Goniosurgery for Glaucoma Complicating Chronic Childhood Uveitis

Objectives  To describe the safety and efficacy of goniotomy in medically uncontrolledglaucoma complicating chronic uveitis and the factors affecting its outcome.

Methods  All goniotomies performed by a single surgeon for refractory childhooduveitic glaucoma were retrospectively reviewed. Success was defined as finalintraocular pressure (IOP) of no greater than 21 mm Hg without medicationsand qualified success as IOP of no greater than 21 mm Hg with medications.Unless otherwise indicated, data are expressed as mean ± SD.

Results  Fifty-four goniotomies were performed in 40 eyes of 31 patients. Juvenilerheumatoid arthritis–associated uveitis was the diagnosis in 30 eyes(75%). Eleven eyes (28%) were aphakic. Mean follow-up was 98.9 months (range,2-324 months). Mean age at surgery was 10.3 ± 4.7 years (range, 4-22years). Mean preoperative IOP was 36.7 ± 6.4 mm Hg while receivinga mean of 2.9 ± 1.1 medications. Overall surgical success was achievedin 29 eyes (72%), including success in 22 (55%) and qualified success in 7(18%) while receiving a mean of 1.6 ± 1.1 medications. Mean postoperativeIOP in the success and qualified-success groups were 14.3 ± 2.8 and15.7 ± 3.1 mm Hg, respectively. Kaplan-Meier survival probabilities(95% confidence interval) at 1, 5, and 10 years were 0.92 (0.82-1.00), 0.81(0.65-0.97), and 0.71 (0.49-0.92), respectively. Phakic eyes, eyes with fewerperipheral anterior synechiae, patients younger than 10 years, and eyes withno prior surgery had significantly better outcomes. Hyphema, typically mildand transient, occurred in 43 procedures (80%).

Conclusions  Goniosurgery is low risk and effective for refractory glaucoma complicatingchronic childhood uveitis. It should be considered the surgical procedureof choice for this condition. Surgical outcome is adversely affected by increasedage, peripheral anterior synechiae, prior surgeries, and aphakia.

Uveitis is a significant cause of secondary glaucoma in childhood, andthe most common systemic association by far is juvenile rheumatoid arthritis(JRA).¹ The overall prevalence of glaucomain children with JRA-related uveitis has been estimated to be as high as 22%.² Glaucoma complicating chronic childhood uveitis canresult in significant visual impairment and is considered the most devastatingcomplication of this disorder.¹ Treatment ofthis secondary glaucoma is challenging because it is often refractory to medicaland surgical therapies.^1,3

Existing surgical treatments that are most commonly chosen after failureof medical therapy include trabeculectomy and glaucoma drainage implants.^4-9 Theoutcome of trabeculectomy in this group of patients is adversely affectedby their young age, increased ocular inflammation, prolonged use of multipletopical drugs, and frequent need for cataract surgery, which all predisposeto excessive fibrosis.¹⁰ Although the use ofantimetabolites improves the success rate, they are associated with increasedrates of complications such as bleb leaks, hypotony, and a lifelong risk ofinfection.^8,10-13 Glaucomadrainage implants have been used with variable success and are also associatedwith serious vision-threatening complications such as corneal decompensationand retinal detachment and an increased need for surgical reinterventions.^6,7,14-18 Hence,there is much room for improvement in the current surgical management of childhooduveitic glaucoma.

Goniotomy is the least invasive surgical procedure in the treatmentof childhood glaucoma, which is an advantage in the treatment of uveitic eyes.However, the reported experience with standard goniotomy in the treatmentof childhood uveitic glaucoma is limited. Haas¹⁹ firstdescribed the use of classic Barkan goniotomy in 3 patients with open-angleglaucoma associated with iridocyclitis in whom at least short-term controlof intraocular pressure (IOP) was achieved. Trabeculodialysis, a modifiedgoniotomy procedure in which the trabecular meshwork is disinserted from thescleral sulcus rather than simply incised as in conventional goniotomy, hasbeen reported to have 44% to 60% success in the short term in small numbersof young patients with inflammatory glaucoma, especially in association withJRA.^20,21 Freedman et al²² recently described the efficacy of conventional goniotomyin 12 patients with childhood uveitic glaucoma and found an overall successof 75% with a mean follow-up of 32.4 months. Factors affecting the outcomeof goniosurgery for this condition have not been previously described.

