Change in Angle of Eyes with Primary Angle Closure Suspects One Year after Laser Peripheral Iridotomy : a randomised study

Background: To report the changes in anterior chamber angle and progression rate to PAC(primary angle closure) following laser peripheral iridotomy (LPI) in primary angle Methods: Prospective, randomized controlled interventional clinical trial conducted at the Handan Eye Hospital, China. 134 bilateral PACS, defined as non-visibility of the posterior trabecular meshwork for ≥ 180 degrees on gonioscopy were randomly assigned to undergo LPI in one eye. Gonioscopy and Goldmann applanation tonometry were performed prior to, on day 7 and 12 months’ post LPI. Results : 80 of 134 patients (59.7%) could be followed up at one year. The mean IOP in treated eyes was 15.9±2.7 mmHg at baseline, 15.4±3.0 mmHg on day 7; 16.5±2.9 mmHg at one month and 15.5±2.9 mmHg at 12 months; the IOP in untreated eyes was similar (p=0.834). One or more quadrants of the angle opened in 93.7% of the LPI treated eyes, but 67.0% (53/79) remained closed in two or more quadrants. The progression rate to PAC in untreated eyes was 3.75% and one developed acute angle closure glaucoma(AACG), the progression rate to PAC(primary angle closure) in treated eyes 2.5% in treated eyes, none had developed PAS or AACG. Conclusion: LPI can open some of the occludable angle in the majority of eyes with PACS, but 67% continue to have non-visibility of the trabecular meshwork for over 180 degrees. IOP remained similar in treated and untreated eyes. Further research is needed to determine the full implications of residual closure as well as the need for follow up and treatment in PACS. The cumulative incidence for PAC/AACG in treated eyes were not significantly different from untreated eyes.

A knowledge of the natural history of PACS and the effect of LPI would help in public health strategy as well as individual management decisions [16]. In a 6 year community based follow up study in urban China, Ye et al. reported that 4.1% (20/485) PACS [defined as less anterior chamber depth (ACD ≤2.0mm, or limbal ACD≤1/4 corneal thickness(CT); or Iris light band ratio≤1/4 with oblique flashlight test), developed PAC or PACG [17]. In a population-based Indian cohort that used gonioscopy for the definition, 22% of PACS progressed to PAC and 28% of PAC progressed to PACG over 5 years; none of the PACS developed PACG in the 5 years of follow up [18].
The results of LPI in PACS are variably reported. In a retrospective hospital based case series, LPI controlled IOP over 5 years in 16/18 (90%) Chinese eyes with PACS [19]. In a two-year hospital based follow up study none of 27 eyes with PACS undergoing LPI progressed to PAC or PACG [20]. However, a population-based clinical trial reported that 6.7% of PACS in Mongolia progressed to PAC after LPI in 6 years [21].
Considering the large number of PACS in China and the potential significance of angle closure as a public health problem, we conducted a randomized trial, the aim of the present study was to investigate the efficacy of LPI in PACS in one eye and chart the course of untreated fellow eyes.
Herein we report the one year results of IOP and angle changes in this trial. Our study adheres to CONSORT guidelines.

Participant and study design
The study was approved by the ethics committee of the PACS was defined as non-visibility of the filtering trabecular meshwork for ≥180 degrees on an "over the hill" view on gonioscopy (one mirror Goldman lens in dim illumination), without PAS and no clinically evident glaucomatous optic damage or visual field change [22,23].
An incident event of AACG or PAC was the primary outcome. AACG was characterized by a combination of acute symptoms of pain, headache, blurred vision and haloes around lights with signs of ischemic iris changes, corneal edema, glaucomflecken, and elevated IOP above 30 mmHg. PAC was defined as PACS with IOP >21mmHg on two separate occasions and / or peripheral anterior synechia of 0.5 clock hours.
Goldmann applanation tonometry was performed by a certified clinical nurse prior to LPI and on day 7, one month and 12 months post LPI. At each visit, the mean of 3 readings was recorded. Gonioscopy was carried out by one glaucoma specialist (FSJ) who was blinded after assignment to the treatment prior to LPI, day 7, one month and 12 months post LPI, using a Goldmann-type 1-mirror lens with low-ambient illumination that did not impinge on the pupil. This was followed by dynamic gonioscopy using the same lens to confirm the absence of peripheral anterior synechia (PAS). The inter-observer reproducibility for gonioscopy between FSJ and another glaucoma specialist for clock hours of PAS was perfect [Intraclass correlation (ICC) =0.972].

