operation at the inferior limbus were also associated with a poor surgical outcome, confirming data from other studies.2- 4 The inclusion of operations performed at the inferior limbus complicated the statistical modeling, both because of the small number of eyes and the fact that these operations were only performed on eyes that had had previous surgery. In fact, the association between previous surgery and outcome disappeared when only surgery performed at the superior limbus was considered. This lack of association could be explained by the greater use of antifibrosis agents in eyes with conjunctival scarring. Similar to the findings of the Fluorouracil Filtering Surgery Study Group,1 we were also unable to find an association between African American race and surgical outcome.
Why was the uric acid level higher in eyes that had unsuccessful surgery? It is unlikely that the excess uric acid was derived from the serum. Experiments in rabbits in which hyperuricemia was induced failed to raise the levels of uric acid in the aqueous humor.15 Furthermore, the levels of uric acid reported in serum16 are comparable to those present in the aqueous humor, so breakdown of the blood-aqueous barrier is unlikely to be an explanation. We know of no studies correlating aqueous humor and plasma uric acid levels, and unfortunately, we did not obtain serum specimens at the time of surgery. A more likely explanation for the higher uric acid concentration is local production. Uric acid is a product of the catabolism of nucleotides; and xanthine oxidase, which catalyzes the conversion of xanthine to uric acid, is present within the eye.17 Uric acid concentration can be expected to increase in situations in which there is tissue breakdown and death. We speculate that the eyes that had failed trabeculectomies were those with high rates of tissue breakdown and cell death, indicated by higher aqueous humor uric acid levels, although there are no data to support this hypothesis.
We would like to perform aqueous humor studies during the postoperative period, since it is the postoperative aqueous humor that bathes the surgical wound. Unfortunately this is incompatible with safe patient care, is impractical in monkeys, and may not be relevant in other laboratory animals such as the rabbit. The measurement of uric acid from aqueous humor obtained at the time of surgery could nevertheless be used as a prognostic marker, although it would be unlikely to affect clinical care. However, if uric acid levels in the aqueous humor could be measured preoperatively and noninvasively, the results could be used to determine whether or not, and how aggressively, to use an antifibrosis agent.
Aqueous humor composition is clearly not the sole determinant of the outcome of trabeculectomy surgery. Nevertheless, we hope this work will stimulate others to consider the composition of the aqueous humor as one determinant of the outcome of trabeculectomy.
Accepted for publication December 1, 1997.
This research was supported in part by grant EY 01765 (Core Facility Grant, Wilmer Institute) from the National Eye Institute, National Institutes of Health, Bethesda, Md.
Corresponding author: Henry D. Jampel, MD, Maumenee B-110, 600 N Wolfe St, Baltimore, MD 21287-9205.