CLINICAL SCIENCE
Effect of Glaucoma Tube Shunt Parameters on Cornea Endothelial Cells in Patients With Ahmed Valve Implants Euna B. Koo, MD,* Jing Hou, MD,*† Ying Han, MD, PhD,* Jeremy D. Keenan, MD,*‡ Robert L. Stamper, MD,* and Bennie H. Jeng, MD§
Purpose: The aim of this study was to assess the effect of various tube parameters on corneal endothelial cell density (ECD) after insertion of Ahmed valves.
Methods: Thirty-nine eyes of 33 patients with previous superotemporal (ST) Ahmed valve implantation and 20 eyes of 13 participants with previous uncomplicated phacoemulsification and intraocular lens implantation but no history of glaucoma surgery were evaluated. Various tube parameters were measured with anterior segment optical coherence tomography. ST, central, and inferonasal (IN) ECD and pachymetry were measured. Endothelial cell loss and corneal thickness in the ST cornea was compared with those in the IN cornea.
Results: The mean age of the operated patients was 58 6 22 years, and the mean time since glaucoma surgery was 2.5 6 2.6 years. Thirty-two of the 39 study eyes were pseudophakic. The ECD was significantly lower in the ST endothelium than in the IN endothelium in eyes with glaucoma tube surgery (P , 0.001), although this relative reduction in ST ECD was not greater than that seen in pseudophakic control eyes (P = 0.16). In univariate analysis, tube angle relative to the cornea and distance from the tip of the tube to the cornea were significant risk factors for decreased ST endothelial cell loss when assessed relative to the IN ECD (P = 0.01 and P = 0.02, respectively). In multivariate analysis, only the distance of the tube tip to the cornea remained significantly associated with ST endothelial cell loss. Received for publication August 11, 2014; revision received September 23, 2014; accepted September 24, 2014. Published online ahead of print November 12, 2014. From the *Department of Ophthalmology, University of California San Francisco, San Francisco, CA; †Department of Ophthalmology, People’s Hospital, Peking University, Beijing, China; ‡Francis I. Proctor Foundation, University of California San Francisco, San Francisco, CA; and §Department of Ophthalmology and Visual Sciences, University of Maryland School of Medicine, Baltimore, MD. Supported in part by a grant from Research to Prevent Blindness, Inc, and That Man May See, Inc. Presented in part at the Association for Research in Vision and Ophthalmology Annual Meeting, Seattle, WA, May 6, 2013; Cornea Society Fall educational symposium, New Orleans, LA, November 15, 2013; and American Academy of Ophthalmology Meeting, New Orleans, LA, November 18–19, 2013. The authors have no conflicts of interest to disclose. E. B. Koo and J. Hou contributed equally as primary coauthors. Reprints: Bennie H. Jeng, MD, Department of Ophthalmology and Visual Sciences, University of Maryland School of Medicine, 419 W. Redwood St, Suite 470, Baltimore, MD 21201 (e-mail: bjeng@som. umaryland.edu). Copyright © 2014 by Lippincott Williams & Wilkins
Cornea Volume 34, Number 1, January 2015
Conclusions: Although this was a retrospective study with inherent limitations, tubes that are closer to the cornea seem to lead to increased loss of adjacent endothelial cells. Key Words: glaucoma tube shunts, endothelial cell densities, anterior segment OCT, specular microscopy (Cornea 2015;34:37–41)
C
orneal endothelial cell loss is a known sequela of glaucoma tube shunt implantation.1–3 Risk factors for postoperative decreased corneal endothelial cell density (ECD) include inflammation, peripheral anterior synechiae, and intermittent tube contact with uveal tissues or with the cornea.4–6 Although tube positioning in the anterior chamber has not yet been found to be associated with decreased ECD after tube shunt surgery, there are reasons to think that it could be important. For example, the corneal endothelium nearest to the tube shows the greatest decrease in ECD after tube placement.7 Corneal graft rejection may be less common in eyes receiving pars plana-based tubes compared with limbal-based tubes, suggesting that tube positioning could play a role in the immune response after corneal transplantation.8 Few studies have assessed the importance of tube positioning on endothelial cells after tube shunt surgery.4 Therefore, we performed anterior segment optical coherence tomography (OCT) and specular microscopy in a series of patients who had undergone tube shunt surgery to assess the effect of various tube parameters on corneal endothelial cell loss.
MATERIALS AND METHODS Ethical approval was obtained from the University of California San Francisco (UCSF) Committee on Human Research. Patients seen at the UCSF Glaucoma Service from 2012 to 2013 who had undergone superotemporal (ST) glaucoma tube implantation surgery at any time in the past and were able to undergo multiple imaging modalities were consented and included in the study. Controls were recruited from other services at the UCSF eye clinic. Those who had previously undergone uncomplicated cataract surgery but no other intraocular surgeries were recruited. Endothelial cell counts and pachymetry were obtained in these controls. The research followed the tenets of the Declaration of Helsinki, and informed consent was obtained from all patients. Patients with preexisting corneal pathology, previous corneal grafts, and inability to complete necessary imaging tests were excluded. www.corneajrnl.com |
37
Koo et al
ST, central, and inferonasal (IN) ECD and corneal thickness were measured by noncontact specular microscopy (CellChek XL Specular Microscope; Konan Medical Inc, Irvine, CA) and ultrasound pachymetry (DGH Pachette 2; DGH Technology, Exton, PA), respectively. Patients were instructed to fixate on the fixation light that was repositioned according to which quadrant was being imaged. The semiautomatic program in CellChek XL necessitating manual identification of individual cells was used to count endothelial cells after images were acquired. The intraclass correlation coefficient for intrarater variability was 0.995, [95% confidence interval (CI), 0.993–0.998]. The tube position, including the angle of the tube relative to the cornea, tube length, and distance between the tip of the tube and the cornea, were measured with anterior segment OCT (Visante OCT Anterior Segment Imaging, Dublin, CA). The best quality image was obtained by similar methods described by Hau et al.4 The scanning axis was positioned along the tube so that the position of the tube, including the tip, in relation to the other structures of the anterior chamber could be clearly visualized (Fig. 1).
