Article Text

Download PDFPDF

Effect of prophylactic laser iridotomy on corneal endothelial cell density over 3 years in primary angle closure suspects
  1. Rajesh S Kumar1,2,
  2. Mani Baskaran1,
  3. David S Friedman3,
  4. Ying Xu4,
  5. Hon-Tym Wong5,
  6. Raghavan Lavanya1,
  7. Paul T Chew6,7,
  8. Paul J Foster8,
  9. Tin Aung1,6,7
  1. 1Singapore National Eye Centre and Singapore Eye Research Institute, Singapore
  2. 2Narayana Nethralaya, Bangalore, India
  3. 3Wilmer Eye Institute, Johns Hopkins Hospital, Baltimore, Maryland, USA
  4. 4Singapore Clinical Research Institute Pte, Singapore
  5. 5Department of Ophthalmology, Tan Tock Seng Hospital, Singapore
  6. 6Department of Ophthalmology, National University Hospital, Singapore
  7. 7Yong Loo Lin School of Medicine, National University of Singapore, Singapore
  8. 8National Biomedical Research Centre for Ophthalmology, UCL Institute of Ophthalmology and Moorfields Eye Hospital, London, UK
  1. Correspondence to Dr Tin Aung, Glaucoma Services, Singapore National Eye Centre, 11 Third Hospital Avenue, Singapore 168751, Singapore; aung_tin{at}


Objective To assess the effect of prophylactic laser peripheral iridotomy (LPI) on corneal endothelial cell density (ECD) and morphology in primary angle closure suspects (PACS) over 3 years.

Methods In this prospective cohort study, subjects underwent LPI in one eye, while the fellow eye was untreated. Specular microscopy was performed at baseline and after 1 and 3 years. Central corneal ECD and morphology of both eyes were assessed using non-contact specular microscopy (Konan SP-9000LC, Konan Inc, Hyogo, Japan).

Results 230 subjects completed 3-year follow-up. The mean age was 62.5±8.0 years, and the majority of subjects were Chinese (92.3%) and women (75.4%). In eyes that underwent LPI, ECD was significantly lower at year 1 (2462.3, 95% CI 2414.5 to 2510.0, p<0.0001) and year 3 (2510.6, 95% CI 2462.1 to 2559.2, p=0.0006) compared with baseline (2609.1, 95% CI 2551.4 to 2666.7). There was also a significant decrease in ECD in fellow untreated eyes from baseline to year 1 (p<0.0001) and year 3 (p=0.01). The decrease in ECD at 3 years compared with the baseline in treated and untreated eyes was similar (2.1% vs 0.9%, p=0.20).

Conclusions In PACS eyes, there was decrease in ECD in LPI-treated and control eyes over 3 years, with no significant difference between groups.

  • Glaucoma
  • Epidemiology
  • Treatment Lasers

Statistics from

Primary angle closure glaucoma (PACG) is a leading cause of blindness in East Asia.1 In China, PACG accounts for 91% of bilateral glaucoma blindness, and 28 million people are classified as primary angle closure suspects (PACS), the anatomical trait predisposing to PACG.1

Laser peripheral iridotomy (LPI) is the established first-line therapy for angle closure.2–4 However, the balance between risk and benefit for prophylactic LPI in PACS eyes is unclear.5 ,6 Reported complications of LPI include laser burns to the cornea, lens or retina, corneal oedema and development of posterior synechiae.7 Reports of the effect of LPI on cornea endothelial cell density (ECD) have been variable, with some articles citing no change over time, and others demonstrating significant decrease in ECD after LPI, and even focal or generalised corneal decompensation.8–11 There are few reports discussing the effect of LPI on endothelial cell morphology.

As PACG is an important problem in many countries where access to care is often limited, it is important to assess the effect of prophylactic LPI on the cornea endothelium in PACS eyes to ensure that a procedure, performed to prevent visual loss from PACG, does not cause further morbidity. The aim of this study was to assess the effect of prophylactic LPI on corneal ECD and morphology in a cohort of Asian PACS over a 3-year period.


