Trabeculectomy is an effective method for lowering

Transcription

1 ORIGINAL STUDY Refractive Outcome of Cataract Surgery in Eyes With Prior Trabeculectomy: Risk Factors for Postoperative Myopia Oliver L. Yeh, MD, Karine D. Bojikian, MD, Mark A. Slabaugh, MD, and Philip P. Chen, MD Purpose: To examine refractive outcomes after phacoemulsification in eyes with prior trabeculectomy. Design: Retrospective observational case-control study Methods: Comparison of eyes of patients undergoing cataract surgery after trabeculectomy (study group) with a matched group with medically controlled (control group). Laser interferometry was used to obtain ocular biometry. We measured the difference between the expected and actual postoperative refraction using third-generation and fourth-generation intraocular lens (IOL) prediction formulae (Haigis, Holladay 2, Hoffer Q, and SRK-T). A residual difference of >1.0 D of hyperopia or myopia was considered a refractive surprise. Results: In total, 86 eyes (85 patients) were included, including 23 eyes (22 patients) in the study group and 63 eyes (63 patients) in the control group. The mean follow-up was 12.2 ± 4.1 months. Eyes (n = 13) with trabeculectomy and a preoperative intraocular pressure (IOP)r9 mm Hg had significantly more large myopic surprises than the control group for all IOL formulae (P = Haigis, P = Holladay 2, P = Hoffer Q, P = SRK-T). Eyes (n = 10) with trabeculectomy and preoperative IOP > 9 mm Hg, however, did not have significantly more myopic errors than the control (P > 0.05, all formulae). An IOP spike defined as a >50% rise in IOP from baseline within 1 month of cataract surgery in the subgroup with preoperative IOPr9mmHg (n = 8) was associated with increased risk of large myopic surprise (3/8 subset vs. 1/63 control eyes for all formulae; P = Haigis, P = Holladay 2, P = Hoffer Q, P = SRK-T) as well as for large myopic and hyperopic surprises overall (4/8 subset vs. r2/63 depending upon formulae; all Pr0.001). Conclusions: Low posttrabeculectomy IOP (r9 mm Hg) is a risk factor for significant myopic surprise when undergoing subsequent cataract surgery despite using laser interferometry to measure ocular biometry and later generation formulae to determine IOL power. In addition, an IOP spike was associated with a 50% risk for large refractive surprise in this low IOP group. Key Words: phacoemulsification,, biometry, refractive error, cataract surgery, trabeculectomy, intraocular lens calculation, IOL Master (J Glaucoma 2017;26:65 70) Received for publication May 28, 2016; accepted August 22, From the Department of Ophthalmology, University of Washington, Seattle, WA. Supported in part by a Departmental grant from Research to Prevent Blindness Inc., New York, NY. Disclosure: The authors declare no conflict of interest.. Reprints: Philip P. Chen, MD, Department of Ophthalmology, University of Washington, P.O. Box , 325 9th Avenue, Seattle 98104, WA ( pchen@u.washington.edu). DOI: /IJG Trabeculectomy is an effective method for lowering intraocular pressure (IOP) for the treatment of. 1 With time, phakic patients may require visual rehabilitation through cataract surgery. However, the refractive outcome in phacoemulsification cataract surgery following trabeculectomy may not be as predictable as expected. In earlier studies using contact A-scan ultrasound technology, trabeculectomy was found to result in an average axial length (AL) decrease of 0.46 mm at various timepoints 2 and 0.91 mm at 12-month follow-up. 3 A later study using IOL Master laser interferometry (Carl Zeiss AGM, Oberkochen, Germany) demonstrated an average AL decrease of 0.32 mm at 1 month and 0.08 mm at 3 months after trabeculectomy, with the magnitude of AL reduction correlating linearly with IOP reduction and final IOP. 4 Another paper using IOL Master found average AL decreases in phacoemulsification cataract surgery alone of 75 mm and in combined phacotrabeculectomy of 117 mm at an average of 7.3 months 5 ; AL reduction demonstrated a linear correlation with final IOP but no correlation with reduction in IOP. These AL changes, along with changes in corneal curvature and shallowing of anterior chamber depth, are thought to predispose patients with phacoemulsification after prior trabeculectomy to greater myopic refractive error. Prior literature has suggested a trend toward myopic shift in phacoemulsification after trabeculectomy, 6 8 and this was found to be statistically significant in 2 retrospective studies. 