Cornea Volume 34, Number 1, January 2015
The distance between the tip of the tube and the cornea, perpendicular to the cornea, was measured. The length of the tube was measured from the point of insertion into the anterior chamber to the tip of the tube. The same investigator (J.H.) obtained all images and measurements of the anterior segment.
Statistical Analysis We calculated the mean difference in (1) ECD and (2) pachymetry between pairs of the 3 measured locations (central, ST, and IN) and tested for significant differences between the locations with a paired test. We created linear regression models adjusted for age and sex to test for differences between eyes that had undergone tube shunt surgery versus control eyes that had undergone cataract surgery. We tested for the relationship between ECD and pachymetry with linear regression. We assumed that glaucoma tube surgery has a greater local effect than a global effect on the corneal endothelial cells and that the ST cornea would be most affected by the tube. Therefore, we created a difference score between the ST and
FIGURE 1. Anterior segment OCT measurements of tube parameters. The top figure represents glaucoma tube in the anterior chamber, and the bottom figure represents the glaucoma tube in the sulcus entering the anterior chamber through the iris.
38
| www.corneajrnl.com
Ó 2014 Lippincott Williams & Wilkins
Cornea Volume 34, Number 1, January 2015
Effect of Glaucoma Tube Shunt Parameters
IN corneas, with the assumption that this represents the best estimate of the change in corneal parameters due to the tube surgery. We tested whether this difference score was associated with several tube parameters in univariate and multivariate linear regression models, using a backward stepwise selection procedure to select the best multivariate model. For all analyses, we used bootstrap resampling at the patient level (9999 repetitions) to account for nonindependence of eyes from the same patient. All analyses were performed with Stata 12 (StataCorp, College Station, TX).
RESULTS Measurements were obtained on 39 eyes that had undergone Ahmed valve placement and on 20 control eyes that had undergone cataract extraction with temporal incisions (Table 1). Of the 33 patients in the post-Ahmed group, the mean age was 58 6 22 years and 17 (51.5%) were female. Of the 13 patients in the control group, the mean age was 68.5 6 8.7 years and 8 (61.5%) were female. The average time since cataract surgery in the control group was 5.7 6 6.1 years. The average time since glaucoma surgery was 2.5 6 2.6 years. Of the eyes in the post-Ahmed group, 1 (2.6%) had congenital glaucoma, 2 (5.1%) had birdshot retinochoroidopathy, 2 (5.1%) had Vogt–Koyanagi–Harada disease and a fluocinolone acetonide intravitreal implant, 6 (15.4%) had previously undergone trabeculectomy, and 32 (82%) were pseudophakic. The average time between cataract surgery and Ahmed tube implantation was 13 6 14.8 months in those who had glaucoma surgery first, and the average time between cataract surgery and Ahmed tube implantation was 82.4 6 114.1 months in those who had cataract surgery first. Ten patients had cataract surgery and glaucoma surgery simultaneously, and 7 never underwent cataract surgery. All of the eyes in the control group had undergone uncomplicated phacoemulsification with intraocular lens implantation but no glaucoma surgery. Phacoemulsification was performed through temporal wounds in both study and control eyes. TABLE 1. Characteristics of Study Eyes Posttube Eyes (N = 39), Mean 6 SD
Control Eyes (N = 20), Mean 6 SD
P
Age of patient, yrs 58 6 22 68.5 6 8.7 0.01 Intraocular pressure, mm Hg* 13.6 6 4.8 16.9 6 2.4 0.002 Time since surgery, yrs 2.5 6 2.6 5.7 6 6.1 — ECD, cells/mm2 Superotemporal 1611.9 6 785.8 2211.3 6 422.5 ,0.002 Central 1902.2 6 841.4 2335.1 6 393.4 0.06 Inferotemporal 1938.4 6 823.5 2382.4 6 368 0.01 Pachymetry, mm ST 654.1 6 95.5 675 6 74.5 0.15 Central 541.2 6 76.5 562.2 6 57.9 0.21 IN 679.1 6 90.3 673.6 6 54.5 0.99 Tube–cornea angle, degrees 29 6 10.5 — — Tube tip–cornea distance, mm 1.1 6 0.6 — — Tube length, mm 3.4 6 0.8 — — *Assessed with pneumotonometry
Ó 2014 Lippincott Williams & Wilkins
In patients with glaucoma tubes, ECD was significantly lower in the ST endothelium compared with either the central or IN endothelium (P , 0.001; Table 2). The pseudophakic control eyes also had ECDs that were lower in the ST endothelium compared with either the central or IN endothelium, although the differences were less marked (Table 2). After adjusting for age and sex, relative differences between the 3 locations did not differ between the posttube and postcataract groups. Figure 2 shows the relationship between ECD and corneal thickness in the ST corneas of the posttube group. ST corneal thickness seemed to be inversely associated with ST ECD, but only at the lower range of ECD values. For example, in eyes with ST ECD , 1000 cells per square millimeter, corneal thickness increased 11 mm (95% CI, 84.5 mm increase to 62.5 mm decrease) for each 100-unit reduction in ECD— although the number of eyes in this stratum was low and the relationship did not achieve statistical significance (P = 0.77, linear regression). In contrast, there seemed to be no association between ST ECD and corneal thickness at the higher range of ECD values (Fig. 2; P = 0.29 in linear regression restricted to eyes with ECD $ 1000 cells/mm2). Various parameters of the tube including tube distance from the cornea, angle of the tube, and tube length obtained by anterior segment OCT are listed in Table 3. In univariate analyses, tube–cornea angle and the closest distance from the tip of the tube to the cornea were statistically significant predictors of ST corneal endothelial cell loss after glaucoma surgery, expressed as the difference between the ST and IN ECD (Table 3). In multivariate analysis, only the distance from the tip of the tube to the cornea remained significant; each millimeter that the tube was closer to the endothelial surface was associated with 353.1 (95% CI, 56.1–650.1; P = 0.02) fewer endothelial cells superotemporally.
