Comparison of Surgically Induced Astigmatism After Glaucoma Filtering and Implant Procedures

Abstract

Purpose

To compare keratometric surgically induced astigmatism (SIA) at 3 months after trabeculectomy (LEC), EX-PRESS shunt (EXP), PRESERFLO MicroShunt (PSM), and Ahmed glaucoma valve (AGV) with ciliary-sulcus tube insertion using vector analysis, and to assess the directionality of SIA.

Study Design

Single-center, nonrandomized, retrospective comparative study.

Subjects and Methods

We retrospectively analyzed 100 eyes with glaucoma undergoing LEC (n = 20), EXP (n = 30), PSM (n = 30), or AGV (n = 20) at a single center. Best-corrected visual acuity (BCVA), intraocular pressure (IOP), and keratometric astigmatism were evaluated preoperatively and at 3 months postoperatively. SIA was calculated using the Jaffe vector method. Directional non-uniformity of the SIA axis was assessed using the Rayleigh test after double-angle transformation. Associations between SIA magnitude and postoperative IOP and BCVA were assessed using Spearman’s rank correlation. In the LEC group, the association between SIA magnitude and the number of residual scleral-flap sutures was also evaluated.

Results

Preoperative and postoperative corneal astigmatism magnitudes did not differ significantly among the four groups. Arithmetic mean individual SIA magnitudes were 1.17 ± 0.80 D in the LEC group, 0.66 ± 0.44 D in the EXP group, 0.65 ± 0.57 D in the PSM group, and 0.58 ± 0.28 D in the AGV group (overall P = 0.035). SIA magnitude was significantly greater after LEC than after EXP, PSM, and AGV (Holm-adjusted P = 0.031, 0.031, and 0.015, respectively), whereas no significant differences were detected among EXP, PSM, and AGV. Significant directional non-uniformity of the SIA axis was detected only in the LEC group, with a circular mean axis of 19.9° (mean resultant length, 0.487; Rayleigh Z = 4.740, P = 0.007), consistent with a with-the-rule–oriented astigmatic change. No significant correlations were detected between SIA magnitude and postoperative IOP or BCVA in any surgical group. In the LEC group, SIA magnitude was not significantly correlated with the number of residual scleral-flap sutures.

Conclusions

At 3 months, LEC was associated with greater keratometric SIA than EXP, PSM, and AGV. Significant with-the-rule–oriented directional non-uniformity was detected only after LEC.

Plain Language Summary

Glaucoma surgery can change the curve of the cornea and cause astigmatism. This can affect how clearly a person sees after surgery. We compared corneal astigmatism 3 months after four glaucoma operations: trabeculectomy, EX-PRESS shunt, PRESERFLO MicroShunt, and Ahmed Glaucoma Valve surgery.

Trabeculectomy was associated with a larger astigmatic change than the other three operations. After trabeculectomy, the astigmatic changes also tended to point in a similar direction. We did not detect this pattern after the other procedures.

These findings suggest that different glaucoma procedures may affect the shape of the cornea differently after surgery. This information may help ophthalmologists anticipate early postoperative visual changes and discuss these possible changes with patients when selecting a glaucoma surgical procedure.

Introduction

Trabeculectomy (LEC) remains one of the most effective options for reducing intraocular pressure (IOP) that cannot be sufficiently lowered with medical therapy.¹ However, LEC has been reported to induce a greater amount of surgically induced astigmatism (SIA) than other glaucoma procedures, which may adversely affect postoperative visual function.^2,3 Postoperative changes in corneal curvature and astigmatism may alter refraction and contribute to transient visual deterioration during refractive recovery.^2,4 Cunliffe et al reported marked early changes in uncorrected visual acuity after trabeculectomy, accompanied by refractive and with-the-rule astigmatic changes.⁴ Dietze et al found that an early increase in corneal astigmatism was associated with a transient reduction in visual acuity.² Claridge et al further demonstrated that postoperative corneal topographic changes may be sufficiently large to affect visual function in some patients.⁵ Thus, even relatively modest procedure-related changes in corneal astigmatism may be clinically relevant when considering postoperative refractive recovery and visual function.^2,4,5 The mechanisms underlying these astigmatic changes remain incompletely understood. Less-invasive filtration procedures such as the EX-PRESS shunt (EXP) and PRESERFLO MicroShunt (PSM), as well as glaucoma drainage devices such as the Ahmed Glaucoma Valve (AGV), have increasingly expanded the surgical options for glaucoma.