7,8 However, these prior studies used various biometry measurement techniques and intraocular lens (IOL) calculation formulae which may be less relevant to current practice, including performance of ocular biometry by contact A-scan ultrasonography 6,8 or unspecified methods 7 and calculations using early-generation IOL formulae such as the SRK/SRK II 7,8 or other unspecified IOL prediction formulae. 6 We sought to evaluate postoperative refraction in patients undergoing phacoemulsification after trabeculectomy by retrospectively calculating predicted outcomes in eyes which had ocular biometry by laser interferometry, using Haigis, Holladay 2, SRK-T, and Hoffer Q formulae for all subjects, with the objective of improving IOL selection. An additional objective was to determine risk factors for refractive surprises in such eyes. METHODS The Human Subjects Division of the University of Washington gave approval for this retrospective casecontrolled study. For our study group, we reviewed the records of 36 eyes (34 patients) with previously J Glaucoma Volume 26, Number 1, January

2 Yeh et al J Glaucoma Volume 26, Number 1, January 2017 TABLE 1. Baseline Characteristics of Posttrabeculectomy (Study) and Medically Controlled Glaucoma (Control) Groups Study Group (N = 23) Control Group (N = 63) 2-Tailed Independent t Test (Except When Otherwise Indicated) (P) Age at phaco (y) ± ± Male/female 10/13 31/ (Pearson w 2 ) Ethnicity [n (%)] 13 (56.5) white 45 (71.4) white (Pearson w 2 ) 7 (30.4) black 4 (6.3) black 0 (0) Hispanic 6 (9.5) Hispanic 3 (13.0) Asian 6 (9.5) Asian 0 (0) other/unknown 2 (3.2) other/unknown Glaucoma type [n (%)] 10 (43.5) primary open 36 (57.1) primary open (Pearson w 2 ) angle angle 2 (8.7) normotension 14 (22.2) Normotension 5 (21.7) pseudoexfoliative 5 (7.9) pseudoexfoliative 1 (4.4) pigmentary 0 (0) pigmentary 5 (21.7) chronic angle closure 8 (12.7) chronic angle closure Trabeculectomy to phaco time (wk) 76.8 ± 66.4 No. preoperative 0.04 ± ± 1.12 < medications Preoperative IOP, average of ± ± 4.4 < (mm Hg) Preoperative LogMAR 0.46 ± ± Pre-cataract extraction axial length ± ± (mm) Pre-cataract extraction anterior chamber depth (mm) 2.96 ± ± Groups were matched by age, axial length, date of surgery, and type. IOP indicates intraocular pressure. Values in bold are statistically significant P < treated by trabeculectomy and subsequently underwent phacoemulsification by 1 surgeon (P.P.C.) between August 2007 and July Eyes in the study group were excluded if they underwent other procedures at the time of phacoemulsification. For the control group, we reviewed the records of eyes with but without prior intraocular surgical intervention, which underwent phacoemulsification cataract extraction by 1 surgeon (P.P.C.) between October 2007 and August We matched control and study eyes for age, AL, date of surgery, and type. We excluded eyes with other ocular diseases affecting final visual acuity measurement or with AL > 3 SDs from the mean of a large populationbased study. 9 Experienced ophthalmic technicians performed IOL Master laser interferometry (Carl Zeiss AGM, Oberkochen, Germany) and all manifest refractions on all eyes. Pertinent data were collected, including IOP using Goldmann applanation tonometry; ophthalmic biometric measurements of keratometry, anterior chamber depth, and AL using IOL Master; and spherical-equivalent manifest refraction for preoperative and postoperative visits. Manifest refraction was checked at 1 month or later after surgery and subsequently at the request of the patient or physician (P.P.C.); the latest postoperative refraction was chosen for analysis. Postoperative visits were scheduled at the discretion of the treating physician (P.P.C.). The baseline IOP was defined as the mean of the 2 preoperative visits before phacoemulsification. Patients did not have IOP measurement on the day of IOL Master measurement. To examine the effects of postoperative IOP elevation in the study group, we defined an IOP spike as an increase of >50% from baseline IOP within the first postoperative month of cataract surgery. 