EyeWorld journal club
Duke University residents' review of "Impact of phacoemulsification on failure of trabeculectomy with mitomycin-C"

Pratap Challa, M.D.
Residency program director,
Duke University

This month, I asked the residents at Duke to review this Japanese paper looking at whether phaco adversely
affects the function of a prior
trabeculectomy.

—David F. Chang, M.D.,
chief medical editor

In the article published this month in JCRS, "Impact of phacoemulsification on failure of trabeculectomy with mitomycin-C," Nanako Awai-
Kasoaka et al. attempted to further elucidate the effect of phacoemulsification on IOP control after trabeculectomy. To date, published studies report conflicting answers to this question. This retrospective study evaluated phacoemulsification after trabeculectomy with mitomycin-C (MMC) as a risk factor for subsequent failure of trabeculectomy surgery. The study included 178 patients who had undergone trabeculectomy with MMC for either primary open-angle glaucoma (POAG) or pseudoexfoliation glaucoma (PXG). Thirty-seven of these patients subsequently underwent phacoemulsification. The primary endpoints were failure of the trabeculectomy as defined by either: condition A (persistent intraocular pressure [IOP] greater than 21 mm Hg or additional glaucoma procedures) or condition B (persistent IOP greater than 18 mm Hg or additional glaucoma procedures). The authors used the Cox proportional hazard model to identify the relative risk of trabeculectomy failure following phacoemulsification cataract surgery.
The authors found the 3-year probability of treatment success for all patients who underwent trabeculectomy with MMC to be 92.7% for condition A and 81.8% for condition B. Of those patients who underwent phacoemulsification after trabeculectomy with MMC, the probability of treatment success at 5 years for patients who underwent phacoemulsification more than 1 year after trabeculectomy with MMC was 92.6%, whereas for patients who underwent phacoemulsification within 1 year of trabeculectomy with MMC, the corresponding probability of success was 48.0%. Using the Cox proportional hazard model, the authors found that for condition A, higher IOP prior to trabeculectomy was a risk factor for surgical failure (p=0.01). For condition B,
significant risk factors included higher IOP prior to trabeculectomy (p=0.0006), as well as phacoemulsification within 1 year after trabeculectomy (p=0.04). The relative risk for surgical failure as defined by condition B for patients undergoing phacoemulsification within 1 year of trabeculectomy was 2.87, while the relative risk of surgical failure increased by 1.09 for each mm Hg additional pre-trabeculectomy IOP.
Considering the complexities and risks associated with trabeculectomy surgery, it is vital to identify and understand risk factors for surgical failure. This study suggests that phacoemulsification within 1 year of trabeculectomy is a risk factor for surgical failure. The authors suggest that this may occur via scarring of the bleb by post-phacoemulsification inflammation. While it has been attempted to address this
question in prior studies, the present investigation uses the Cox proportional hazard model (a time-to-event analysis estimating the relative risk of surgical failure based on multiple patient characteristics) to estimate the relative contribution of individual risk factors to trabeculectomy failure. However, with a borderline p-value and lack of an adjustment for multiple comparisons, further studies are warranted to support the authors' conclusion.
While the statistical analysis is unique for this specific question, the study design introduces several limitations. As noted by the authors, the retrospective design introduces susceptibility to bias from patient selection and other factors. Another limitation is sample size: 37 patients underwent phacoemulsification, and only 10 underwent phacoemulsification within 1 year post-trabeculectomy. The data show that these 10 patients had a much lower probability of good long-term IOP control. However, given the small sample size, an analysis of baseline characteristics of this group compared to the other patients could have been informative. In addition, the article would benefit from an analysis of whether these patients' IOP control might have fared better had they been randomized to wait longer for their cataract surgery. It may have also been informative to provide an analysis of average IOP change after cataract surgery in these patients since cataract surgery itself can affect IOP. No patient details are presented to address the question of whether patients who underwent phacoemulsification soon after trabeculectomy might have had a more complex post-op course (e.g., flat anterior chamber) contributing to surgical failure, or whether patients having phacoemulsification earlier within 1 year fared worse than those having it 12 months post-trabeculectomy.
Another limitation is the use of multiple similar endpoints. The distinguishing factor between condition A and condition B is the IOP limit of 21 versus 18. This 3 mm Hg difference does not seem clinically significant given that target pressures for patients undergoing trabeculectomy are usually much lower depending on the degree of glaucoma severity. However, including both of these endpoints doubles the number of statistical comparisons and could decrease the strength of the study's conclusions due to multiple comparisons.
Another important issue is the post-phacoemulsification treatment regimen, which in this study included fluorometholone and levofloxacin for 1 month. It is typical practice in the United States to use more potent topical steroids following phacoemulsification, especially for eyes that might be susceptible to inflammation, e.g., those with thick irides or those who have recently undergone trabeculectomy. It is not possible based on this study to determine whether a more aggressive anti-inflammatory post-op regimen would have decreased the risk of surgical failure.
An additional limitation of the study is the inclusion of patients with both POAG and PXG. It is not possible to determine based on the data presented whether the glaucoma diagnosis played any significant role in the patients' post-op outcomes. Although the univariable analysis was not significant, it is not clear if the study was adequately powered to detect a difference. Furthermore, additional analyses would be useful to address the impact of other factors such as why different MMC dosages were used in the study, the amount of phaco power required during surgery (reflects density of cataract), and whether there is a difference in failure rates among limbus or fornix-based incisions. Finally, the exclusion criteria, which include pre-trabeculectomy IOP less than 21 as well as any prior ocular surgery, may limit the generalizability of these conclusions to glaucoma patients that do not meet these criteria.
In summary, this study provides important considerations prior to performing phacoemulsification surgery in post-trabeculectomy patients. The authors report that higher pre-trabeculectomy IOP and phacoemulsification within 1 year after trabeculectomy are important risk factors for trabeculectomy failure. Due to the limitations noted above, the study cannot address some important clinical questions such as whether a more intensive anti-inflammatory regimen after phacoemulsification helps prevent surgical failure and whether it would be beneficial for patients who develop cataracts soon after trabeculectomy to defer phacoemulsification until 1 year post-trabeculectomy. Future prospective randomized trials will be required in order to address these questions and provide important guidance for clinicians.

Contact information

Challa: pratap.challa@duke.edu