Several mechanisms have been proposed to explain SIA after trabeculectomy, including scleral-flap creation, suture-related effects, trabecular meshwork excision, and postoperative bleb-related changes; however, the mechanisms underlying these astigmatic changes remain incompletely understood.^5–8 Comparing SIA among glaucoma procedures that differ in surgical technique may therefore help characterize the procedure-specific patterns of postoperative astigmatic change and improve our understanding of how glaucoma surgery affects corneal shape. Such comparisons may be clinically useful when considering the refractive consequences associated with different glaucoma surgical approaches. Previous studies have evaluated SIA after LEC and EXP, and more recent studies have investigated postoperative SIA after PSM implantation, including direct comparisons with trabeculectomy.^3,9,10 However, data regarding SIA after glaucoma drainage device surgery remain limited, particularly for AGV implantation with ciliary-sulcus tube placement. Moreover, direct vector-based comparisons among LEC, EXP, PSM, and AGV within the same study remain scarce.

We evaluated SIA at 3 months postoperatively because previous studies, including recent studies of PSM, have commonly assessed SIA at this time point, and postoperative astigmatic changes after glaucoma surgery have been reported to approach relative stability by approximately 3 months.^3,8,10 The primary outcome of this study was the magnitude of SIA at 3 months postoperatively among the four surgical procedures. Based on our study hypothesis, we expected LEC to induce a greater SIA magnitude than EXP, PSM, and AGV. We also evaluated the directional characteristics of the SIA axis using vector and circular analyses.

Materials and Methods

We retrospectively reviewed data from 100 eyes of 100 patients who met the study eligibility criteria and underwent glaucoma surgery at Japanese Red Cross Gifu Hospital between March 2020 and April 2025. Cases were reviewed on a consecutive basis from the available surgical and clinical records. Among exclusions that could be confirmed from the available records, three PSM cases were excluded: one because of a history of corneal transplantation and two because they were reoperations. One LEC case was excluded because it was a reoperation, and two EXP cases were excluded: one because it was a reoperation and one because of a history of corneal transplantation. However, because a complete screening log for the entire study period was not available, the exact total number of potentially eligible cases screened and any additional exclusions could not be confirmed retrospectively. All procedures within each surgical group were performed by the same surgeon. The choice of surgical procedure was individualized according to the target IOP, patient age, residual visual field, and overall clinical status.

Patients with intraoperative complications, pre-existing corneal or ocular-surface disease that could affect keratometric measurements (including a history of corneal transplantation), a history of previous glaucoma filtering or implant surgery, or missing data required for the study analyses were excluded. Patients who underwent any other ophthalmic surgery within 3 months before or after the study surgery were also excluded. Lens status was not used as an eligibility criterion, provided that no ophthalmic surgery was performed during this perioperative period.

Data on best-corrected visual acuity (BCVA), corneal keratometry, and intraocular pressure (IOP) were recorded preoperatively and at 3 months postoperatively. Corneal keratometry was obtained during routine clinical care using an autorefractor/keratometer (TONOREF II; NIDEK Co., Ltd., Aichi, Japan), and IOP was measured by Goldmann applanation tonometry. Because this was a retrospective study, the exact number of repeated keratometric acquisitions obtained at each visit and the specific measurement-acceptance criteria were not consistently documented and could not be reliably reconstructed; therefore, these details are not specified.

Surgical Procedure

All surgeries were performed under local anesthesia with sub-Tenon injection of lidocaine (2%). Postoperative medications were moxifloxacin (0.5%; four times daily for 1 month; Nitto Medic, Toyama, Japan), bromfenac (0.1%; twice daily for 3 months; Nitto Medic, Toyama, Japan), and betamethasone (0.1%; four times daily for up to 3 months; Sanbetazon; Santen, Osaka, Japan).

LEC

Surgery was performed in the superior quadrant. A conjunctival peritomy of less than one quadrant was created; a half‑thickness scleral flap measuring 3 × 3 mm² was fashioned. Within the remaining scleral bed, a second scleral flap (approximately 80% thickness) was created. Mitomycin C 0.4 mg/mL was then applied to the surgical site for 5 minutes and thoroughly irrigated with 50 mL of balanced salt solution. Within this second flap, the trabecular meshwork was excised en bloc. Following peripheral iridectomy, the initial scleral flap was secured with a minimum number of 10–0 nylon sutures (Mani, Utsunomiya, Japan) required to maintain the anterior chamber. Typically, three sutures were placed—one at each corner of the flap and one in between. The conjunctiva was reapproximated using 10–0 absorbable sutures (Vicryl; Johnson and Johnson, New Brunswick, NJ, USA). Postoperative laser suture lysis was initiated as early as the first postoperative day and completed within one month, depending on IOP and bleb formation. Typically, suture lysis was performed when IOP exceeded 12 mmHg.