10 Predicted refractive outcomes were calculated using the recorded IOL Master measurements of corneal curvature, anterior chamber depth, and AL for Haigis using published formula and for Holladay 2, SRK-T, and Hoffer Q using Holladay IOL Consultant, version (Holladay Consulting, Bellaire, TX). In addition, preoperative manifest refraction and white-to-white measurements were used for the Holladay 2 calculation. Published optimized IOL A- constants were used. Statistical analysis was performed using SPSS, version 21 (IBM, Somers, NY). The main outcome measure was the difference between the expected and actual postoperative refraction. A difference of >1.0 D was considered a refractive (myopic or hyperopic) surprise. RESULTS We included 23 study eyes and 63 control eyes. All study eyes had trabeculectomy >3 months before cataract surgery. The indication for cataract surgery in all patients was visually significant cataract; in no study eyes was the indication hypotony, or low IOP. 11 Patient characteristics are summarized in Table 1. The mean follow-up time overall was 12.2 ± 4.1 months (study group, 11.6 ± 5.3 mo; control group, 12.5 ± 3.7 mo; P = 0.490, 2-tail t test). The mean time to final manifest refraction was 6.0 ±

3 J Glaucoma Volume 26, Number 1, January 2017 Cataract Surgery in Eyes With Prior Trabeculectomy FIGURE 1. Most recent postphacoemulsification refraction versus predicted for posttrabeculectomy study group (A) and medically controlled group (B) using Holladay 2, SRK-T, Hoffer Q, and Haigis formulae. Posttrabeculectomy group had significantly more eyes with >1.0 D of myopic postphacoemulsification refractive surprise than the medically controlled group when using Holladay 2, SRK-T, and Hoffer Q formula; however, the Haigis formula resulted in significantly more hyperopic errors of >1.0 D in the study group. Figure 1 can viewed in color online at months (study group, 4.1 ± 4.8 mo; control group, 6.4 ± 7.6 mo; P = 0.092). There was a significant difference in patient ethnicity (P = 0.023, Pearson w 2 ), but not for age at phacoemulsification or AL. The study group had a mean preoperative IOP of 9.4 ± 3.6 mm Hg (vs. control group 15.9 ± 4.4 mm Hg, P < 0.001, 2-tail t test) on 0.04 ± 0.21 (vs ± 1.12, P < 0.001) medications; at 12-month or latest 67

4 Yeh et al J Glaucoma Volume 26, Number 1, January 2017 TABLE 2. Trabeculectomy Group With Tr9 mm Hg Preoperatively and Intraocular Pressure Spike of >50% From Baseline Within Postoperative Month 1 (n = 8), Versus Control Group of Medically Controlled Glaucoma Patients Without Trabeculectomy (n = 63) Haigis Holladay 2 Hoffer Q SRK-T > 1.0 D myopia 3/8 vs. 1/63 P = /8 vs. 1/63 P = /8 vs. 0/63 P = /8 vs. 1/63 P = > 1.0 D hyperopia 1/8 vs. 0/63 1/8 vs. 1/63 1/8 vs. 0/63 1/8 vs. 1/63 P = > 1.0 D myopia or hyperopia 4/8 vs. 1/63 P < All P-values are obtained using Fisher exact test. Values in bold are statistically significant P < P = /8 vs. 2/63 P = P = /8 vs. 0/63 P < P = /8 vs. 2/63 P = postoperative follow-up, IOP was 12.0 ± 5.9 mm Hg (vs ± 3.3 mm Hg, P = 0.102) on 0.43 ± (vs ± 1.07, P < 0.001) medications. The mean visual acuity was LogMAR 0.465, Snellen equivalent 20/58 preoperatively (vs. control group LogMAR 0.192, Snellen equivalent 20/31, P = 0.063), and LogMAR 0.158, Snellen equivalent 20/29 (vs. LogMAR 0.045, Snellen equivalent 20/22, P = 0.042) at latest follow-up. The study group of posttrabeculectomy eyes had significantly more eyes with >1.0 D of myopic postphacoemulsification refractive surprise than the control group of medically controlled eyes when using Holladay 2, SRK-T, and