The aim of this report is to describe the long-term efficacy and safetyof standard goniotomy for the treatment of glaucoma secondary to chronic childhooduveitis and the prognostic factors affecting its outcome in the largest knownclinical series with the longest follow-up.

We undertook a retrospective review of all medical records of patientsfor whom goniotomies were performed by a single surgeon (D.S.W.) for childhooduveitic glaucoma refractory to medical therapy. Goniotomy is the first-linesurgical treatment after failed maximal medical therapy in this practice.Patients may have had prior surgeries, including cataract extraction and trabeculectomy,before referral to this practice for uncontrolled glaucoma. Uveitis was controlledwith topical and/or systemic anti-inflammatory therapy as necessary untilanterior chamber cells numbered less than 5 per high-power field using a 3× 1-mm slitlamp beam for at least 1 week before surgery in all eyes.

The main outcome measures were IOP at the last follow-up and time tosurgical failure. A complete surgical success was defined as an IOP of nogreater than 21 mm Hg without glaucoma medications at the final follow-up;qualified success, an IOP of no greater than 21 mm Hg with medications atthe final follow-up; and failure, an IOP of greater than 21 mm Hg despitemedical therapy. Overall success was used to indicate all cases in which successwas complete or qualified. The secondary outcome measure was the number ofmedications needed to achieve an IOP of no greater than 21 mm Hg after surgery.

Intraocular pressures were measured using a Perkins tonometer in alleyes. Direct gonioscopy using a Koeppe lens was performed in all eyes on theoperating table before surgery and at least 2 weeks after goniosurgery wheneverexamination under topical anesthesia in the office setting was possible orunder general anesthesia when a separate surgical procedure was required.The gonioscopic findings, including the number of clock hours of peripheralanterior synechiae (PAS), were recorded by means of detailed drawings in agoniogram at each examination.

Tonography was performed using the standard technique²³ beforeand after goniotomy for the affected eye of an 11-year-old patient in thestudy who achieved complete success with an electronic indentation tonometer(V Mueller and Co, Chicago, Ill). The indentation tonometer was used to measurethe baseline IOP (P₀) after instillation of topical anesthesia.A 4-minute pressure tracing was subsequently obtained with the needle recorderlinked to the tonometer while it was gently applied to the cornea. Its positionon the cornea was maintained until a smooth tracing for 4 minutes was obtained.The P₀ and the change in scale reading during the 4 minutes werethen used to obtain the coefficient of outflow facility (C) from tonographictables.²³

All goniotomies were performed under general anesthesia using the standardgoniotomy technique that has been previously described.²⁴ Briefly,under direct visualization of the angle using a Barkan goniotomy lens, theanterior chamber was entered with a needle goniotomy knife through the peripheralclear cornea opposite the area of the angle to be treated. The knife was directedinto the angle, and an incision was made in the posterior trabecular meshworkfor 4 to 6 clock hours. The anterior chamber was then reformed with balancedsalt solution instilled through the corneal incision, which was then closedwith a 10-0 Vicryl (ETHICON Inc, Somerville, NJ) suture if leakage was observed.Postoperative topical antibiotics and steroids were administered. Topicalsteroid therapy was tapered to the preoperative frequency, usually within2 weeks, according to the extent of anterior chamber inflammation observed.All preoperative systemic anti-inflammatory regimens were continued postoperatively.If a postoperative hyphema was observed, 0.5% apraclonidine hydrochloridewas given for 1 to 2 days after surgery to attempt to minimize it. Head elevationand avoidance of excessive crying, straining, or vigorous activity in theearly postoperative period were recommended to reduce bleeding.