EXP

Surgery was performed in the superior quadrant. After creating a half-thickness rectangular scleral flap (3 × 3 mm²), mitomycin C (0.4 mg/mL) was applied to the surgical site for 5 minutes and the area was thoroughly irrigated with 50 mL of balanced salt solution.

A model P-50 EXP (Alcon Japan, Tokyo, Japan) was inserted into the anterior chamber beneath the flap through a scleral tunnel fashioned with a 25-gauge needle.

The scleral flap was secured with a single 10–0 nylon suture (Mani, Utsunomiya, Japan); thereafter, the conjunctiva was reapproximated with 10–0 absorbable sutures (Vicryl; Johnson and Johnson, New Brunswick, NJ, USA).

Postoperatively, laser suture lysis followed the same protocol as that for LEC—initiated as early as postoperative day 1, typically performed when IOP exceeded 12 mmHg and completed within 1 month according to IOP and bleb status.

PSM

Surgery was performed in the superotemporal quadrant. After a conjunctival peritomy, mitomycin C (0.4 mg/mL) was applied to the surgical site for 5 minutes and the area was thoroughly irrigated with 50 mL of balanced salt solution. The PSM (Santen Pharmaceutical Co., Ltd., Osaka, Japan) was inserted into the anterior chamber through a scleral tunnel fashioned with a double-step knife, with the entry site 3 mm posterior to the corneal limbus and the tunnel directed parallel to the iris plane, in accordance with the manufacturer’s instructions. The conjunctiva was reapproximated with 10–0 absorbable sutures (Vicryl; Johnson and Johnson, New Brunswick, NJ, USA).

AGV

Surgery was performed in the superotemporal quadrant. After a conjunctival peritomy, a half-thickness rectangular scleral flap (3 × 3 mm²) was fashioned. The plate of AGV (model FP7; New World Medical, Rancho Cucamonga, CA, USA) was positioned between the superior and lateral rectus muscles and secured to the sclera 8.5 mm posterior to the limbus with 8–0 nylon sutures (Mani, Utsunomiya, Japan). A 23-gauge scleral tunnel was then created 1.5 mm posterior to the limbus, directed toward the ciliary sulcus. After priming the device according to the manufacturer’s instructions, the tube was inserted through the tunnel into the sulcus. The scleral flap was lightly secured with two corner 10–0 nylon sutures (Mani, Utsunomiya, Japan) to approximate the flap without undue tension. The conjunctiva was reapproximated with 10–0 absorbable sutures (Vicryl; Ethicon, Johnson and Johnson, Raritan, NJ, USA).

Astigmatism Analysis

SIA was calculated using the Jaffe vector method.¹¹ The arithmetic mean SIA magnitude was defined as the mean of the individual SIA magnitudes within each surgical group and was used for comparisons among procedures. For graphical presentation, the centroid SIA vector was calculated by averaging the Cartesian components of the individual SIA vectors within each group and converting the resulting mean vector back to polar coordinates.

For directional analysis, the circular mean SIA axis was calculated separately from the individual SIA axes. Because astigmatism axis is axial data, with 0° and 180° representing the same orientation, each SIA axis was first doubled, the resulting unit-direction components were averaged, and the mean direction was then halved to return it to the 0°–180° astigmatic axis scale. The circular mean SIA axis therefore represents the mean orientation of the individual SIA axes irrespective of SIA magnitude and is distinct from the axis of the centroid SIA vector.

LEC and EXP were performed in the superior quadrant, and right- and left-eye data were combined for directional analysis. In contrast, PSM and AGV were performed in the superotemporal quadrant, which is mirror-symmetric between right and left eyes. Therefore, right and left eyes were analyzed separately for PSM and AGV without reflection or anatomical transformation of the left-eye axes. Directional analyses were thus performed in six groups: LEC, EXP, PSM-right, PSM-left, AGV-right, and AGV-left.

Statistical Analysis

Continuous variables, including age, BCVA, IOP, preoperative and postoperative astigmatism magnitude, and SIA magnitude, were compared among the four surgical groups using Welch’s one-way analysis of variance. For continuous variables other than SIA magnitude, when a significant overall difference was detected, pairwise comparisons were performed using Welch’s t-tests with Holm adjustment for all six pairwise comparisons. Categorical variables were compared using Pearson’s chi-square test. Continuous data are presented as mean ± standard deviation unless otherwise specified.

