Bleb Analysis by Using Anterior Segment Optical Coherence Tomography in Two Different Methods of Trabeculectomy

Abstract

Purpose.: To investigate the correlation between bleb morphology and IOP control via the modified Indiana Bleb Appearance Grading Scale (IBAGS) and anterior-segment optical coherence tomography (AS-OCT) in two different trabeculectomy (TLE) groups.

Methods.: This study involved 94 eyes with primary open angle glaucoma that underwent two different TLE methods: limbal-based TLE (group I, 62 eyes) and fornix-based TLE (group II, 32). IOP control was defined as successful with an IOP ≤ 20 mm Hg and ≥20% reduction of preoperative IOP. IBAGS and various parameters of the bleb height, extent, wall thickness, ciliochoroidal detachment (CCD), and lake under the scleral flap (LUSF) were obtained by slit-lamp and AS-OCT, respectively. Correlation between IOP control and IBAGS/AS-OCT parameters were assessed by SAS.

Results.: Both groups had the same success rate. As to correlation between IOP control and IBAGS, extent and Seidel were the best-paired parameters in group I (Cp = 3.0402, R = 0.6401), yet no parameter was significant in group II (maximum R = 0.1599). As to correlation between IOP control and AS-OCT, the combinations of height, extent, and the minimum value of bleb wall thickness were significant (Cp = 0.2037, 0.2314, R = 0.4336, 0.4330) in group I. In group II, no parameter was significant, except CCD and/or LUSF (P = 0.032). As to coincidence of IBAGS and AS-OCT parameters, height and extent in group I (P = 0.000, P = 0.000) and height in group II were statistically significant (P = 0.020).

Conclusions.: IOP control in limbal-based TLE seemed to be more dependent on the large size and thinned-wall bleb than that in fornix-based TLE.

Introduction

The success of trabeculectomy (TLE) depends on the formation of a functioning filtering bleb. Therefore, bleb morphology is an important predictor for a successful surgical outcome. The association between the function and the appearance of the bleb created by TLE has been investigated. ^{1 –5} The Indiana Bleb Appearance Grading Scale (IBAGS) ¹ and the Moorfields Bleb Grading System (MGS) ⁵ were recently introduced for the assessment of bleb morphology. Those two grading systems include various bleb parameters, such as height, extent, vascularity, bleb leakage, width, microcystic changes, and encystment, that are evaluated and graded independently by use of slit-lamp photographs that are viewed and scored using the IBAGS photograph. IBAGS and MGS are similar grading systems; however, Seidel (bleb leak) is not applied in MGS. Ultrasound biomicroscope (UBM) reportedly enables the visualization of the internal structure of the bleb. ⁶ By use of UBM, Yamamoto et al. ³ classified 4 bleb types and found that IOP control was associated with those bleb types. It has also been reported that UBM has a 91% sensitivity for predicting functioning blebs and a 70% specificity for predicting nonfunctioning blebs. ⁶ In a UBM examination, an eye cup must be put on the patient's eye, and the examination process is known to be time-consuming. ⁷ Thus, the time needed to perform UBM creates the possible risk for a bleb infection. On the other hand, anterior segment optical coherence tomography (AS-OCT) is a quick and easily tolerated procedure for the patients and allows for noncontact high-resolution imaging. ² Recently, AS-OCT was introduced for not only the study and assessment of anterior chamber angles, ^4,5 but also for investigating the morphology of blebs. ^8,9 Bleb appearance is very important for evaluating the IOP-lowering effect of TLE; however, it is thought that the observed deep sclera, such as “lake under the scleral flap” (LUSF) or “ciliochoroidal detachment” (CCD), created by TLE may also play a role in the IOP-lowering effect. Spectral domain OCT (SD-OCT) reportedly fared poorly compared with AS-OCT in imaging deep structures in blebs, such as the bleb cavity, scleral flap, the subflap space, and internal ostium. ¹⁰ Therefore, time-domain AS-OCT may be more useful for assessing bleb function, as it penetrates more deeply into the sclera than does SD-OCT. 