Data were expressed as mean ± SD unless otherwise specified.We used the 2-tailed paired t test to compare preoperativeand postoperative quantitative data and the McNemar χ² testfor related categorical data. Kaplan-Meier survival analysis was performedto calculate probabilities of overall surgical success at different time periodsafter surgery. We measured the association between surgical outcome and possibleprognostic variables using the Pearson correlation for interval and ratiovariables, φ for nominal variables, and Spearman ρ for ranked variables.We compared each of these variables between the different outcome groups usingthe χ² test for categorical variables, t test for continuous variables, and Mann-Whitney U test for ranked variables. All statistical tests were 2 sided, and P values of less than .05 were considered significant.We used SPSS software, version 11.0 (SPSS Inc, Chicago, Ill) in the statisticalanalysis.

We included 40 eyes of 31 patients in this study. A total of 54 goniotomyprocedures were performed. Twenty-eight eyes (70%) underwent 1 goniotomy,10 (25%) underwent 2, and 2 (5%) underwent 3. All except 7 patients were female.All patients were white except 1, who was Asian. The mean age at goniosurgerywas 10.3 ± 4.7 years (range, 4-22 years). Diagnoses included JRA-relateduveitis in 30 eyes (75%), idiopathic uveitis in 8 (20%), and sarcoidosis inthe remaining 2 (5%). Twenty-nine eyes (72%) were phakic, and 11 (28%) eyeswere aphakic at the time of surgery. Sixteen (55%) of the phakic eyes hadposterior subcapsular cataracts before surgery. The mean follow-up was slightlymore than 8 years (98.9 ± 87.8 months; range, 2-324 months). The meanpreoperative IOP was 36.7 ± 6.4 mm Hg, and the mean number of glaucomamedications used preoperatively was 2.9 ± 1.1.

At the last follow-up, 22 eyes (55%) achieved surgical success withoutany glaucoma medications, and 7 eyes (18%), qualified success while receivinga mean of 1.6 ± 1.1 medications. Failure occurred in 11 eyes (28%).This gave an overall success rate of 72% (in 29 eyes), which was achievedwith a single goniotomy in 20 (69%) of these eyes. Postoperative mean IOPwas 14.3 ± 2.8 mm Hg in the complete-success group and 15.7 ±3.1 in the qualified-success group. Figure1 compares the IOP and glaucoma medications preoperatively and postoperativelywithin the different surgical outcome groups. In eyes with overall surgicalsuccess, the mean IOP was reduced from 35.8 ± 5.9 mm Hg to 14.7 ±2.9 mm Hg (P<.001), and the mean number of glaucomamedications was reduced from 3.0 ± 1.2 to 0.4 ± 0.9 (P<.001). The mean number of clock hours of PAS preoperatively was1.6 ± 2.6 compared with 3.1 ± 2.5 postoperatively, indicatinga significant increase in PAS extent after surgery (P<.001).There were no significant differences in the mean frequency of topical steroidsused per day preoperatively and at the last postoperative follow-up in eyeswith a successful outcome (2.1 ± 1.8 and 1.6 ± 1.7, respectively)or in eyes with failed surgery (2.2 ± 1.5 and 1.8 ± 1.5, respectively).There was also no statistically significant difference between the numbersof eyes that required steroids for uveitis control before surgery (29/40 eyes)and at the last follow-up after surgery (28/40 eyes).

The Kaplan-Meier survival plot for overall surgical success is shownin Figure 2. Mean cumulative survivalprobabilities (95% confidence interval) in all eyes with 1 or more goniotomiesat the specified time intervals after surgery were as follows: 0.92 (0.82-1.00)at 1 year, 0.81 (0.65-0.97) at 5 years, 0.71 (0.49-0.92) at 10 years, and0.56 (0.28-0.84) at 13 years, with no failure events occurring after 12 years(144 months). Beyond 12 years after the first goniotomy, 5 eyes were stillbeing followed up and demonstrated surgical success at last follow-up.

Tonography performed in the affected eye of the 11-year-old boy withidiopathic uveitis who had complete success after a single goniotomy showedC values of 0.00 µL/min/mm Hg preoperatively at an IOP of 42 mm Hg,which increased to 0.07 µL/min/mm Hg postoperatively at an IOP of 18mm Hg without glaucoma medications.