This was a prospective cohort study nested within a randomised controlled trial (NCT00347178, Subjects over the age of 50 years diagnosed as PACS underwent prophylactic LPI in one randomly selected eye, while the fellow eye served as a control. The randomisation was computer generated with concealment of allocation. Approval for the study was obtained from the Institutional Review Board of Singhealth, and the study adhered to the tenets of the Declaration of Helsinski. All patients signed a written informed consent.

PACS was defined as the presence bilaterally of appositional contact between the iris and posterior trabecular meshwork for at least 180° on gonioscopy, with intraocular pressure (IOP) <21 mm Hg, absence of peripheral anterior synechiae or glaucomatous optic neuropathy (defined as focal notching or thinning of neuroretinal rims or cup:disc asymmetry >0.2) or visual field changes compatible with glaucoma. Subjects with secondary angle closure, history of intraocular surgery or penetrating eye injury, corneal disorders, evidence of a prior acute angle closure attack, cataract with visual acuity worse than 20/40, and retinal diseases were excluded.

All subjects underwent an ophthalmic examination at baseline, including assessment of visual acuity, IOP by Goldmann applanation tonometry, dynamic indentation gonioscopy using a Sussman four-mirror gonioscope (Ocular Instruments Inc, Washington, USA), fundus examination and automated static perimetry (SITA-Standard 24-2 test, Humphrey Visual Field Analyzer, Carl Zeiss Meditec, Dublin, California, USA). Measurements of axial length (AL), anterior chamber depth (ACD) and lens thickness (LT) with A-scan ultrasonography (Echo Scan, Nidek, Aichi, Japan.) and central corneal thickness (CCT) (Nidek US-1800, Aichi, Japan) were also performed.

Sequential argon-Nd:YAG LPI was performed on eyes randomised to LPI within a week of the baseline visit. These eyes were pretreated with pilocarpine 2% and brimonidine 0.15% eye drops. LPI was performed in the superior region (10 to 2 o'clock) using an Abraham lens. IOP was measured 1 h after completion of LPI; if IOP increased >5 mm Hg, subjects were treated with oral acetazolamide 250 mg (provided there was no contraindication). All patients were prescribed betamethasone 0.1% eye drops applied four times a day for 1 week after LPI.

All test procedures that were performed at baseline were repeated at years 1 and 3 after LPI.

Endothelial cell imaging and analysis

The central corneal endothelium of both eyes was assessed using non-contact specular microscopy (Konan Noncon Robo SP-9000LC, Konan Inc, Hyogo, Japan). This instrument automatically captures images of the endothelium once the subject fixates on a target and records over a field size of 0.10 mm2. ECD, mean cell area, coefficient of variation of cell area (CV) and percentage of hexagonal cells were determined using the Konan-Robo center endothelial analysis method. This method is semi-automatic; the centre of each field (at least 100 contiguous cells) is hand marked on the screen and a computer algorithm calculates the values. Cell density was recorded as the number of cells per mm2. The mean cell area (mm2) and CV (%) in cell area (SD divided by the mean cell area) were used as an index of the extent of variation in cell area (polymegathism). The percentage of hexagonal cells in the area analysed (hexagonality, %) was used as an index of variation in cell shape (pleomorphism). Each image was graded by one of two independent observers who performed cell marking. Inter-observer reliability for all images was determined using intra-class correlation.

Statistical analysis was performed using the STATA statistical programme (V.10.1). Paired t test was used to compare the difference in the continuous outcomes. A two-sided p value <0.05 was considered to be statistically significant. An analysis of covariance analysis was also performed to adjust for potential prognostic covariates, including AL, ACD and LT.


Two hundred and thirty subjects who were PACS completed 3 years of follow-up and were included in the analysis. The majority of subjects were Chinese (92.3%) and women (75.4%). The mean age was 62.5±8.0 years (range 58–68 years). Sixteen subjects had repeat LPI for small size of the LPI; all had a maximum of one repeat procedure and this was within the first year. Analysis comparing specular microscopy data at baseline, year 1 and 3 was performed only on 219 patients with non-missing values for all endpoints.