Hoffer Q formula; however, the Haigis formula resulted in significantly more hyperopic errors of >1.0 D in the study group (Fig. 1). Comparison of study group eyes with preoperative IOPr9 mm Hg (n = 13) with the control group demonstrated significantly more large myopic errors of >1.0 D regardless of IOL prediction formula used. All IOL formulae performed similarly in the number of errors. In contrast, comparison of study group eyes with IOP > 9 mm Hg (n = 10) with control eyes showed no significant difference in large refractive errors from predicted for any of the formulae, with the exception of the Haigis, which had more large hyperopic errors in the study group than in the control. There was no significant difference in preoperative average corneal curvature (44.1 ± 1.3 vs ± 2.9 D, P = 0.459, 2-tail t test), anterior chamber depth (3.07 ± 0.58 vs ± 0.47 mm, P = 0.261), or AL (23.9 ± 1.0 vs ± 1.4 mm, P = 0.899) for the IOPr9 and IOP > 9 mm Hg posttrabeculectomy subgroups. We defined an IOP spike as a >50% increase from baseline IOP 10,12 within the first postoperative month of cataract surgery, and found it to be associated with myopic surprise for all 4 formulae (data not shown; Pr0.007). We found a strong association of the presence of an IOP spike in the study group with IOPr9 mm Hg with an increased frequency of myopic surprise measured at 1 month postoperatively or later regardless of choice of IOL prediction formula (Table 2). In the study group, 8/23 eyes had IOPr9 mm Hg and an IOP spike at any time within the first postoperative month. This group (n = 8) had a mean IOP on day of spike of 19.9 ± 7.8 mm Hg, versus mean preoperative IOP of 6.4 ± 1.6 mm Hg. At 1 month postoperative, this group had IOP of 14.4 ± 8.1 mm Hg on 0.13 ± 0.35 medications versus 11.7 ± 4.0 mm Hg (P = 0.293) on 0.27 ± 0.46 medications (P = 0.456) for the other 15/23 eyes. Final IOP was measured to be 12.6 ± 8.6 mm Hg on 0.38 ± 0.52 medications versus 11.7 ± 4.4 mm Hg on 0.47 ± medications (P = 0.760). Of the 8 eyes in this group, 3 eyes had large myopic surprises; additionally, 4 eyes had sustained IOP rise at the end of month 1. Two of the 4 eyes with sustained IOP rise had a large myopic surprise, whereas 1 of the 4 eyes with temporary IOP rise had a large myopic surprise. Four of the 23 study eyes (1/8 in the preoperative IOPr9 mm Hg subgroup with IOP spike vs. 3/15 of the remainder, P = 1.00 Fisher exact test), were treated with an additional medication in the first postoperative month, and in all cases only a single topical medication was added. No eyes had paracentesis nor temporary reopening of the wound to treat an IOP spike. We repeated the analysis using an alternate definition for IOP spike of an increase of 5 mm Hg, and found the same 7/8 patients with preoperative IOPr9 mm Hg had IOP spikes of Z5 mm Hg. The 1 patient who did not TABLE 3. Average Diopters of Refractive Surprise From Predicted in Posttrabeculectomy Groups Stratified by Intraocular Pressure (IOP) Versus Medically Controlled Glaucoma Eyes Haigis Holladay 2 Hoffer Q SRK-T Study group with preoperative IOPr9mm Hg 0.26 ± ± ± ± 1.07 (n = 13) (D) P = P = P = P = Study group with preoperative IOP > 9 mm Hg 0.55 ± ± ± ± 0.64 (n = 10) (D) P = P = P = P = Medically controlled (n = 63) (D) 0.11 ± ± ± ± 0.52 Correlation between change in IOP and predicted r = r = r = refraction, study group (n = 23) P = P = P = r = P = (Pearson correlation) No overall trend in refractive surprise found in the posttrabeculectomy study group when analyzing either the preoperative IOPr9mm Hg or the IOP > 9 mm Hg subsets versus the medically controlled group. All P-values are obtained using Fisher exact except when indicated. Values in bold are statistically significant P <