DISCUSSION In this study, we used anterior segment OCT and specular microscopy to assess the role of tube positioning on corneal endothelial cell loss after placement of glaucoma drainage devices. We estimated endothelial cell loss as the difference in ECD between the ST and IN corneas, and found that the distance from the tip of the tube was significantly associated with this metric of decreased ECD. Our study confirms that endothelial cells may be better preserved with tubes placed further away from the cornea. Previous studies have reported loss of endothelial cells in patients with glaucoma drainage devices. A 2-year prospective study evaluating corneal endothelial cell counts in various locations of the cornea before and after Ahmed valve placement in the ST portion of the anterior chamber showed that, compared with glaucomatous control eyes with no glaucoma surgery, study eyes had significantly decreased endothelial cell counts superotemporally at all time points after 6 months of follow-up.7 That study reported that the reduction in ST ECD from baseline to 2 years after surgery was 22.6%, which represented a 26.3% lower ECD in comparison with nonoperated control eyes. Our study is consistent with this because we found that the ST ECD in posttube eyes was approximately 27% lower than that of pseudophakic control eyes (1612 vs. 2211 cells/mm2). www.corneajrnl.com |
39
Cornea Volume 34, Number 1, January 2015
Koo et al
TABLE 2. ECD and Pachymetry Measurements Taken at the ST, Central, and IN cornea Posttube (N = 39) Difference in ECD ST minus central ST minus IN Central minus IN Difference in pachymetry ST minus central ST minus IN Central minus IN
Control (N = 20)
Mean Difference (95% CI)
P*
Mean Difference (95% CI)
P*
P†
2290.3 (2407.2 to 2173.4) 2326.5 (2478.8 to 2174.3) 236.3 (2125.5 to 53)
,0.001 ,0.001 0.43
2123.8 (2255.1 to 7.6) 2171.1 (2320.1 to 222) 247.3 (2104 to 9.4)
0.07 0.03 0.10
0.31 0.46 0.92
112.8 (95.2 to 130.5) 225.1 (245.4 to 24.7) 2137.9 (2155.2 to 2120.7)
,0.001 0.02 ,0.001
112.8 (89.9 to 135.7) 1.4 (218.2 to 21) 2111.4 (2122.5 to 2100.3)
,0.001 0.89 ,0.001
0.96 0.22 0.10
*Paired t test comparing values at pairs of corneal locations. †Age- and sex-adjusted linear regression.
One mechanism that has been proposed to explain endothelial cell loss in patients with glaucoma drainage devices is turbulence present at the tip of the implant.9 Our finding of decreased ECD near the tube supports this hypothesis. All tubes in our study had a bevel-up tip potentially directing turbulent flow toward endothelial cells anterior to the tip. In this model, the ST cornea would be most affected by jet flow or turbulence compared with other areas of the cornea. In our study, there was a much larger difference in ST and IN ECD in eyes with a tube (2326.5 6 493.1 cells/mm2) compared with the pseudophakic eyes (2320.1 to 222 cells/ mm2) (Table 2). This suggests that the tube likely did cause a local reduction in ECD, although we could not demonstrate a statistically significant difference between the posttube and control groups after adjusting for age and sex. In a cross-sectional study, disruption of endothelial cells by physical contact of the tube to the cornea at its insertion point was evaluated as another possible explanation
for endothelial cell loss.10 Microscopic fibrotic tissue between the tube and corneal endothelium was detected using highresolution Fourier domain OCT to evaluate anterior chamber tube shunts.10 The interface between the tube and the cornea at the point of insertion was not assessed in our study. Interestingly, tube angle, which hypothetically could affect the extent of fibrosis between the tube and the cornea, was a significant risk factor for endothelial cell loss in univariate analysis in our study. However, the effect of the tube angle on the endothelium could be mediated by tube distance. In future studies, evaluating this interface with OCT by adjusting resolution settings at the time of image acquisition may be useful in further elucidating the effect of drainage devices and their parameters on adjacent tissues. Our study suggests that the distance from the tip of the tube to the cornea is a crucial risk factor for endothelial cell loss. This finding is consistent with a previous study that found that the distance from the tube to the posterior cornea was the most important of several tube parameters tested, although that study did not find a statistically significant association (P = 0.08).4 It is important to note that this previous study was conducted primarily in patients with uveitis and tested for differences in ECD over the tube site, but not relative to an area of a normal cornea as we did in this study. Moreover, the previous study measured the distance between the tip of the tube and the cornea, perpendicular to the tube, whereas we analyzed the distance between the tip of the tube
TABLE 3. Univariate Analyses of Tube Parameters to Predict Endothelial Cells Loss in ST Compared to IN Cornea
Covariate
FIGURE 2. Relationship between ST corneal thickness and ST ECD in patients who underwent tube shunt surgery. In eyes with ST ECD , 1000 cells per square millimeter, the corneal thickness increased 11 mm (95% CI, 84.5 mm increase to 62.5 mm decrease) for each 100-unit reduction in ECD.