With regard to TLE surgery, there is currently a worldwide trend to shift from limbal-based TLE to fornix-based TLE, because limbal-based TLE reportedly has a greater chance of creating an avascular bleb with a high risk of blebitis than does fornix-based TLE. ^11,12 Eyes with an avascular bleb, which are often observed in limbal-based TLE, have low IOP; however, eyes with a flat bleb, which are sometimes experienced in fornix-based TLE, also show good IOP control. Therefore, it could be speculated that there is a different IOP-lowering mechanism between the two TLE methods. To the best of our knowledge, no morphological study has been conducted to compare the bleb between fornix-based and limbal-based TLE by using AS-OCT. Therefore, the purpose of this study was to investigate the functional differences for lowering IOP mechanism between the two different TLE methods by using modified IBAGS and AS-OCT. 

Materials and Methods

This is a retrospective cross-sectional study of bleb morphology evaluated by modified IBAGS and AS-OCT. Enrolled in this study were POAG patients who underwent TLE and were consecutively recruited when they came for routine follow-up visits at the Japanese Red Cross Medical Center from July 2007 to March 2008. Patients with secondary glaucoma due to pseudo-exfoliation, uveitis, or neovascular glaucoma and those within the 1-year follow-up period after TLE were excluded from this study. The methods of TLE were divided into the following two groups: limbal-based conjunctiva (group I) and fornix-based conjunctiva (group II). Except for the incision area (location), the method of TLE was the same in both groups. The size of the conjunctival flap and TLE were 4.0 × 4.0 mm and 1.5 × 3.0 mm, respectively. Mitomycin C with a concentration of 0.04% was topically applied under and over the flap and 5 to 7 mm away from the edge of the flap for 3 minutes in all cases. Written informed consent was obtained from all patients, and the study was approved by the institutional review board of the Japanese Red Cross Medical Center and conducted in accordance with tenets set forth in the Declaration of Helsinki. In an attempt to ascertain reproducibility of modified IBAGS, every patient was evaluated by two of the authors (TH, TO) independently using slit-lamp microscopy. If the results of those evaluations differed, a third author (SS) assessed them using the bleb photographs. The Seidel test was evaluated only by the first author. For bleb analysis, bleb height (Z1), bleb extent (Z2), vascularity (Z3), and Seidel (Z4) were evaluated by modified IBAGS. For the assessment of IBAGS bleb height (Z1), grading was dividing into one of the following five stages (H0–4): H0: no bleb, H1: less than the corneal thickness (CT), H2: 1 CT < H2 ≤ 2 CT, H3: 2 CT < H3 ≤ 3 CT, and H4: H4 > 3 CT. The bleb was then imaged on the same day using AS-OCT (ZEISS Visante OCT Model 1000; Carl Zeiss Meditec, Dublin, CA). ² In a sitting position, the subject was asked to look down and the upper lid was gently retracted to expose the bleb as fully as possible in the superior bulbar conjunctiva. Taking care to avoid pressure on the globe and bleb, four directions of standard AS-OCT images were obtained for each bleb, as shown in Figure 1A. Every scan was adjusted to the center of the scleral flap. Each AS-OCT scan image was obtained and printed at the size of 7.3 × 18.6 cm, and the following various parameters were obtained: inside area of the bleb in cut B∼D (X1∼3 = height in B∼D × extent, maximum: X_1∼3max), height of the bleb in cut B∼D (X4∼6, maximum: X_4∼6max), extent of the bleb in cut A (X7), bleb wall thickness in cut B∼D (X8∼10, minimum: X_8∼10min), existence of CCD in cut B∼D (X11), extent of LUSF in cut B∼D (X12∼14, maximum: X_12∼14max), the height of LUSF in cut B∼D (X15∼17, maximum: X_15∼17max), and the area of LUSF (X18 = X12∼14 × X15∼17, maximum: X_18max = X_12∼14max × X_15∼17max). Bleb wall thickness was defined as the distance between the first reflective signals of the conjunctiva to the top of the subconjunctival fluid space. If subconjunctival space was not observed, the bleb wall was regarded as unmeasurable. Height of the bleb was defined as the distance between the first reflective signals of the conjunctiva to perpendicular to the sclera at the highest of the images (Fig. 1B). Height of the bleb was measured at the maximum area. Extent of the bleb (Fig. 1C) (X7) was measured at the length of the bulging part of the bleb in cut A in Figure 1A. All of the measurement values except height and extent in AS-OCT were expressed as actual size on the printed images. Height in AS-OCT was classified into one of the five stages (H0–4) by comparing CT at the limbus. Extent in AS-OCT was expressed by the same four stages with IBAGS (E0: 0 < 1 Clock Hours, E1: 1–2 Clock Hours, E2: 2 < Clock Hours < 4, and E3: Clock Hours ≥ 4). We defined LUSF as the fluid space (lake) under the scleral flap. The existence of CCD was defined as positive if ciliary body and/or choroidal detachment were observed. For the purpose of analysis, the status of postoperative IOP was classified into the following three criteria: “excellent” (postoperative IOP ≤ 20 mm Hg and ≥30% reduction of preoperative IOP), “good” (postoperative IOP ≤ 20 mm Hg and ≥20% reduction of preoperative IOP), and “fail” (postoperative IOP ≥ 21 mm Hg or ≤20% reduction of preoperative IOP). Success in IOP control was defined as the eyes that showed excellent and good IOP control. 