A statistically significant association was found between surgical outcomegroups (success, qualified success, and failure) and lens status (P = .004), age at surgery (P = .03), numberof past ocular surgeries (P = .001), and extent ofpreoperative (P = .004) and postoperative PAS (P = .002) in the angle. Phakic eyes had an overall successrate of 86% (25 eyes) compared with 36% (4 eyes) in aphakic eyes (P = .004) (Figure 3A). Patients10 years or younger had greater overall success of 79% (19 eyes) comparedwith 62% (10 eyes) for those older than 10 years (P =.03) (Figure 3B). Success was greatestin eyes without prior surgery (87%; 25 eyes), whereas eyes with more than2 previous ocular surgeries before goniotomy such as cataract extraction ortrabeculectomy had a uniformly poor outcome (100% failure) (Figure 4). The mean number of previous surgeries was 0.2 ±0.6 in eyes with successful outcomes and 1.3 ± 1.3 in eyes with surgicalfailures (P = .03). Eyes with no PAS preoperativelyhad the highest success rate (82%), whereas those with more than 6 clock hoursof PAS had a 100% failure rate. All eyes with no PAS at the last postoperativegonioscopy had operative successes. The mean number of clock hours of preoperativePAS and postoperative PAS was 0.9 ± 1.5 and 2.3 ± 2.0, respectively,in eyes with surgical success and 3.0 ± 3.6 and 4.8 ± 2.9, respectively,in eyes with surgical failure (P = .01 and P = .007 for preoperative and postoperative PAS, respectively). Figure 5 illustrates the relationship betweenpreoperative and postoperative PAS and surgical outcomes.

Sex, diagnosis (type of uveitis), preoperative IOP, number of glaucomamedications before surgery, presence or severity of postoperative hyphema,frequency and use of topical steroids before or after surgery, and presenceof persistent inflammation (defined as the persistent presence of cells orflare requiring topical steroids ≥3 times a day or the recurrent inflammationrequiring increased topical steroids ≥3 times a day for more than 1 weekwithin the last 6 months of follow-up) had no statistically significant correlationwith the surgical outcome. Specifically, there was no difference in the surgicaloutcomes when comparing JRA-related uveitis with other types of uveitis asa group.

Hyphema occurred after 43 goniotomies (80%). This was mild and transientwith complete spontaneous clearance within a week in most affected eyes. In2 eyes, an anterior chamber washout had to be performed for moderate hyphemasassociated with elevated IOP. Deterioration of preexisting posterior subcapsularcataracts occurred early postoperatively in both eyes of a single patientrequiring extraction at 1 and 4 months after surgery. Six other eyes requiredcataract extraction for progression of posterior subcapsular cataracts thatexisted before goniosurgery and became visually significant after goniosurgery(mean time after goniosurgery, 63.9 ± 57.2 months; range, 8-138 months).Of the 8 eyes that had cataract extraction after goniosurgery, 7 continuedto have successful IOP control after that surgery (6 without need for glaucomamedications). In the remaining eye, cataract extraction was performed after3 goniotomies that had resulted in successful IOP control without medications.In this eye, loss of IOP control occurred at 3.5 years after cataract extractionand was refractory to maximal medical therapy, necessitating a trabeculectomywith mitomycin. These progressive posterior subcapsular cataracts were mostlikely due to continued topical steroid use for the uveitis and were not likelyto be related to the goniosurgery. No exacerbations of uveitis in the earlypostoperative period or other significant complications such as infection,iatrogenic damage to intraocular structures, hypotony, or posterior segmentchanges were encountered.

Secondary glaucoma is a frequent and serious complication of uveitis.In this largest known study to report the long-term results of standard goniotomyin childhood uveitic glaucoma, successful IOP control was achieved in 72%of eyes and in 55% without medications at the end of a mean follow-up of morethan 8 years. Cumulative probabilities of overall success of goniosurgeryat 1, 5, 10, and 13 years were 92%, 81%, 71%, and 56%, respectively. The onlyprevious report on the subject by Freedman et al²² showeda similarly high overall success rate of 75% at a mean follow-up of 32.4 months,although most of their patients (75%) required glaucoma medications, and cumulativesuccess probabilities were lower (79% at 1 year and 70% at 15 months). Thedifference was probably due in part to the smaller number of eyes and shorterfollow-up. These results are exciting and suggest that goniotomy can improvethe management of one of the most visually destructive and refractory formsof glaucoma in childhood and young adults.