40
| www.corneajrnl.com
Tube angle, per degree closer to the cornea Tube distance, per mm closer to the cornea* Tube length, per mm
Reduction in ECD (Cells/mm), ST Compared With IN Cornea (95% CI) 16.3 (3.6–29) 353.1 (56.1–650.1) 2 (133.4 to 2129.5)
P 0.01 0.02 0.98
*The only variable that remained significant after multivariate backward stepwise algorithm.
Ó 2014 Lippincott Williams & Wilkins
Cornea Volume 34, Number 1, January 2015
and the cornea, perpendicular to the cornea (Fig. 1). Without knowing exact physics of turbulence at the tip of the tube, it is difficult to know which distance is more consistent with actual events. Nonetheless, studies are generally consistent and do provide support for the importance of tube distance from the cornea as a significant predictor of endothelial cell loss. Limitations of our study include a small population, lack of standardization of time of follow-up after surgery, and lack of preoperative and longitudinal data. In our study, we did not include slit-lamp or gonioscopic findings. For example, we did not assess peripheral anterior synechiae, which have been reported to alter endothelium integrity.4 Also, we did not compare various types of tubes because all of our patients had Ahmed tubes implanted in the eye. However, a comparison done by Nassiri et al11 of the Ahmed valves to the single plate Molteno implants showed no difference in corneal ECD or corneal thickness at 2 years. High-resolution noninvasive imaging modalities have made studies like ours more feasible. Given the relatively recent availability of these modalities, longitudinal studies measuring progression of corneal endothelial cell loss after shunt placement are still few. The long-term clinical significance of endothelial cell loss after glaucoma tube implantation remains unclear. It is possible that the presence of a tube near the endothelium presents a constant insult to the endothelium, which would be quite different from the presumably 1-time insult of cataract surgery. Longitudinal studies that incorporate imaging of the tube would be helpful in determining clinical importance of minimizing the distance of the tube from the cornea. In the meantime, our study
Ó 2014 Lippincott Williams & Wilkins
Effect of Glaucoma Tube Shunt Parameters
reinforces the widely held view that tubes be placed as far from the cornea as safely possible.
REFERENCES 1. Minckler DS, Francis BA, Hodapp EA, et al. Aqueous shunts in glaucoma: a report by the American Academy of Ophthalmology. Ophthalmology. 2008;115:1089–1098. 2. Barton K, Heuer DK. Modern aqueous shunt implantation: future challenges. Prog Brain Res. 2008;173:263–276. 3. Chihara E, Kubota H, Takanashi T, et al. Outcome of White pump shunt surgery for neovascular glaucoma in Asians. Ophthalmic Surg. 1992;23: 666–671. 4. Hau S, Scott A, Bunce C, et al. Corneal endothelial morphology in eyes implanted with anterior chamber aqueous shunts. Cornea. 2011;30:50–55. 5. Kim CS, Yim JH, Lee EK, et al. Changes in corneal endothelial cell density and morphology after Ahmed glaucoma valve implantation during the first year of follow up. Clin Experiment Ophthalmol. 2008;36: 142–147. 6. Topouzis F, Coleman AL, Choplin N, et al. Follow-up of the original cohort with the Ahmed glaucoma valve implant. Am J Ophthalmol. 1999; 128:198–204. 7. Lee EK, Yun YJ, Lee JE, et al. Changes in corneal endothelial cells after Ahmed glaucoma valve implantation: 2-year follow-up. Am J Ophthalmol. 2009;148:361–367. 8. Ritterband DC, Shapiro D, Trubnik V, et al. Penetrating keratoplasty with pars plana glaucoma drainage devices. Cornea. 2007;26:1060–1066. 9. McDermott ML, Swendris RP, Shin DH, et al. Corneal endothelial cell counts after Molteno implantation. Am J Ophthalmol. 1993;115:93–96. 10. Jiang C, Li Y, Huang D, et al. Study of anterior chamber aqueous tube shunt by fourier-domain optical coherence tomography. J Ophthalmol. 2012;2012:189580. 11. Nassiri N, Majdi-N M, Salehi M, et al. Corneal endothelial cell changes after Ahmed valve and Molteno glaucoma implants. Ophthalmic Surg Lasers Imaging. 2011;42:394–399.
www.corneajrnl.com |
41
Effect of Glaucoma Tube Shunt Parameters on Cornea Endothelial Cells in Patients With Ahmed Valve Implants Euna B. Koo, MD,* Jing Hou, MD,*† Ying Han, MD, PhD,* Jeremy D. Keenan, MD,*‡ Robert L. Stamper, MD,* and Bennie H. Jeng, MD§
Purpose: The aim of this study was to assess the effect of various tube parameters on corneal endothelial cell density (ECD) after insertion of Ahmed valves.