For SIA magnitude, the arithmetic mean of the individual SIA magnitudes was used for comparisons among surgical procedures. Based on the study hypothesis that the LEC group would have a greater SIA magnitude than the other three groups, the primary pairwise comparisons were LEC vs EXP, LEC vs PSM, and LEC vs AGV, using Welch’s t-tests with Holm adjustment for these three comparisons. As secondary exploratory analyses, pairwise comparisons among EXP, PSM, and AGV were also performed using Welch’s t-tests, with Holm adjustment for these three comparisons.

Directional non-uniformity of the SIA axis was assessed using the Rayleigh test after double-angle transformation.¹² The Rayleigh test was performed separately for the six directional-analysis groups defined above. For each group, the circular mean direction, 95% confidence interval, mean resultant length, Rayleigh Z statistic, and P value were reported. The 95% confidence intervals for the circular mean directions were estimated using nonparametric bootstrap resampling with 10,000 replications. Eyes with an SIA magnitude of 0 D were excluded from directional analysis because the SIA axis was undefined.

Associations between SIA magnitude at 3 months and postoperative IOP and postoperative logMAR BCVA were assessed using Spearman’s rank correlation coefficient (ρ). In the LEC group, the association between the number of residual scleral-flap sutures at 3 months and SIA magnitude was also evaluated using Spearman’s ρ. All statistical tests were two-sided, and P < 0.05 was considered statistically significant. Statistical analyses were performed using R software, version 4.5.0 (R Foundation for Statistical Computing, Vienna, Austria). Patients with missing data required for the analyses were excluded, and no imputation was performed. No a priori sample-size calculation was performed because this retrospective study analyzed the available eligible cases during the study period.

Results

The baseline and postoperative characteristics of the four surgical groups are summarized in Table 1. Significant between-group differences were observed in age (P = 0.013) and glaucoma subtype (P = 0.001), whereas sex distribution (P = 0.249) and laterality (P = 0.856) did not differ significantly among the groups. Age was significantly higher in the AGV group than in the LEC group after Holm adjustment (P = 0.021).

Table 1.

Subject Characteristics

Characteristic	LEC	EXP	PSM	AGV	P value
N (eyes)	20	30	30	20
Age, years (mean ± SD)	69.5 ± 12.7	75.2 ± 9.1	78.1 ± 9.6	79.7 ± 6.1	0.013
Holm-adjusted pairwise P: vs EXP	0.281
Holm-adjusted pairwise P: vs PSM	0.077	0.487
Holm-adjusted pairwise P: vs AGV	0.021	0.178	0.487
Right/Left	12/8	18/12	15/15	11/9	0.856
Sex, male/female	12/8	13/17	11/19	12/8	0.249
Glaucoma type, n (%)
POAG	6 (30.0)	20 (66.7)	16 (53.3)	3 (15.0)	0.001
SOAG	14 (70.0)	10 (33.3)	14 (46.7)	17 (85.0)
Preoperative BCVA (logMAR)	0.41 ± 0.60	0.16 ± 0.27	0.17 ± 0.30	0.72 ± 0.82	0.018
Holm-adjusted pairwise P: vs EXP	0.370
Holm-adjusted pairwise P: vs PSM	0.370	0.860
Holm-adjusted pairwise P: vs AGV	0.370	0.042	0.043
Postoperative BCVA (logMAR)	0.43 ± 0.60	0.19 ± 0.29	0.12 ± 0.48	0.67 ± 0.81	0.028
Holm-adjusted pairwise P: vs EXP	0.329
Holm-adjusted pairwise P: vs PSM	0.222	0.602
Holm-adjusted pairwise P: vs AGV	0.602	0.099	0.063
Preoperative IOP (mmHg)	29.45 ± 8.64	26.60 ± 7.69	25.50 ± 6.79	29.90 ± 7.74	0.143
Postoperative IOP (mmHg)	9.60 ± 3.57	9.57 ± 3.31	7.98 ± 2.67	11.20 ± 2.63	0.002
Holm-adjusted pairwise P: vs EXP	0.974
Holm-adjusted pairwise P: vs PSM	0.281	0.230
Holm-adjusted pairwise P: vs AGV	0.281	0.235	<0.001

Notes: Overall P values for continuous variables were calculated using Welch’s one-way analysis of variance; categorical variables were compared using Pearson’s chi-square test. When a significant overall difference was detected for a continuous variable, pairwise comparisons were performed using Welch’s t-tests with Holm adjustment. For pairwise rows, each displayed P value compares the group shown in that column with the group named in the row label.

Abbreviations: LEC, trabeculectomy; EXP, EX-PRESS shunt; PSM, PRESERFLO MicroShunt; AGV, Ahmed Glaucoma Valve; BCVA, best-corrected visual acuity; IOP, intraocular pressure; logMAR, logarithm of the minimum angle of resolution; SD, standard deviation; n, number; %, percent; mmHg, millimeters of mercury; POAG, primary open-angle glaucoma; SOAG, secondary open-angle glaucoma.