Compared with the currently favored surgical options to increase aqueousoutflow, standard goniotomy seems to have comparable or higher success ratesbut much fewer risks of complications. Success of trabeculectomy without antimetabolitesin young patients or in eyes with inflammatory glaucoma is generally poor.^1,2,20,25-29 Fewpublished reports address the use of trabeculectomy for uveitic glaucoma inthe young alone. Most include adult patients or other forms of pediatric glaucoma.Trabeculectomy with antimetabolites in adults with uveitic glaucoma have yieldedcumulative success probabilities of 78% at 1 year and 62% at 2 years,⁸ although this is probably lower in children.^11,30,31 Apart from the usualincreased risks of hypotony, bleb leak, cataract formation, and bleb-relatedinfection,^{8,10-13,30,32} thereare further risks of complications associated with the uveitis such as postoperativefibrinous and cellular anterior chamber reaction, which can result in surgicalfailure.^6,33,34 Ahistopathological study of conjunctival biopsy specimens from patients withuveitic glaucoma found that the uveitic conjunctiva contained significantlymore fibroblasts, lymphocytes, and macrophages compared with that of controls,which may help explain the increased risk of bleb fibrosis and failure offiltration surgery in these patients.³⁵ Asa result, glaucoma drainage implants are often recommended for uveitic glaucoma.^4,36 Various types of implants have beenused with reported life-table success rates in adults of up to 94% (1 year),90% (52 months), and 91.7% (24 months) for the Ahmed,⁹ Molteno,⁷ and Baerveldt³⁷ implants,respectively, but this is variable, and relatively lower success rates of57% to 76% have been found in other series.^38-40 Vision-threateningcomplications such as choroidal effusion or hemorrhage, cataract, cornealdecompensation, hypotony, and cystoid macular edema can occur, and surgicalreinterventions for these or other complications such as encapsulated bleb,plate extrusion, corneal-tube touch, and flat anterior chamber are commonlyrequired.^{6,7,9,18,37,38,40}

In contrast, goniotomy has very few complications, and these are limitedalmost exclusively to hyphemas in experienced hands in this study and thatby Freedman et al.²² Due to the minimal invasionand manipulation of intraocular and extraocular tissues, exacerbation of uveitisdid not occur in either study, and the conjunctiva of the eyes undergoingoperation was spared for future filtration surgery if needed. Trabeculotomyab externo, the alternative to goniotomy, has been used with a success ratecomparable to that of goniotomy in the treatment of primary infantile glaucoma.However, its value in childhood uveitic glaucoma is unknown. Although widelyregarded as a different means to the same end, it involves considerable conjunctivalmanipulation and is more traumatic than goniotomy. This may result in significantinflammation, which is undesirable and could affect its success rate in auveitic eye.