Methods: Thirty-nine eyes of 33 patients with previous superotemporal (ST) Ahmed valve implantation and 20 eyes of 13 participants with previous uncomplicated phacoemulsification and intraocular lens implantation but no history of glaucoma surgery were evaluated. Various tube parameters were measured with anterior segment optical coherence tomography. ST, central, and inferonasal (IN) ECD and pachymetry were measured. Endothelial cell loss and corneal thickness in the ST cornea was compared with those in the IN cornea.
Results: The mean age of the operated patients was 58 6 22 years, and the mean time since glaucoma surgery was 2.5 6 2.6 years. Thirty-two of the 39 study eyes were pseudophakic. The ECD was significantly lower in the ST endothelium than in the IN endothelium in eyes with glaucoma tube surgery (P , 0.001), although this relative reduction in ST ECD was not greater than that seen in pseudophakic control eyes (P = 0.16). In univariate analysis, tube angle relative to the cornea and distance from the tip of the tube to the cornea were significant risk factors for decreased ST endothelial cell loss when assessed relative to the IN ECD (P = 0.01 and P = 0.02, respectively). In multivariate analysis, only the distance of the tube tip to the cornea remained significantly associated with ST endothelial cell loss. Received for publication August 11, 2014; revision received September 23, 2014; accepted September 24, 2014. Published online ahead of print November 12, 2014. From the *Department of Ophthalmology, University of California San Francisco, San Francisco, CA; †Department of Ophthalmology, People’s Hospital, Peking University, Beijing, China; ‡Francis I. Proctor Foundation, University of California San Francisco, San Francisco, CA; and §Department of Ophthalmology and Visual Sciences, University of Maryland School of Medicine, Baltimore, MD. Supported in part by a grant from Research to Prevent Blindness, Inc, and That Man May See, Inc. Presented in part at the Association for Research in Vision and Ophthalmology Annual Meeting, Seattle, WA, May 6, 2013; Cornea Society Fall educational symposium, New Orleans, LA, November 15, 2013; and American Academy of Ophthalmology Meeting, New Orleans, LA, November 18–19, 2013. The authors have no conflicts of interest to disclose. E. B. Koo and J. Hou contributed equally as primary coauthors. Reprints: Bennie H. Jeng, MD, Department of Ophthalmology and Visual Sciences, University of Maryland School of Medicine, 419 W. Redwood St, Suite 470, Baltimore, MD 21201 (e-mail: bjeng@som. umaryland.edu). Copyright © 2014 by Lippincott Williams & Wilkins
Cornea Volume 34, Number 1, January 2015
Conclusions: Although this was a retrospective study with inherent limitations, tubes that are closer to the cornea seem to lead to increased loss of adjacent endothelial cells. Key Words: glaucoma tube shunts, endothelial cell densities, anterior segment OCT, specular microscopy (Cornea 2015;34:37–41)
C
orneal endothelial cell loss is a known sequela of glaucoma tube shunt implantation.1–3 Risk factors for postoperative decreased corneal endothelial cell density (ECD) include inflammation, peripheral anterior synechiae, and intermittent tube contact with uveal tissues or with the cornea.4–6 Although tube positioning in the anterior chamber has not yet been found to be associated with decreased ECD after tube shunt surgery, there are reasons to think that it could be important. For example, the corneal endothelium nearest to the tube shows the greatest decrease in ECD after tube placement.7 Corneal graft rejection may be less common in eyes receiving pars plana-based tubes compared with limbal-based tubes, suggesting that tube positioning could play a role in the immune response after corneal transplantation.8 Few studies have assessed the importance of tube positioning on endothelial cells after tube shunt surgery.4 Therefore, we performed anterior segment optical coherence tomography (OCT) and specular microscopy in a series of patients who had undergone tube shunt surgery to assess the effect of various tube parameters on corneal endothelial cell loss.
MATERIALS AND METHODS Ethical approval was obtained from the University of California San Francisco (UCSF) Committee on Human Research. Patients seen at the UCSF Glaucoma Service from 2012 to 2013 who had undergone superotemporal (ST) glaucoma tube implantation surgery at any time in the past and were able to undergo multiple imaging modalities were consented and included in the study. Controls were recruited from other services at the UCSF eye clinic. Those who had previously undergone uncomplicated cataract surgery but no other intraocular surgeries were recruited. Endothelial cell counts and pachymetry were obtained in these controls. The research followed the tenets of the Declaration of Helsinki, and informed consent was obtained from all patients. Patients with preexisting corneal pathology, previous corneal grafts, and inability to complete necessary imaging tests were excluded. www.corneajrnl.com |
37
Koo et al
ST, central, and inferonasal (IN) ECD and corneal thickness were measured by noncontact specular microscopy (CellChek XL Specular Microscope; Konan Medical Inc, Irvine, CA) and ultrasound pachymetry (DGH Pachette 2; DGH Technology, Exton, PA), respectively. Patients were instructed to fixate on the fixation light that was repositioned according to which quadrant was being imaged. The semiautomatic program in CellChek XL necessitating manual identification of individual cells was used to count endothelial cells after images were acquired. The intraclass correlation coefficient for intrarater variability was 0.995, [95% confidence interval (CI), 0.993–0.998]. The tube position, including the angle of the tube relative to the cornea, tube length, and distance between the tip of the tube and the cornea, were measured with anterior segment OCT (Visante OCT Anterior Segment Imaging, Dublin, CA). The best quality image was obtained by similar methods described by Hau et al.4 The scanning axis was positioned along the tube so that the position of the tube, including the tip, in relation to the other structures of the anterior chamber could be clearly visualized (Fig. 1).