Preoperative BCVA differed significantly among the four groups (P = 0.018). In pairwise comparisons, preoperative BCVA was significantly worse in the AGV group than in the EXP group (Holm-adjusted P = 0.042) and the PSM group (Holm-adjusted P = 0.043). Postoperative BCVA also differed significantly among the groups (P = 0.028); however, none of the pairwise comparisons remained statistically significant after Holm adjustment. Preoperative IOP did not differ significantly among the groups (P = 0.143). In contrast, postoperative IOP differed significantly among the groups (P = 0.002), with a significantly higher postoperative IOP in the AGV group than in the PSM group (Holm-adjusted P < 0.001). Astigmatic parameters are summarized in Table 2. Preoperative and postoperative astigmatism magnitudes did not differ significantly among the four surgical groups (P = 0.512 and P = 0.720, respectively). The arithmetic mean individual SIA magnitudes were 1.17 ± 0.80 D in the LEC group, 0.66 ± 0.44 D in the EXP group, 0.65 ± 0.57 D in the PSM group, and 0.58 ± 0.28 D in the AGV group, with a significant overall difference among the four groups (Welch’s ANOVA, P = 0.035).

Table 2.

Comparison of astigmatism parameters among the surgical groups

A. Astigmatism magnitude
Characteristic	LEC	EXP	PSM	AGV	Overall P value
					(Welch’s ANOVA)
Preoperative astigmatism (D)	1.25 ± 0.97	1.10 ± 0.89	1.46 ± 0.94	1.20 ± 0.78	0.512
Postoperative astigmatism (D)	1.48 ± 1.30	1.36 ± 0.95	1.45 ± 0.82	1.21 ± 0.71	0.72
Arithmetic mean SIA magnitude (D)	1.17 ± 0.80	0.66 ± 0.44	0.65 ± 0.57	0.58 ± 0.28	0.035
Pairwise comparisons of SIA magnitude
Analysis	Comparison	Mean difference,	95% CI	Holm-adjusted P
		first − second (D)
Primary	LEC vs EXP	0.51	0.10–0.91	0.031
	LEC vs PSM	0.52	0.10–0.94	0.031
	LEC vs AGV	0.58	0.19–0.98	0.015
Secondary exploratory	EXP vs PSM	0.01	−0.25–0.28	1
	EXP vs AGV	0.08	−0.13–0.28	1
	PSM vs AGV	0.06	−0.18–0.31	1
B. Centroid SIA vector
Group	N	Centroid SIA magnitude (D)	Centroid SIA axis (°)
LEC	20	0.64	14.8
EXP	30	0.12	79.3
PSM-right	15	0.16	9.5
PSM-left	15	0.13	95.3
AGV-right	11	0.19	124.7
AGV-left	9	0.41	42.6
C. Circular analysis of SIA axis*
Group	N	Circular mean SIA axis	Mean resultant	Rayleigh Z	Rayleigh P value
		(95% CI), °	length (r)
LEC	20	19.9 (3.6–38.6)	0.487	4.74	0.007
EXP	28	85.1 (14.7–150.3)	0.174	0.845	0.433
PSM-right	15	73.1 (2.5–114.7)	0.181	0.494	0.618
PSM-left	11	93.3 (50.2–136.2)	0.392	1.689	0.187
AGV-right	10	105.5 (38.0–171.0)	0.204	0.415	0.671
AGV-left	9	43.2 (24.2–70.1)	0.559	2.816	0.056

Notes: *Eyes with an SIA magnitude of 0 D were excluded from the circular analysis because the SIA axis was undefined.

Abbreviations: D, diopters; CI, confidence interval; SIA, surgically induced astigmatism; LEC, trabeculectomy; EXP, EX-PRESS shunt; PSM, PRESERFLO MicroShunt; AGV, Ahmed Glaucoma Valve.

In the primary pairwise comparisons, SIA magnitude was significantly greater in the LEC group than in the EXP group (mean difference, 0.51 D; 95% CI, 0.10–0.91 D; Holm-adjusted P = 0.031), the PSM group (mean difference, 0.52 D; 95% CI, 0.10–0.94 D; Holm-adjusted P = 0.031), and the AGV group (mean difference, 0.58 D; 95% CI, 0.19–0.98 D; Holm-adjusted P = 0.015). In the secondary exploratory comparisons, no significant differences in SIA magnitude were detected among the EXP, PSM, and AGV groups (all Holm-adjusted P = 1.000).