5 J Glaucoma Volume 26, Number 1, January 2017 Cataract Surgery in Eyes With Prior Trabeculectomy requalify under the 5 mm Hg spike criteria had an IOP of 8.5 preoperative and Tmax of 13 within month 1; he did not have a refractive surprise. This finding did not appreciably change the results, and in fact using any of the 4 IOL formula resulted in 3/7 large myopic and 1/7 hyperopic errors with all P < Despite the increased frequency of myopic surprises in the posttrabeculectomy group with IOPr9mm Hg versus controls, there was no overall shift toward myopic surprise, regardless of IOL formula. There was also no difference in mean refractive surprise for the IOP > 9 mm Hg study group (Table 3). Among all study eyes (n = 23) we found no correlation between IOP change after phacoemulsification and predicted refractive error for 3 of 4 formulae; only the Haigis formula was weakly correlated (r = 0.417, P =0.048). There were 3 eyes in the study group with preoperative hypotony, defined as an IOP < 6 mm Hg. No significant differences were found in refractive surprises >1.0 D except for more hyperopia when using the Haigis formula (1/3 in the study group, 0/63 in the control; P = 0.045, Fisher exact test). We found no association of large refractive errors with age, type, posttrabeculectomy AL, preoperative use of medication, or intraoperative factors such as posterior synechiolysis or pupil stretching. DISCUSSION Phacoemulsification after trabeculectomy improves vision in patients with. However, we found that cataract surgery in an eye with preoperative IOPr9mm Hg after trabeculectomy is at greater risk for unexpected large myopic refractive surprises despite use of modern biometry and IOL prediction formulae. In addition, we found that a postoperative IOP spike of >50% above the baseline IOP may contribute to a myopic surprise in such eyes; in our study, 37.5% of these preoperatively low IOP eyes with subsequent IOP spike had a myopic surprise of over 1 D. These findings suggest that in an eye with a lower IOP, a period of higher IOP after phacoemulsification could stretch the eye to a longer AL, inducing a myopic shift. 6,8 Notably, the preoperative average keratometry, anterior chamber depth, and AL was similar for the IOPr9 and >9 mm Hg posttrabeculectomy groups. Other authors have found that eyes having phacoemulsification following trabeculectomy had significantly greater myopic surprises versus control eyes, using older biometric methods and IOL formulae. 6,8 Zhang et al 8 using contact A-scan ultrasonography to determine ocular biometry, and the SRK II/T formula, noted that the magnitude of myopic surprise correlated with the change between preoperative and postoperative IOP; we found this to be true only with the Haigis formula. An earlier retrospective study 6 also using contact A-scan ultrasonography found no difference in overall refractive outcomes for posttrabeculectomy and control eyes, but observed more large myopic surprises in a posttrabeculectomy group with an IOPr5 mm Hg. Contact A-scan biometry may induce bias by probe deformation of the softer IOP eye, thereby resulting in a falsely shorter AL measurement and greater myopic refractive surprise. Interestingly, we had similar results using exclusively laser interferometry, despite the known advantages of noncontact biometry. 13 In our study, we found a significantly higher frequency of myopic surprises (> 1.0 D) for posttrabeculectomy eyes with preoperative IOPr9 mm Hg, yet even within this subset of eyes, no myopic shift was seen on an average. There was neither an increase in frequency of myopic surprises nor a myopic shift for the IOP > 9 mm Hg group. These findings were consistent across all 4 formulae analyzed. However, the occurrence of an IOP spike, defined as a >50% IOP rise from baseline within the first postoperative month, demonstrated a strong association with myopic surprises in the low preoperative IOP study group. Given the small sample size of eyes with sustained IOP rise, we will not extrapolate more broadly upon the effect of duration of IOP spike on myopic surprise. Future prospective studies could explore whether the administration of medications to minimize IOP spikes would reduce the risk of these myopic surprises. We acknowledge that specific formulae have been considered to be more appropriate for some eyes than others, depending