Cornea Volume 34, Number 1, January 2015
The distance between the tip of the tube and the cornea, perpendicular to the cornea, was measured. The length of the tube was measured from the point of insertion into the anterior chamber to the tip of the tube. The same investigator (J.H.) obtained all images and measurements of the anterior segment.
Statistical Analysis We calculated the mean difference in (1) ECD and (2) pachymetry between pairs of the 3 measured locations (central, ST, and IN) and tested for significant differences between the locations with a paired test. We created linear regression models adjusted for age and sex to test for differences between eyes that had undergone tube shunt surgery versus control eyes that had undergone cataract surgery. We tested for the relationship between ECD and pachymetry with linear regression. We assumed that glaucoma tube surgery has a greater local effect than a global effect on the corneal endothelial cells and that the ST cornea would be most affected by the tube. Therefore, we created a difference score between the ST and
FIGURE 1. Anterior segment OCT measurements of tube parameters. The top figure represents glaucoma tube in the anterior chamber, and the bottom figure represents the glaucoma tube in the sulcus entering the anterior chamber through the iris.
38
| www.corneajrnl.com
Ó 2014 Lippincott Williams & Wilkins
Cornea Volume 34, Number 1, January 2015
Effect of Glaucoma Tube Shunt Parameters
IN corneas, with the assumption that this represents the best estimate of the change in corneal parameters due to the tube surgery. We tested whether this difference score was associated with several tube parameters in univariate and multivariate linear regression models, using a backward stepwise selection procedure to select the best multivariate model. For all analyses, we used bootstrap resampling at the patient level (9999 repetitions) to account for nonindependence of eyes from the same patient. All analyses were performed with Stata 12 (StataCorp, College Station, TX).
RESULTS Measurements were obtained on 39 eyes that had undergone Ahmed valve placement and on 20 control eyes that had undergone cataract extraction with temporal incisions (Table 1). Of the 33 patients in the post-Ahmed group, the mean age was 58 6 22 years and 17 (51.5%) were female. Of the 13 patients in the control group, the mean age was 68.5 6 8.7 years and 8 (61.5%) were female. The average time since cataract surgery in the control group was 5.7 6 6.1 years. The average time since glaucoma surgery was 2.5 6 2.6 years. Of the eyes in the post-Ahmed group, 1 (2.6%) had congenital glaucoma, 2 (5.1%) had birdshot retinochoroidopathy, 2 (5.1%) had Vogt–Koyanagi–Harada disease and a fluocinolone acetonide intravitreal implant, 6 (15.4%) had previously undergone trabeculectomy, and 32 (82%) were pseudophakic. The average time between cataract surgery and Ahmed tube implantation was 13 6 14.8 months in those who had glaucoma surgery first, and the average time between cataract surgery and Ahmed tube implantation was 82.4 6 114.1 months in those who had cataract surgery first. Ten patients had cataract surgery and glaucoma surgery simultaneously, and 7 never underwent cataract surgery. All of the eyes in the control group had undergone uncomplicated phacoemulsification with intraocular lens implantation but no glaucoma surgery. Phacoemulsification was performed through temporal wounds in both study and control eyes. TABLE 1. Characteristics of Study Eyes Posttube Eyes (N = 39), Mean 6 SD
Control Eyes (N = 20), Mean 6 SD
P
Age of patient, yrs 58 6 22 68.5 6 8.7 0.01 Intraocular pressure, mm Hg* 13.6 6 4.8 16.9 6 2.4 0.002 Time since surgery, yrs 2.5 6 2.6 5.7 6 6.1 — ECD, cells/mm2 Superotemporal 1611.9 6 785.8 2211.3 6 422.5 ,0.002 Central 1902.2 6 841.4 2335.1 6 393.4 0.06 Inferotemporal 1938.4 6 823.5 2382.4 6 368 0.01 Pachymetry, mm ST 654.1 6 95.5 675 6 74.5 0.15 Central 541.2 6 76.5 562.2 6 57.9 0.21 IN 679.1 6 90.3 673.6 6 54.5 0.99 Tube–cornea angle, degrees 29 6 10.5 — — Tube tip–cornea distance, mm 1.1 6 0.6 — — Tube length, mm 3.4 6 0.8 — — *Assessed with pneumotonometry
Ó 2014 Lippincott Williams & Wilkins
In patients with glaucoma tubes, ECD was significantly lower in the ST endothelium compared with either the central or IN endothelium (P , 0.001; Table 2). The pseudophakic control eyes also had ECDs that were lower in the ST endothelium compared with either the central or IN endothelium, although the differences were less marked (Table 2). After adjusting for age and sex, relative differences between the 3 locations did not differ between the posttube and postcataract groups. Figure 2 shows the relationship between ECD and corneal thickness in the ST corneas of the posttube group. ST corneal thickness seemed to be inversely associated with ST ECD, but only at the lower range of ECD values. For example, in eyes with ST ECD , 1000 cells per square millimeter, corneal thickness increased 11 mm (95% CI, 84.5 mm increase to 62.5 mm decrease) for each 100-unit reduction in ECD— although the number of eyes in this stratum was low and the relationship did not achieve statistical significance (P = 0.77, linear regression). In contrast, there seemed to be no association between ST ECD and corneal thickness at the higher range of ECD values (Fig. 2; P = 0.29 in linear regression restricted to eyes with ECD $ 1000 cells/mm2). Various parameters of the tube including tube distance from the cornea, angle of the tube, and tube length obtained by anterior segment OCT are listed in Table 3. In univariate analyses, tube–cornea angle and the closest distance from the tip of the tube to the cornea were statistically significant predictors of ST corneal endothelial cell loss after glaucoma surgery, expressed as the difference between the ST and IN ECD (Table 3). In multivariate analysis, only the distance from the tip of the tube to the cornea remained significant; each millimeter that the tube was closer to the endothelial surface was associated with 353.1 (95% CI, 56.1–650.1; P = 0.02) fewer endothelial cells superotemporally.