Directional analysis of the SIA axis was performed separately in the LEC, EXP, PSM-right, PSM-left, AGV-right, and AGV-left groups (Figure 1). A statistically significant directional non-uniformity was detected only in the LEC group, with a circular mean SIA axis of 19.9°, a mean resultant length of 0.487, and a Rayleigh Z statistic of 4.740 (P = 0.007). No statistically significant directional non-uniformity was detected in the EXP, PSM-right, PSM-left, AGV-right, or AGV-left groups (Rayleigh P = 0.433, 0.618, 0.187, 0.671, and 0.056, respectively).

Figure 1.

Distribution of surgically induced astigmatism magnitude and axis according to surgical procedure. Double-angle polar plots show surgically induced astigmatism (SIA) at 3 months postoperatively in the (A) LEC, (B) EXP, (C) PSM-right, (D) PSM-left, (E) AGV-right, and (F) AGV-left groups. Each orange point represents the SIA vector of an individual eye, and the blue point represents the centroid SIA vector for each group. The arithmetic mean SIA magnitude shown below each panel represents the mean of the individual SIA magnitudes and is distinct from the centroid SIA vector. Concentric rings indicate SIA magnitudes of 1, 2, and 3 D; values exceeding 3 D are plotted beyond the outer ring. Angular labels are shown at 15° intervals. SIA axes near 0° correspond to a with-the-rule–oriented astigmatic change, whereas axes near 90° correspond to an against-the-rule–oriented astigmatic change. For PSM and AGV, right and left eyes were analyzed separately because the superotemporal surgical location is mirror-symmetric between eyes; no reflection or anatomical transformation of the left-eye axes was performed.

No significant correlations were observed between SIA magnitude and postoperative IOP in any surgical group (LEC: ρ = −0.040, P = 0.868; EXP: ρ = −0.002, P = 0.990; PSM: ρ = −0.241, P = 0.199; AGV: ρ = −0.133, P = 0.577). Similarly, SIA magnitude was not significantly correlated with postoperative logMAR BCVA in any group (LEC: ρ = −0.340, P = 0.142; EXP: ρ = −0.222, P = 0.238; PSM: ρ = 0.173, P = 0.360; AGV: ρ = −0.157, P = 0.508). In the LEC group, no significant correlation was observed between SIA magnitude and the number of residual scleral-flap sutures at 3 months (ρ = 0.015, P = 0.949).

Discussion

Comparative reports using vector analysis to evaluate SIA after glaucoma surgery are limited.^13–15 Furthermore, detailed comparisons among patients undergoing different surgical techniques are particularly scarce.^3,9,16

In this study, the mean SIA magnitude in the LEC group was significantly greater than that in the EXP, PSM, and AGV groups. In contrast, no significant differences in SIA magnitude were detected among the EXP, PSM, and AGV groups. These findings are generally consistent with previous comparative studies suggesting that trabeculectomy induces greater corneal astigmatic changes than less-invasive filtration or implant procedures.^3,9,16

More recently, Ishiyama et al compared SIA after PSM implantation and trabeculectomy at 3 months postoperatively. They reported that mean SIA was significantly lower after PSM implantation than after trabeculectomy when the procedures were performed in the superonasal quadrant, whereas no significant difference was observed for the superotemporal surgical location.¹⁰ They also found that the centroid SIA tended to shift toward the surgical site in both procedures. These findings suggest that postoperative SIA may be influenced not only by the type of procedure but also by surgical location. Differences in surgical quadrant, study population, and analytic approach should therefore be considered when comparing SIA across studies.

Previous studies have also suggested relatively limited astigmatic changes after angle-based minimally invasive glaucoma surgery (MIGS). Tanito et al directly compared SIA among LEC, EXP, ab externo trabeculotomy, and microhook ab interno trabeculotomy and reported significantly less SIA after microhook ab interno trabeculotomy than after LEC.³ More recently, Nakagawa et al evaluated angle-based MIGS combined with cataract surgery and reported mean SIA magnitudes of 0.77 D, 0.79 D, and 0.76 D after iStent, iStent inject W, and microhook ab interno trabeculotomy, respectively, with no significant differences among the three procedures.¹⁷ These findings may be consistent with the relatively less invasive nature of angle-based MIGS, which avoids conjunctival dissection, scleral-flap creation, and filtering-bleb formation and therefore involves less manipulation of the sclera and ocular surface than filtering surgery. However, direct comparison with the present study should be made cautiously because the procedures in the study by Nakagawa et al were combined with phacoemulsification, which itself may influence postoperative corneal astigmatism, and because the surgical approaches and study populations differed.