Figure 1

 

(A) Scan mode in four directions. Scan A: a tangential section to the limbus; Scan B: an oblique section 45 degrees to the limbus; Scan C: a radial section perpendicular to the limbus; Scan D: an oblique section 135 degrees to the limbus. (B) AS-OCT image of a radial section perpendicular to the limbus. Cut at section C. Arrow indicates the height of bleb. Patient 21 in group I. (C) AS-OCT image of a tangential section to the limbus. Cut at section A. Arrow indicates the extent of bleb. Patient 21 in group I.

In the two groups (group I and group II), the following five items were assessed: IOP control in the two groups, the correlation between IOP control and IBAGS, the correlation between IOP control and AS-OCT, the coincidence of height and extent between modified IBAGS and AS-OCT, and topical glaucoma medication and IOP control in the two groups. For statistical analysis, we adopted the correlation coefficient of modified IBAGS and AS-OCT as a measure of coincidence on modified IBAGS and AS-OCT. SAS software (SAS Institute, Inc., Cary, NC) was used for statistical analysis in this study. 

Results

This study involved 94 eyes of 75 patients (62 eyes of 46 patients in group I and 32 eyes of 29 patients in group II). The median follow-up period was 4.86 ± 1.61 years in group I and 3.54 ± 1.57 years in group II. Final assessment of each patient was performed in the following manner: if a result of the third author was the same as the first or the second author's results, we used the third author's result. If a result of the third author was different from the other authors, we used the middle evaluation. There was no difference of more than two steps on each parameter between the first and the second author's assessment. In group I, the rates of agreement of the first two authors' assessments for each parameter (height, extent, and vascularity) were 90%, 92%, and 98%, respectively. However, in group II, the rates of agreement of the first two authors' assessments for each parameter were 81%, 75%, and 100%, respectively. The findings of IOP control in both groups were as follows: excellent: 56.5% (group I), 53.3% (group II); good: 22.6% (group I), 28.1% (group II); fail: 21.0% (group I), 18.8% (group II). No significant differences were found between both groups with regard to the three IOP criteria; excellent (P = 0.759), good (P = 0.553), and fail (P = 0.800). 

For correlation between IOP control and parameters of modified IBAGS for both groups, we investigated which set of explanatory variables was the best as a predictive model, based on the model selection criterion (Cp). According to the method of Cp, the smaller value indicates the more significant correlation. In group I, the Cp value of extent (Z2) and Seidel (Z4) was the minimum value (Cp = 3.0402, Table 1). Wide extent of bleb with leak showed success in IOP control (Fig. 2). In group II, if any parameter (Z1–Z4) was considered, the correlation of IOP control was not significant because the maximum of multiple correlation coefficient (R) was 0.1992 (Table 2). 

Figure 2

 

A slit-lamp photograph of a bleb with excellent IOP control (patient 5 in group I). Large extent and avascular bleb is seen. The bleb showed point leaks. IBAGS: H2E1V1S1.