DISCUSSION In this study, we used anterior segment OCT and specular microscopy to assess the role of tube positioning on corneal endothelial cell loss after placement of glaucoma drainage devices. We estimated endothelial cell loss as the difference in ECD between the ST and IN corneas, and found that the distance from the tip of the tube was significantly associated with this metric of decreased ECD. Our study confirms that endothelial cells may be better preserved with tubes placed further away from the cornea. Previous studies have reported loss of endothelial cells in patients with glaucoma drainage devices. A 2-year prospective study evaluating corneal endothelial cell counts in various locations of the cornea before and after Ahmed valve placement in the ST portion of the anterior chamber showed that, compared with glaucomatous control eyes with no glaucoma surgery, study eyes had significantly decreased endothelial cell counts superotemporally at all time points after 6 months of follow-up.7 That study reported that the reduction in ST ECD from baseline to 2 years after surgery was 22.6%, which represented a 26.3% lower ECD in comparison with nonoperated control eyes. Our study is consistent with this because we found that the ST ECD in posttube eyes was approximately 27% lower than that of pseudophakic control eyes (1612 vs. 2211 cells/mm2). www.corneajrnl.com |
39
Cornea Volume 34, Number 1, January 2015
Koo et al
TABLE 2. ECD and Pachymetry Measurements Taken at the ST, Central, and IN cornea Posttube (N = 39) Difference in ECD ST minus central ST minus IN Central minus IN Difference in pachymetry ST minus central ST minus IN Central minus IN
Control (N = 20)
Mean Difference (95% CI)
P*
Mean Difference (95% CI)
P*
P†
2290.3 (2407.2 to 2173.4) 2326.5 (2478.8 to 2174.3) 236.3 (2125.5 to 53)
,0.001 ,0.001 0.43
2123.8 (2255.1 to 7.6) 2171.1 (2320.1 to 222) 247.3 (2104 to 9.4)
0.07 0.03 0.10
0.31 0.46 0.92
112.8 (95.2 to 130.5) 225.1 (245.4 to 24.7) 2137.9 (2155.2 to 2120.7)
,0.001 0.02 ,0.001
112.8 (89.9 to 135.7) 1.4 (218.2 to 21) 2111.4 (2122.5 to 2100.3)
,0.001 0.89 ,0.001
0.96 0.22 0.10
*Paired t test comparing values at pairs of corneal locations. †Age- and sex-adjusted linear regression.
One mechanism that has been proposed to explain endothelial cell loss in patients with glaucoma drainage devices is turbulence present at the tip of the implant.9 Our finding of decreased ECD near the tube supports this hypothesis. All tubes in our study had a bevel-up tip potentially directing turbulent flow toward endothelial cells anterior to the tip. In this model, the ST cornea would be most affected by jet flow or turbulence compared with other areas of the cornea. In our study, there was a much larger difference in ST and IN ECD in eyes with a tube (2326.5 6 493.1 cells/mm2) compared with the pseudophakic eyes (2320.1 to 222 cells/ mm2) (Table 2). This suggests that the tube likely did cause a local reduction in ECD, although we could not demonstrate a statistically significant difference between the posttube and control groups after adjusting for age and sex. In a cross-sectional study, disruption of endothelial cells by physical contact of the tube to the cornea at its insertion point was evaluated as another possible explanation
for endothelial cell loss.10 Microscopic fibrotic tissue between the tube and corneal endothelium was detected using highresolution Fourier domain OCT to evaluate anterior chamber tube shunts.10 The interface between the tube and the cornea at the point of insertion was not assessed in our study. Interestingly, tube angle, which hypothetically could affect the extent of fibrosis between the tube and the cornea, was a significant risk factor for endothelial cell loss in univariate analysis in our study. However, the effect of the tube angle on the endothelium could be mediated by tube distance. In future studies, evaluating this interface with OCT by adjusting resolution settings at the time of image acquisition may be useful in further elucidating the effect of drainage devices and their parameters on adjacent tissues. Our study suggests that the distance from the tip of the tube to the cornea is a crucial risk factor for endothelial cell loss. This finding is consistent with a previous study that found that the distance from the tube to the posterior cornea was the most important of several tube parameters tested, although that study did not find a statistically significant association (P = 0.08).4 It is important to note that this previous study was conducted primarily in patients with uveitis and tested for differences in ECD over the tube site, but not relative to an area of a normal cornea as we did in this study. Moreover, the previous study measured the distance between the tip of the tube and the cornea, perpendicular to the tube, whereas we analyzed the distance between the tip of the tube
TABLE 3. Univariate Analyses of Tube Parameters to Predict Endothelial Cells Loss in ST Compared to IN Cornea
Covariate
FIGURE 2. Relationship between ST corneal thickness and ST ECD in patients who underwent tube shunt surgery. In eyes with ST ECD , 1000 cells per square millimeter, the corneal thickness increased 11 mm (95% CI, 84.5 mm increase to 62.5 mm decrease) for each 100-unit reduction in ECD.