Randomization and Allocation Concealment.
The SPSS program generated a series of numbers to randomly select the right or left eye of the participants to be treated with LPI. Allocation concealment was achieved by involving a research nurse(ZXY) in the process: when a patient met the criteria for enrollment, the ophthalmologist(FSJ) involved in this study contacted the research nurse who communicated the allocation.

Interventions
This study followed routine clinical practice. LPI was performed with an Abraham contact lens in the superior (10:00 to 2:00 o'clock) region of the iris by FSJ or LYB using a Nd:YAG laser (YL-1600; NIDEK Co., LTD, Japan). If the IOP measured 1 hour after iridotomy was ≥ 30 mmHg oral acetazolamide (250mg) was given.
Due to non-availability of plain steroid drops, the eye undergoing LPI eye was treated with Tobradex eye drops (Alcon Laboratories, Fort Worth, TX four times daily for 3 days.

Sample size estimation
Based on an expected 22% incidence of PAC in control PACS and reduction to 5% with LPI 22 , a sample size of 116 patients was calculated to allow demonstration of superiority at the 5.0 % significance level with a power of 80%. Anticipating a loss to follow up of 10% per year, the sample size was increased to 177. Enrollment was slow and 134 eligible subjects were recruited between October 2005 and January 2008.

Statistical Analysis
All analyses were performed using SAS 9.0.3 statistical software (SAS Inc., Chicago, IL). Data from the 1-year visit were used for analysis.
A paired t-test was used to compare the change in VA (LogMAR), IOP, Spherical equivalent (SE), anterior chamber depth (ACD), lens thickness (LT) and axial length (AXL) in the treated eye to that in the untreated fellow eye. We used a general linear model to test the difference in IOP with repeated measurements. The means and standard deviations (SD) were calculated for continuous outcome variables with a normal distribution. Statistical significance was determined using the Student's t-test (normal distribution) or rank-sum test (non-normal distribution). To compare the incidence rate of PAC/AACG between treated eyes and untreated eyes, we used Fisher exact test (1-sided). P< 0.05 was considered to be statistically significant.

Characteristics of the participants
191 subjects were eligible for the study. 12 patients who declined to participate and 45 who refused randomization were excluded. 134 patients were followed up for one year and one eye was treated with LPI at random. 10 subjects were lost to follow up on day 7, 23 subjects were lost to follow up at The baseline characteristics and the quadrants of non-visible trabecular meshwork in the treated eyes was not significantly different from the fellow eye (Table 1). Since the drop off rate was high, we compared the baseline characteristics between the those who attended follow up and those who did not attend follow up: there was no significant difference in age, gender, ocular parameters or quadrants of non-visible meshwork. The IOP in the participants who missed the 1-year follow-up was a little lower, and they had better VA than the participants who attended (Table 2).

IOP outcomes
The mean IOP in the treated eyes was 15.9±2.7 mmHg at baseline, 15.4±3.0 mmHg on day 7; 16.5±2.9 mmHg at one month and 15.5±2.9 mmHg at 12 months. The change in IOP between the baseline and follow up visits were very similar in the treated eyes and the untreated fellow eyes at all follow up visits (Figure 2). IOP in eyes with four, three and two quadrants of non-visible trabecular meshwork pre operatively decreased by 0.82 ±3.3 mmHg, 0.14 ±3.4mmHg and 1.6±3.5mmHg respectively. There was no difference between treated and untreated eyes (p=0.440~0.612).

Gonioscopy outcomes
79 patients underwent gonioscopy at the 12 month visit. Five of the untreated eyes (6%) showed one quadrant of increase in "closure" but none developed PAS (Table 3) (Table 3 ).

Progression rate to PAC outcomes
Five of the 80 patients who attended the 1 year follow up had developed PAC or AACG. Those who progressed were females aged 49 to 69 years. Of the untreated eyes, one developed AACG while two eyes recorded an IOP >21 and were classified as PAC, the progression rate to PAC in untreated eyes was therefore 3.75% (95% CI, 0 -7.9%). Two of the treated eyes had an IOP above 21mmHg and were classified as PAC (2.5%; 95% CI, 0 -5.9%); none had developed PAS or AACG. The cumulative incidence for PAC/AACG in treated eyes were not significantly different from untreated eyes (p=0.650).