Table 1

 

Relationships Between IOP Control and Modified IBAGS in Group I

No. of Variables	Cp	Multiple Correlation Coefficient, R	Sets of Independent Variables
2	3.0402	0.6401	Z2,Z4
3	3.2970	0.6535	Z2,Z3,Z4
3	4.6386	0.6432	Z1,Z2,Z4
2	4.7886	0.6263	Z2,Z3
4	5.0000	0.6558	Z1,Z2,Z3,Z4
1	5.7188	0.6024	Z2
3	5.7929	0.6342	Z1,Z2,Z3
2	6.3036	0.6141	Z1,Z2

Table 2

 

Relationships Between IOP Control and Modified IBAGS in Group II

No. of Variables	Cp	Multiple Correlation Coefficient, R	Sets of Independent Variables
1	−0.5599	0.1599	Z3
1	−0.2174	0.1153	Z1
1	−0.1207	0.0995	Z2
1	0.1425	0.0223	Z4
2	1.0399	0.1992	Z1,Z3
2	1.0965	0.1942	Z2,Z3
2	1.4401	0.1597	Z3,Z4
2	1.7204	0.1245	Z1,Z4

For correlation between IOP control and parameters of AS-OCT for both groups, in group I, Cp in the combination of the maximum of height (X_4∼6max) and the minimum of the bleb wall thickness (X_8∼10min) was the smallest of all variable sets (Cp = 0.2037, Table 3). These data indicated that high blebs with a thin wall in the AS-OCT image showed a significantly high success rate of IOP control (Fig. 3). Cp in the combination of extent (X7) and the minimum bleb wall thickness (X_8∼10min) was the secondary smallest (Cp = 0.2314), and this value was almost identical to that in the combination of X_4∼6max and X_8∼10min (Cp = 0.2037, Table 3). These data indicated that wide blebs with thin walls were also a significantly high success rate of IOP control. Therefore, blebs with a thick wall tended to fail even if the blebs were rather high and wide (Fig. 4). We investigated whether the eyes with no or slight bleb formation have any correlation between IOP control and the existence of LUSF CCD. There were 26 eyes (41.9%) that had no or slight bleb formation (less than H1 by AS-OCT). Among those eyes, there were 6 eyes (23.1%) with CCD, 16 eyes (61.5%) with LUSF, and 4 eyes (15.4%) with both LUSF and CCD. The existence of LUSF and/or CCD did not have a significant correlation in IOP control compared with the eyes of no LUSF and CCD (13 eyes, 61.9%) (P = 0.125). 

Figure 3

 

AS-OCT image of a bleb with excellent IOP control (patient 6 in group I), IBAGS: H2E2V1S1.

Figure 4

 

AS-OCT image of a thick bleb wall with failed IOP control (patient 63 in group I), IBAGS: H2E2V0S0. A hyporeflex space can be observed (asterisk).

Table 3

 

Relationships Between IOP Control and AS-OCT in Group I

No. of Variables	Cp	Multiple Correlation Coefficient, R	Sets of Independent Variables
2	0.2037	0.4336	X4∼6max, X8∼10min
2	0.2314	0.4330	X7, X8∼10min
1	0.8603	0.3684	X4∼6max
3	1.3162	0.4533	X4∼6max, X7, X8∼10min

In group II, the maximum of multiple correlation coefficient (R) was 0.1507 in all sets of parameters (Table 4); therefore, the correlation of IOP control and all parameters did not show any significance. The occurrence of no or slight bleb formation (less than H1 evaluated by AS-OCT) in group II (65.6%) was higher compared with that in group I (41.9%). Among the eyes with no or slight bleb formation, there were 3 eyes (14.3%) with CCD (Fig. 5), 14 eyes (66.7%) with LUSF (Fig. 6), and 0 eyes (0%) had both. The existence of LUSF and/or CCD showed significant correlation in IOP control compared with the eyes of no LUSF and CCD (9 eyes, 69.2%) (P = 0.016). 

Figure 5

 

AS-OCT image of a flat bleb (no height and extent) with CCD (arrows). IOP control was excellent (patient 75 in group II). IBAGS: H0E0V0S0.

Figure 6

 

AS-OCT image of a flat bleb with the LUSF. IOP control was good (patient 76 in group II). IBAGS: H0E0V2S0.