At baseline, AL, ACD, LT, mean ECD, hexagonality, number of cells analysed and mean CCT were similar for LPI-treated and fellow eyes (table 1). However, the mean cell area and CV differed between the two eyes (p<0.05) at baseline.

Table 1

A-scan, specular biomicroscopy and pachymetry parameters for treated and fellow eyes at baseline

In eyes that underwent LPI, ECD was significantly lower at year 1 (p<0.0001) and year 3 (p=0.0006) compared with baseline (table 2). There was also a significant decrease in ECD in fellow untreated eyes from baseline to year 1 (p<0.0001) and year 3 (p=0.01). The percentage decrease in ECD at year 1 in the treated eyes and fellow eyes was comparable (p=0.71) (table 2). There was no significant difference in ECD decrease at year 3 between treated and control eyes (p=0.20).

Table 2

Endothelial cell density (per mm2) in treated and fellow eyes for the 219 subjects with complete data at baseline, year 1 and year 3 study visits

The mean cell area increased at year 3 compared with baseline (p=0.005) in LPI-treated eyes (table 3), but not in fellow eyes; the change over the 3-year follow-up was not statistically significant between the treated and fellow eyes (p=0.12) (table 4). The CV was significantly different in the treated eyes at year 3 compared with baseline (p=0.002) and in untreated fellow eyes (p=0.0002). However, the change in CV over time was similar for treated and fellow eyes at year 3 (p=0.57) (table 4). Cell hexagonality also increased over 3 years in the treated group (p=0.0001) and fellow eyes (p=0.0001), but again the change was similar for both groups (p=0.93) (table 4). The mean CCT was significantly different in the treated eyes at year 3 compared with baseline (p=0.009) (table 3) and in untreated fellow eyes (p=0.002). The change in mean CCT over time was similar for treated and fellow eyes at year 3 (p=0.13) (table 4).

Table 3

Mean changes in specular biomicroscopy and pachymetry for treated eyes at baseline and year 3

Table 4

Comparison of changes over 3 years in endothelial cell characteristics and pachymetry in treated and fellow eyes

Table 5 shows that AL, ACD and LT did not affect the change in ECD after LPI. CCT was comparable for treated and fellow eyes at baseline (p=0.37) and at year 3 (p=0.33).

Table 5

Effect of laser peripheral iridotomy on endothelial cell density change (year 3 – baseline), after adjusting for baseline characteristics (using repeated measure analysis of covariance)

Four of 230 patients had an IOP spike (defined as IOP >21 mm Hg) an hour after LPI; all were prescribed oral acetazolamide as per protocol. None of these eyes had IOP >21 mm Hg at week 1 follow-up or required additional topical IOP-lowering treatment. There was no significant change in the endothelial parameters in any of these patients at year 1 or 3.

Intra-class correlation for inter-observer reliability for the ECD measurement was high (0.88; 95% CI 0.72 to 0.95).


Three years after sequential argon-YAG LPI, fellow and treated PACS eyes had declines in ECD. While the decline was nearly twice as large in LPI-treated eyes, this difference was not statistically significant. Several hypotheses have been proposed for the cause of endothelial cell loss relating to LPI: direct endothelial laser burns, thermal damage from the iris or aqueous temperature spikes, transient increase in IOP and mechanical forces with turbulent flow of aqueous during the LPI.7–12 More recently, time-dependent effect of shear stress on the endothelium has been proposed to cause endothelial cell loss.13