Discussion
This randomized study found that at one year 3.75% of untreated PACS fellow eyes progressed to PAC / AACG; however, in this sample with a small number of events LPI did not significantly reduce the incidence of PAC. There was no significant reduction of IOP following LPI and 67.0% (53/79) of treated eyes continued to have non visibility of trabecular meshwork in two or more quadrants.
In our study, we found that the angle opened in at least one quadrants in 93.7% of the PACS eyes which is consistent with the reported role of pupillary block in angle closure disease among the Chinese population [24,25]. However, following LPI about 2/3 of the PACS eyes did not open in 2 or more quadrants; and 17.8% did not open in 3 quadrants or more. This result is very similar to that of a population-based study from southern China in which about 19.4% still had 3 or more quadrants of non-visibility of meshwork following laser iridotomy [24]. Previous studies had reported that 37% to 60% of Chinese eyes undergoing LPI for early primary angle closure were still positive on the dark room prone provocative test [26,27]. Non responsive cases may have some of the multiple mechanisms of angle-closure reported in Asian eyes [24,25,[28][29][30].
Several studies have reported an association of IOP and angle width. Foster estimated a 0.2 mm Hg increase per 10° change in width in all four quadrants [31,32]. He et al reported a 3.1mmHg reduction in mean IOP at 2 weeks' post LPI [25], while Hisao et al observed a reduction of 2.3mmHg in mean IOP after LPI [33]. We did not observe significant IOP reduction following LPI at any of the follow up visits, did not find an association of IOP with the number of non-visible quadrants, and the change in IOP was similar to the fellow untreated eyes. In the present study, we found that at 7 days, one month and one year after LPI, mean IOP rise and fall in the treated and fellow untreated eyes almost simultaneously ( Figure 2). Such effect seems to occur after trabeculectomy [34], Diestelhorst [32] studied the effect of trabeculectomy on the aqueous humour flow of the unoperated fellow eye, he concluded that filtration surgery in one eye triggers a CNS mediated, reflective increase in aqueous flow to maintain physiological stability in the anterior chamber of the surgically treated eye,we supposed that LPI may have the same effect as trabeculectomy.
The incidence of PAC/AACG in the untreated eyes in our study was 3.75%, which was very similar to that reported in a population based Indian cohort 4.4% per year [22]. In Liwan study, approximately one in five people aged 50 years and older developed some form of angle closure over a 10-year period [36]. However, two of the LPI treated eyes also developed increased IOP without PAS in our study. All cases classified as progressing to PAC were based on recording an IOP > 21 mmHg. While a cut off is required for trial purposes, a single IOP recording could be erroneous and would not be considered clinically significant. While it is possible that indentation gonioscopy may have revealed differences in PAS between groups, it seems that any benefit of LPI at one year in preventing PAC is likely to be minimal, would not justify laser iridotomy for all and therefore cannot justify population based screening for PACS [16,18,21,26,28,[37][38][39].
The major limitation of our study is the loss to follow up of 40% at one year, much higher than expected; 26 of those who did not attend follow up were contacted by phone and confirmed absence any symptoms of AAC. In addition, HEH is the only eye hospital in local area. As all subjects were informed and aware about the symptoms of angle-closure glaucoma and that free eye care would be available, it is unlikely that they had symptoms but did not attend. Also the baseline characteristics of patients who did or did not attend follow were similar for the treated eyes and the untreated eyes.
Based on the PAC rate in this study, the power is low. The referred rate was 22% at 5 years. It may be due to different population. Accordingly, we believe pathology if any, in subjects lost to follow up was likely to be PAC, not AACG.
Another limitation is the subjective nature of gonioscopy for angle closure. Although the intraobserver agreement of gonioscopy for angle closure sounds good, in the untreated group one third of the cases had a wider angle comparing to the baseline, which probably represents variablility in goniosocpy.
The changes in the angle were however different in treated compared to untreated eyes.

Conclusions
The present registry study indicated that 3.75% of untreated PACS fellow eyes progressed to PAC / AACG, a rate of progression similar to that reported in the literature [22]. The progression rate to PAC in LPI treated eyes was lower than untreated eyes.IOP was not reduced significantly after LPI and about two thirds of PACS continued to have two quadrants of non-visible trabecular meshwork,