The mean differences in SIA magnitude between LEC and the other procedures in the present study were approximately 0.5–0.6 D. Such differences may be clinically relevant to postoperative refractive recovery in some eyes, particularly when pre-existing corneal astigmatism or other optical factors are present. However, the present study did not directly assess uncorrected visual acuity, changes in spectacle prescription, or patient-reported visual quality. Thus, the observed differences may have clinical relevance for postoperative refractive recovery, although their impact on visual function warrants direct evaluation in future studies.

Only the LEC group demonstrated statistically significant directional non-uniformity of the SIA axis, with a circular mean axis consistent with a with-the-rule (WTR)-oriented change. This finding is consistent with previous studies^6,7,18 reporting a WTR shift after LEC. Several mechanisms have been proposed to explain astigmatic changes after LEC, including (i) relative meridional flattening on the flap side due to separation or gaping of the wound edges associated with scleral-flap creation,⁶ (ii) localized corneal steepening in the flap direction attributable to suture tension,⁵ and (iii) localized deformation near the corneal limbus associated with trabecular meshwork excision.⁸ Excessive tissue cautery⁷ and the formation of a large or overhanging filtering bleb⁵ have also been suggested as possible contributors.

In the present study, no significant differences in SIA magnitude were detected among the EXP, PSM, and AGV groups. Furthermore, no statistically significant directional non-uniformity was detected in the EXP, PSM-right, PSM-left, AGV-right, or AGV-left groups. These findings differ to some extent from previous reports suggesting that scleral manipulation or scleral-flap creation may contribute to postoperative SIA.^3,13 Although modest directional tendencies cannot be excluded given the sample sizes, a statistically significant directional pattern was identified only after LEC in the present study.

This discrepancy may partly reflect differences in surgical technique and postoperative management. In the EXP group, laser suture lysis was completed within 1 month postoperatively, and no scleral-flap sutures remained at the 3-month examination. In addition, electrocautery for hemostasis was not used, and no markedly elevated or overhanging blebs were observed. Under these conditions, factors such as residual suture tension, thermal effects, and marked bleb-related deformation may have had relatively limited effects on corneal astigmatism. The lack of a significant difference in SIA magnitude between the EXP and PSM groups may therefore suggest that scleral-flap creation alone is not sufficient to account for the greater SIA observed after LEC. Together, these findings support the possibility that the greater SIA observed after LEC reflects the combined effects of multiple procedure-specific factors rather than scleral-flap creation alone.

In the LEC group, the mean number of residual scleral-flap sutures at 3 months was 2.35 ± 1.90. However, the number of residual sutures was not significantly correlated with SIA magnitude. Although superior sutures, which might be expected to influence the vertical corneal meridian, were preferentially lysed, significant directional non-uniformity consistent with a WTR-oriented change remained at 3 months. These findings suggest that residual suture number alone may not explain the astigmatic changes observed after LEC. However, residual suture number does not fully reflect other potentially relevant factors, including suture tension, position, timing of laser suture lysis, or biomechanical effects occurring earlier in the postoperative period. Other structural effects associated with trabeculectomy, including scleral-flap configuration, limbal tissue alteration, trabecular meshwork excision, and filtering-bleb formation, may also contribute. These findings suggest that residual suture number alone does not account for the WTR-oriented astigmatic change after LEC and highlight the potential contribution of procedure-related structural factors not captured by suture count.

The relationship between SIA and postoperative IOP remains inconclusive. Several studies have reported a negative correlation, whereby lower postoperative IOP is associated with greater SIA.^3,15,18,19 However, other studies have suggested that this association is more apparent during the early postoperative period and diminishes from approximately 3 months onward, indicating a possible time-dependent effect.^8,20 Some studies have also reported no clear association between SIA and the percentage reduction in IOP, suggesting that the absolute postoperative IOP level rather than the magnitude of IOP reduction may be relevant.²¹ Over longer follow-up periods, studies of both LEC and XEN microstent implantation have similarly reported no significant association between SIA and postoperative IOP.²²

In the present study, no significant correlation between SIA magnitude and postoperative IOP was detected within any of the four surgical groups. These findings may partly reflect the relatively small sample size of each group and the restricted range of postoperative IOP values after filtration or implant surgery. Moreover, the relationship between IOP and corneal shape may be influenced by surgical technique, wound healing, scleral biomechanics, and postoperative time point. Therefore, the absence of a statistically significant correlation in the present study should not be interpreted as evidence that postoperative IOP has no influence on SIA.