There have been variable reports of the effect of LPI on the corneal endothelium, with most reporting no significant short-term or long-term endothelial damage.14–16 Some case series have documented significant endothelial loss and corneal decompensation in eyes that underwent argon LPI.10 ,14–16 Interestingly, in a trial of argon LPI in one eye of subjects with PACG and YAG LPI in fellow eyes, there was a significantly lower percentage decrease in endothelial cell counts in the YAG group (0±5% vs 8±7% in the argon group, p<0.01).17 Recently, Ang et al reported that almost 30% of patients undergoing penetrating keratoplasty for bullous keratopathy in a hospital from Japan had previously undergone prophylactic argon LPI (after an average of 4 years).11 However, the high risk of endothelial failure in this series may be related to the laser technique used. As stated by Ang et al, the technique predominantly used was that of solely argon LPI. In our clinical practice, we perform sequential argon-YAG LPI18 as it allows the production of a patent iridotomy using a fraction of the energy needed by argon laser alone.19 In the UK, Nd:YAG LPI is the standard technique in Caucasians with thin irides. However, a recent randomised controlled trial from London demonstrated better patency rates and lower side effect rates in Asian and African people undergoing sequential LPI.20

There are large variations in the reported change in ECD over time. The mean exponential cell loss in normal Caucasian eyes over a 10-year period has been reported to be 0.6%±0.5% per year or about 16±13 cells/mm/year.21 The rate of cell loss in normal Indian and Chinese eyes has been found to be around 0.3%.22 ,23 In our cohort of PACS subjects, the loss was around 3.6% at 1 year and 2% at 3 years in those who underwent LPI versus 3.2% at 1 year and 0.9% at 3 years in fellow eyes. The high rate of endothelial loss in the untreated eyes in our cohort is a striking finding, as it is substantially higher than reported before. The key difference was that ours was a longitudinal study (not cross-sectional, as were previous studies) and our participants had irido-trabecular contact (ITC). There are no studies to date on the natural history of endothelial cell changes in eyes with ITC. It is possible that some form of subclinical intermittent asymptomatic closure in some of these eyes may be causing a higher rate of endothelial cell loss than expected. Although mean CCT showed statistically significant change at year 3 follow-up, the values were not clinically significant (541 μ vs 543 μ), which can be attributed to inter-observer variation.

An error of 10% for ECD measurement has been reported to be acceptable, and changes of less than 200 cells may be attributed to measurement variability due to the instrument.24 Bourne found a 7% inter-visit variability for ECD counting using the specular microscopy.25 In our study, the difference in the number of endothelial cells was less than 100 cells/mm2 at year 3 in the treated group and around 70 cells/mm2 in fellow eyes (table 4); this difference could have been an instrument-related variation. We also noticed that there was an ‘increase’ in the number of cells (80–100 cells/mm2) at year 3 compared with year 1 in treated and fellow eyes (table 2), which could be attributable to inter-visit variability. It is relevant that the 95% CIs of our baseline, year 1 and year 3 ECD figures for treated and untreated eyes overlapped.

An important limitation of this study was that we examined parameters only in the central field of the cornea whereas LPI was performed in the periphery. This is because it is technically difficult to record the ECD and morphology in the peripheral cornea with currently available instrumentation. The inter-visit variability that we discussed earlier could also have been a confounder.

In summary, we observed a significant decline in ECD in LPI-treated and untreated eyes after 1 and 3 years, relative to baseline. Although the decline in treated eyes appears to have been greater, this difference was not statistically significant at both time points. The question of whether the observed decline in ECD is clinically significant remains to be fully answered. There is a need for additional longitudinal data on ECD in angle-closure disease, and on the effect of prophylactic LPI on the corneal endothelium as these findings may have implications for the prevention of angle closure blindness worldwide.



  • Contributors Design of the study (DSF, PTC, PJF, TA); conduct of the study (RSK, MB, RL, HTW, PTC, TA); collection and management of data (RSK, MB, RL, TA); analysis and interpretation of data (RSK, MB, YX, TA); preparation of manuscript (RSK, MB, TA); review or approval of manuscript (DSF, HTW, PTC, PJF, TA).

  • Funding This study was funded by a grant (no. R358/16/2004) from the National Medical Research Council, Singapore. The authors acknowledge support to Paul Foster from the UK National Institute for Health Research, Fight for Sight and the RD Crusaders Charitable Trust.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval Institutional Review Board of Singhealth. The study adhered to the tenets of the Declaration of Helsinski.

  • Provenance and peer review Not commissioned; externally peer reviewed

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.