Table 4

 

Relationships Between IOP Control and AS-OCT in Group II

No. of Variables	Cp	Multiple Correlation Coefficient, R	Sets of Independent Variables
1	−1.5308	0.1507	X11
1	−1.5076	0.1449	X7
1	−1.3290	0.1330	X4∼6max
1	−1.3236	0.0877	X18max
1	−1.2519	0.0500	X8∼10min

Coincidence of height and extent between modified IBAGS and AS-OCT was investigated in each group (Table 5). In group I, the correlation coefficient of height (Z1 [modified IBAGS] and X_4∼6max [AS-OCT]) was very highly correlated (R = 0.768, P = 0.000), and that of extent (Z2 [IBAGS] and X7 [AS-OCT]) was highly correlated (R = 0.584, P = 0.000). In group II, although the correlation coefficient of Z1 and X_4∼6max was smaller than that in group I, there was significant correlation between those parameters (R = 0.410, P = 0.020). On the other hand, the correlation coefficient of Z2 and X7 was not significant (R = 0.291, P = 0.106). 

Table 5

 

Significance on Correlation Coefficients Between Each Parameter in Modified IBAGS and AS-OCT

	Correlation Coefficient, R	P Value
Group I
Z1 and X4∼6max	0.7680	0.000
Z2 and X7	0.5840	0.000
Group II
Z1 and X4∼6max	0.4100	0.020
Z2 and X7	0.2910	0.106

As for the glaucoma medications, no patient received oral acetazolamide. The respective topical glaucoma medications in the two groups (group I, group II) were latanoprost (48%, 38%), beta-blocker (48%, 44%), dorzolamide (27%, 31%), and bunazosin (2%, 6%). The number of topical glaucoma medications in each IOP control classification (excellent, P = 0.928; good, P = 0.745; fail, P = 0.769) did not show any significance between group I and group II (Table 6). 

Table 6

 

Number of Topical Glaucoma Medications in the Classification of IOP Control (Excellent, Good, Fail)

	Excellent	Good	Fail
Group I	0.91 ± 1.09 (0–3)	1.86 ± 1.23 (0–4)	1.54 ± 1.39 (0–3)
Group II	0.88 ± 1.22 (0–3)	1.67 ± 1.41 (0–3)	1.33 ± 1.37 (0–3)
P value	0.928	0.745	0.769

 Numbers in parentheses indicate the range of topical glaucoma medication. P value was evaluated by t-test.

Discussion

IBAGS ¹ and MGS ⁵ are excellent tools for investigating the relationship between bleb morphology and IOP control. We adopted IBAGS because the Seidel test, which is important for affecting IOP, is not included in MGS. Although the original IBAGS method involves the use of slit-lamp photography, ¹ we applied the modified IBAGS method by slit-lamp microscopy observation without photographs in this present study. We opted to use the modified method because the extent of bleb height is difficult to evaluate by slit-lamp photographs, even when the photographs are taken with slit-lamp projection. Bleb leak in the Seidel test is also difficult to judge via a slit-lamp photograph, because the bleb leak appearance will change according to the amount of elapsed time post application of fluorescein on the bleb. Moreover, the bleb leak can change, so we therefore posited that the judgment of bleb leak should be evaluated by slit-lamp observation, and that it was enough that the judgment be performed by the first author alone. In the original method of IBAGS, classification of the height is divided into four scores: flat (H0), low (H1), medium (H2), and high (H3). In this present study, the bleb height was classified into five scores (H0–4) by comparing the thickness of the cornea in the slit-lamp observation. UBM has reportedly been used to estimate bleb and investigate the correlation between IOP control and bleb morphology. ⁶ However, UBM is time consuming and has a risk for bleb infection. Recently, AS-OCT was introduced, and it allows for a quick, direct, and noncontact examination of the patient's eye. ^2,4,8,9 AS-OCT has gained favor because it is a method that produces higher resolution compared with UBM. Various morphological studies have been performed using different types of OCT, ^{2,4,8 –10} and SD-OCT is reportedly excellent for viewing the superficial part of the bleb; however, it has limited clinical utility because it does not provide useful information about deep features in blebs. ¹⁰ On the other hand, time-domain AS-OCT is an excellent device for the observation of deep structures of the bleb, such as LUSF or CCD, which may be very important to evaluate bleb function. We investigated the correlation between IOP control and bleb morphology using the parameters of modified IBAGS and AS-OCT images by dividing the materials into two groups (group I: limbal-based TLE, group II: fornix-based TLE). The coincidence of height and extent between modified IBAGS and AS-OCT in the two groups was also evaluated. 