No significant correlation between SIA magnitude and postoperative logMAR BCVA was detected in any surgical group. Tanito et al³ reported a weak but significant positive correlation between SIA and logMAR BCVA at 3 months, suggesting that greater SIA may be associated with poorer visual acuity. Cunliffe et al⁴ demonstrated prospectively that early postoperative changes in corneal topography can contribute to visual deterioration, with substantial recovery occurring during the early postoperative period. Bormann et al²² reported changes in BCVA during longer-term follow-up after LEC without a significant correlation between SIA and BCVA. A systematic review⁸ also suggested that surgically induced corneal changes may contribute to early postoperative visual deterioration but tend to stabilize over time.

Changes in corneal higher-order aberrations (HOAs) associated with postoperative IOP reduction have also been reported.^23–25 Because SIA primarily reflects the regular astigmatic component of corneal optical change, visual function after glaucoma surgery may also be affected by optical factors that are not captured by SIA magnitude alone, including irregular astigmatism and HOAs. This may partly explain the absence of a detectable association between SIA magnitude and BCVA at 3 months in the present study. Furthermore, because only BCVA was evaluated, the present study cannot determine whether the observed differences in SIA affected uncorrected visual acuity, spectacle dependence, or subjective visual quality.

This retrospective study has several limitations. First, the surgical groups differed significantly in age, glaucoma subtype, preoperative BCVA, postoperative BCVA, and postoperative IOP. In particular, the AGV group included a greater proportion of eyes with secondary glaucoma and had poorer preoperative visual acuity than some of the other groups. Because the surgical procedure was selected according to the clinical characteristics of each eye rather than by randomization, indication-related selection bias and residual confounding cannot be excluded. The relatively small sample size limited the ability to perform robust multivariable adjustment for these differences. Therefore, the observed between-procedure differences in SIA should be interpreted with caution.

Second, SIA was assessed only at 3 months after surgery. Although this time point has commonly been used in previous studies and postoperative astigmatic changes may approach relative stability by approximately 3 months, corneal shape may continue to change as wound healing, suture manipulation, and bleb morphology evolve.^3,8,10 Thus, the present findings should be interpreted as 3-month postoperative outcomes and may not represent longer-term refractive changes.

Third, each surgical group included only 20–30 eyes, and some directional subgroups contained even fewer eyes after separation by laterality. Consequently, the study may have had insufficient statistical power to detect weak associations or modest directional tendencies. In addition, keratometry was derived from routine clinical practice, and the exact number of repeated acquisitions and formal measurement-acceptance criteria could not be standardized retrospectively. Although cases were reviewed on a consecutive basis from the available records and six exclusions could be confirmed, a complete screening log for the entire study period was unavailable; therefore, the exact total number of potentially eligible cases screened and any additional exclusions could not be confirmed retrospectively. Larger prospective studies with longer follow-up, standardized keratometric measurements, predefined measurement-quality criteria, and adjustment for baseline differences are needed to confirm these findings.

Conclusions

In conclusion, in this single-center retrospective study, LEC was associated with a greater magnitude of keratometric SIA at 3 months than EXP, PSM, and AGV. Statistically significant directional non-uniformity of the SIA axis, consistent with a with-the-rule–oriented change, was detected only in the LEC group; in the other surgical groups, a statistically significant directional preference was not detected in the available sample. The mechanism underlying the observed directional change cannot be determined from the present study.

Funding Statement

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. No sponsor was involved in the study design, data collection, data analysis, data interpretation, manuscript preparation, or the decision to submit the article for publication.

Abbreviations

SIA, surgically induced astigmatism; D, diopter; CI, confidence interval; LEC, trabeculectomy; EXP, EX-PRESS shunt; PSM, PRESERFLO MicroShunt; AGV, Ahmed Glaucoma Valve.

Data Sharing Statement

The datasets generated and/or analyzed during the current study are not publicly available because they contain clinical data and are subject to institutional privacy and ethical restrictions. De-identified data may be available from the corresponding author on reasonable request, subject to approval by the relevant institutional review board and institutional data-sharing policies.

Ethics Approval and Informed Consent

This retrospective observational study using existing clinical records was approved by the Institutional Review Board of the Japanese Red Cross Gifu Hospital (approval No.: 025052601) and complied with the principles of the Declaration of Helsinki. The Institutional Review Board waived the requirement for individual written informed consent for participation before study commencement because the study involved no intervention or additional procedures and all analyses were performed using de-identified clinical data. In accordance with institutional policy, information about the study was publicly disclosed before data collection and analysis, and patients were provided the opportunity to decline participation through an opt-out process.

Consent for Publication

This manuscript does not contain any identifiable individual person’s data, images, videos, recordings, or other materials requiring consent for publication.

Disclosure

The author declares that there are no financial or non-financial competing interests related to this work.