40
| www.corneajrnl.com
Tube angle, per degree closer to the cornea Tube distance, per mm closer to the cornea* Tube length, per mm
Reduction in ECD (Cells/mm), ST Compared With IN Cornea (95% CI) 16.3 (3.6–29) 353.1 (56.1–650.1) 2 (133.4 to 2129.5)
P 0.01 0.02 0.98
*The only variable that remained significant after multivariate backward stepwise algorithm.
Ó 2014 Lippincott Williams & Wilkins
Cornea Volume 34, Number 1, January 2015
and the cornea, perpendicular to the cornea (Fig. 1). Without knowing exact physics of turbulence at the tip of the tube, it is difficult to know which distance is more consistent with actual events. Nonetheless, studies are generally consistent and do provide support for the importance of tube distance from the cornea as a significant predictor of endothelial cell loss. Limitations of our study include a small population, lack of standardization of time of follow-up after surgery, and lack of preoperative and longitudinal data. In our study, we did not include slit-lamp or gonioscopic findings. For example, we did not assess peripheral anterior synechiae, which have been reported to alter endothelium integrity.4 Also, we did not compare various types of tubes because all of our patients had Ahmed tubes implanted in the eye. However, a comparison done by Nassiri et al11 of the Ahmed valves to the single plate Molteno implants showed no difference in corneal ECD or corneal thickness at 2 years. High-resolution noninvasive imaging modalities have made studies like ours more feasible. Given the relatively recent availability of these modalities, longitudinal studies measuring progression of corneal endothelial cell loss after shunt placement are still few. The long-term clinical significance of endothelial cell loss after glaucoma tube implantation remains unclear. It is possible that the presence of a tube near the endothelium presents a constant insult to the endothelium, which would be quite different from the presumably 1-time insult of cataract surgery. Longitudinal studies that incorporate imaging of the tube would be helpful in determining clinical importance of minimizing the distance of the tube from the cornea. In the meantime, our study
Ó 2014 Lippincott Williams & Wilkins
Effect of Glaucoma Tube Shunt Parameters
reinforces the widely held view that tubes be placed as far from the cornea as safely possible.
REFERENCES 1. Minckler DS, Francis BA, Hodapp EA, et al. Aqueous shunts in glaucoma: a report by the American Academy of Ophthalmology. Ophthalmology. 2008;115:1089–1098. 2. Barton K, Heuer DK. Modern aqueous shunt implantation: future challenges. Prog Brain Res. 2008;173:263–276. 3. Chihara E, Kubota H, Takanashi T, et al. Outcome of White pump shunt surgery for neovascular glaucoma in Asians. Ophthalmic Surg. 1992;23: 666–671. 4. Hau S, Scott A, Bunce C, et al. Corneal endothelial morphology in eyes implanted with anterior chamber aqueous shunts. Cornea. 2011;30:50–55. 5. Kim CS, Yim JH, Lee EK, et al. Changes in corneal endothelial cell density and morphology after Ahmed glaucoma valve implantation during the first year of follow up. Clin Experiment Ophthalmol. 2008;36: 142–147. 6. Topouzis F, Coleman AL, Choplin N, et al. Follow-up of the original cohort with the Ahmed glaucoma valve implant. Am J Ophthalmol. 1999; 128:198–204. 7. Lee EK, Yun YJ, Lee JE, et al. Changes in corneal endothelial cells after Ahmed glaucoma valve implantation: 2-year follow-up. Am J Ophthalmol. 2009;148:361–367. 8. Ritterband DC, Shapiro D, Trubnik V, et al. Penetrating keratoplasty with pars plana glaucoma drainage devices. Cornea. 2007;26:1060–1066. 9. McDermott ML, Swendris RP, Shin DH, et al. Corneal endothelial cell counts after Molteno implantation. Am J Ophthalmol. 1993;115:93–96. 10. Jiang C, Li Y, Huang D, et al. Study of anterior chamber aqueous tube shunt by fourier-domain optical coherence tomography. J Ophthalmol. 2012;2012:189580. 11. Nassiri N, Majdi-N M, Salehi M, et al. Corneal endothelial cell changes after Ahmed valve and Molteno glaucoma implants. Ophthalmic Surg Lasers Imaging. 2011;42:394–399.
www.corneajrnl.com |
41