Some reports have shown a significant difference in IOP control between limbal- and fornix-based TLE, ^13,14 but other reports have not. ^15,16 In this study, no statistical difference in IOP control was found between the two groups (P = 0.800). Moreover, as for the topical glaucoma medications, no statistical difference was found between the two groups with regard to the different IOP control classifications (i.e., excellent, good, and fail) (Table 6, P = 0.769). These results suggest that the number of topical medications had no influence on IOP control in both groups. In the analysis of IOP control and bleb morphology evaluated by modified IBAGS, extent (Z2) and Seidel (Z4) were found to be the most important parameters for predicting success in IOP control in limbal-based TLE (group I) (Table 1). On the other hand, in the assessment of AS-OCT, the combination of the maximum of height (X_4∼6max) and the minimum of bleb wall thickness (X_8∼10min) or the combination of extent (X7) and the minimum of bleb wall thickness (X_8∼10min) were found to be the most important parameters for IOP control (Table 3). The bleb with a thinner wall thickness accompanies avascular bleb (Figs. 2, 3) in which Seidel is more possible to occur. Therefore, the analysis of IOP control and bleb morphology assessed by modified IBAGS seemed to show the same result as those by AS-OCT. From these results, bleb wall thickness (Fig. 4) and the size of the bleb in AS-OCT was important for assessment of bleb function, especially in the limbal-based TLE. 

On the other hand, in group II, the multiple correlation coefficient (R) between IOP control and all parameters of modified IBAGS and AS-OCT were significantly low (Tables 2, 4). The main reason for this fact may be that more than 50% of the eyes (65.6%) in group II showed no or slight bleb formation (less than H1 by AS-OCT). Among the eyes with no or slight bleb formation, the existence of CCD and/or LUSF in group I did not show significant correlation for IOP control compared with no CCD and/or LUSF. However, the existence of CCD (Fig. 5) and/or LUSF (Fig. 6) in group II showed significant correlation for IOP control. It has been speculated that CCD is a result of enhanced uveoscleral outflow. ¹⁷ Direct routes from the anterior chamber to the supraciliary space may be formed by TLE in eyes with CCD. LUSF may also enhance uveoscleral outflow by newly created routes from the LUSF to the uvea through supraciliary spaces. Therefore, the existence of CCD and/or LUSF may be important for uveoscleral outflow, especially in eyes with no or slight bleb. To that regard, the eyes with a successful IOP-lowing effect were divided into two groups. The reason for the successful IOP control in group I was a large and thin wall bleb; however, the successful IOP control in group II may be more dependent on uveoscleral outflow, possibly through CCD and/or LUSF than through the bleb. 

As to the coincidence of modified IBAGS and AS-OCT parameters, height in both groups and extent in group I were statistically significant (Table 5); especially, height in group I was found to very highly correlated (R = 0.768). This may be because the height in avascular bleb, which was more frequent in group I, seemed to be easy to assess by comparing the thickness of the cornea in modified IBAGS. This might also be the reason for the slightly lesser agreement in the assessment of bleb height in group II (81%) when compared with that in group I (90%). The reason for no coincidence about extent in group II (P = 0.106) may be because the border of the bulging part was very difficult to determine in the vascular bleb when extent was evaluated by modified IBAGS and AS-OCT. The poor agreement in relation to extent in group II (75%) compared with group I (92%) between the authors may reflect the fact of no coincidence about extent in group II assessed by modified IBAGS and AS-OCT. 

In summary, the IOP-lowering effect in groups I and II is dependent on the size and thickness of the bleb, and CCD and/or LUSF, respectively. Further investigation is needed to elucidate how uveoscleral outflow is enhanced via CCD or LUSF. 

Acknowledgments

We express our deep gratitude to Tetsuro Sakurai (Center of Education, Tokyo University Science, Suwa, Japan) for assistance with the statistical evaluation. 

Disclosure: T. Hamanaka, None; T. Omata, None; S. Sekimoto, None; T. Sugiyama, None; Y. Fujikoshi, None