Open-angle and closed-angle mechanisms play a role in the pathogenesisof glaucoma in uveitis, although the open-angle mechanisms are more common.^4,41 Complex interactions between severalbiochemical and cellular mechanisms inherent in the inflammatory process occurto cause an elevation of IOP in the presence of an open angle.^4,5,41 Thesuccess of goniotomy suggests that in many cases of glaucoma secondary tochildhood uveitis, an incision in the inner portion of the meshwork wherethe angle is open reduces the resistance to aqueous outflow. The tonographyresults in the uveitic eye with uncontrolled glaucoma confirmed the preoperativereduction in outflow facility (C value of 0.00 µL/min per millimetersof mercury compared with a mean of 0.28 µL/min per millimeters of mercuryfor healthy eyes⁴²), which was increased butnot normalized (C = 0.07 µL/min per millimeters of mercury) after goniotomy.The underlying mechanism for improved outflow facility is unclear. Goniotomymay result in a change of the cellular or biochemical milieu that reducesthe outflow resistance. However, its long-lasting benefit, despite recurrentor persistent inflammation and the quantitative nature of its effect (moreIOP lowering with additional or increased extent of goniotomy incision), suggeststhat a mechanical obstruction of aqueous outflow may exist in the inner portionof the meshwork that is reduced by the goniotomy incision. There was histopathologicalevidence of a persistent communication between the anterior chamber and theSchlemm canal in a specimen taken from an eye with sarcoid uveitis 1 monthafter trabeculodialysis.⁴³ Closure of the angleby PAS formation before or after goniosurgery limits aqueous access to themeshwork. Significant outflow reduction can occur if angle closure is extensiveor obliterates a previously functional goniotomy cleft. The mean number ofclock hours of PAS in eyes with surgical failures in this series was 3 to4 clock hours and was not markedly more extensive compared with eyes withsurgical successes. This was because the surgical failures included eyes withopen angles and other predispositions to failure such as aphakia or multipleprior surgeries. The effect of PAS on surgical outcome could best be illustratedin Figure 5, which shows that eyeswith no PAS had the greatest likelihood of success, whereas eyes with extensivePAS of more than 6 clock hours had no chance of success (100% failure rate).

Older age, prior ocular surgery, and aphakia are probably associatedwith different or more advanced changes in the meshwork and/or more distalparts of the outflow system, which are less likely to be relieved by incisionsin the meshwork by goniotomy alone, hence their association with adverse outcomes.Aphakia in young children is itself a definite risk factor for the developmentof glaucoma for reasons as yet unknown. These findings are in contrast tothose of Kanski and McAllister²¹ on trabeculodialysisfor inflammatory glaucoma in children and young adults, which found no associationbetween aphakia or synechial angle-closure and outcome, and those of Freedmanet al,²² which found no relation between ageand crystalline lens status at surgery with surgical outcome. This may bedue at least in part to the smaller sample size in these studies.

Cataract extraction after goniosurgery did not seem to compromise thesurgical outcome in 7 (88%) of 8 eyes in this series. The only eye that experiencedfailure 3.5 years after cataract surgery also had other predictors of failuresuch as increased postoperative PAS (4 clock hours) and older age at surgery(21.5 years).

Use of topical steroids is known to cause IOP elevations. As the amountand presence of topical steroid use was not significantly different beforeand after goniosurgery, it is unlikely that the IOP reduction was due to reductionof steroid use with time.

Although goniotomy is widely recognized as an effective first-line surgeryfor primary congenital glaucoma, it also now deserves consideration as a relativelysafe and effective first-line surgical treatment for medically uncontrolledglaucoma secondary to chronic childhood uveitis. Unlike filtration surgery,it is less traumatic, has fewer risks, and facilitates aqueous outflow throughthe normal drainage system of the eye rather than creating an artificial pathwaythat bypasses the physiological route. It is effective not only for JRA-relateduveitis but also in idiopathic and sarcoid-related uveitis when the angleis open. Although the success rate of goniosurgery is highest in children10 years or younger, older age up to young adulthood does not preclude success.Aphakia, multiple previous ocular surgeries, and extensive PAS are much strongerpredictors of surgical failure.

In conclusion, goniosurgery is safe and effective for the treatmentof glaucoma complicating chronic childhood uveitis. It should be consideredthe procedure of choice because it restores function of the trabecular tissuewithout bypassing it, is the least traumatic of the glaucoma surgeries, andspares the conjunctiva. Most treated patients do not require glaucoma medicationsafter the procedure. Surgical success is highest in children younger than10 years. Aphakia, multiple prior ocular surgeries, and extensive PAS areadverse prognostic factors for a successful surgical outcome.

Corresponding author and reprints: Ching Lin Ho, FRCSEd, 2 LongfellowPl, Suite 201, Boston, MA 02114-2224 (e-mail: hocl8@yahoo.com.sg).

Submitted for publication June 2, 2003; final revision received September5, 2003; accepted February 12, 2004.

This study was presented in part as a poster at the 73rd Meeting ofthe Association of Vision Research in Ophthalmology; May 5, 2003; Fort Lauderdale,Fla.