on AL. 14 However, in this study we sought to evaluate refractive outcomes by using predicted refractions for all 4 formulae for every eye, regardless of AL, and then calculating actual versus expected refractive outcomes. In fact our results show no significant difference between the formulae in accuracy despite the range of ALs. Our study has clear limitations. The number of patients studied is small. The retrospective nature of our study might have resulted in errors related to inaccuracy and incompleteness of data recording, and variability of postoperative regimens and follow-up. Postoperative IOL Master biometry was not obtained in our patients, and therefore the Haigis formula was not optimized for the individual surgeon. However, unlike prior retrospective studies that used ultrasonography to determine ocular dimensions and reported solely on the surgeon s choice (sometimes unspecified) of IOL calculation formulae, our study used IOL Master biometric data to recalculate Haigis, Holladay 2, SRK-T, and Hoffer Q prediction outcomes. We also used the average of 2 preoperative IOP measurements as a baseline, which was not done in prior studies 6 8 and may reduce errors resulting from regression to the mean in postoperative IOP measurements. We advise that cataract extraction after trabeculectomy in an eye with a low IOP may result in a refractive outcome that is significantly more myopic than expected compared with that of cataract extraction in medically controlled even when using laser interferometry for ocular biometry and later generation IOL prediction formulae. We speculate that stricter control of IOP postoperatively to minimize the chance of an IOP spike may help prevent such refractive surprises and may be more predictable than attempting to modify IOL power in anticipation of possible postoperative AL elongation. REFERENCES 1. Landers J, Martin K, Sarkies N, et al. A twenty-year follow-up study of trabeculectomy: risk factors and outcomes. Ophthalmology. 2012;119: Cashwell LF, Martin CA. Axial length decrease accompanying successful filtration surgery. Ophthalmology. 1999;106: Kook MS, Kim HB, Lee SU. Short-term effect of mitomycin-c augmented trabeculectomy on axial length and corneal astigmatism. J Cataract Refract Surg. 2000;27:

6 Yeh et al J Glaucoma Volume 26, Number 1, January Francis BA, Wang M, Lei H, et al. Changes in axial length following trabeculectomy and drainage device surgery. Br J Ophthalmol. 2005;89: Law SK, Mansury AM, Vasudev D, et al. Effects of combined cataract surgery and trabeculectomy with mitomycin C on ocular dimensions. Br J Ophthalmol. 2005;89: Muallem JS, Nelson GA, Osmanovic S, et al. Predicted refraction versus refraction outcome in cataract surgery following trabeculectomy. J Glaucoma. 2009;18: Tan H, Wu S. Refractive error with optimum intraocular lens power calculation after filtering surgery. J Cataract Refract Surg. 2004;30: Zhang N, Tsai PL, Catoira-Boyle YP, et al. The effect of prior trabeculectomy on refractive outcomes of cataract surgery. Am J Ophthalmol. 2013;155: Lee KE, Klein BE, Klein R, et al. Association of age, stature, and education with ocular dimensions in an older white population. Arch Ophthalmol. 2009;127: Slabaugh MA, Bojikian KD, Moore DB, et al. The effect of phacoemulsification on intraocular pressure in medically controlled open-angle patients. Am J Ophthalmol. 2014;157: Doyle JW, Smith MF. Effect of phacoemulsification surgery on hypotony following trabeculectomy surgery. Arch Ophthalmol. 2000;118: Chen PP, Musch DC, Niziol LM, et al. The effect of early posttrabeculectomy intraocular pressure spike in the Collaborative Initial Glaucoma Treatment Study. J Glaucoma. 2011;20: Findl O, Drexler W, Menapace R, et al. Improved prediction of intraocular lens power using partial coherence interferometry. J Cataract Refract Surg. 2001;27: Patel AS, O Brien C, Shahzed HSF. Biometry for intra-ocular lens (IOL) power calculation, choice of formulas Available at: Ocular_Lens_(IOL)_power_calculation#Choice_of_formulas. Accessed August 7,