Factors associated with lamina cribrosa displacement after trabeculectomy measured by optical coherence tomography in advanced primary open-angle glaucoma

Abstract

Purpose

To investigate the relationship of lamina cribrosa displacement to corneal biomechanical properties and visual function after mitomycin C-augmented trabeculectomy.

Method

Eighty-one primary open-angle eyes were imaged before and after trabeculectomy using an enhanced depth spectral domain optical coherence tomography (SDOCT). Corneal biomechanical properties were measured with the ocular response analyzer before the surgery. The anterior lamina cribrosa (LC) was marked at several points in each of the six radial scans to evaluate LC displacement in response to intraocular pressure (IOP) reduction. A Humphrey visual field test (HVF) was performed before the surgery as well as 3 and 6 months, postoperatively.

Results

Factors associated with a deeper baseline anterior lamina cribrosa depth (ALD) were cup-disc ratio (P = 0.04), baseline IOP (P = 0.01), corneal hysteresis (P = 0.001), and corneal resistance factor (P = 0.001). After the surgery, the position of LC became more anterior (negative), posterior (positive), or remained unchanged. The mean LC displacement was − 42 μm (P = 0.001) and was positively correlated with the magnitude of IOP reduction (regression coefficient = 0.251, P = 0.02) and negatively correlated with age (regression coefficient = − 0.224, P = 0.04) as well as baseline cup-disk ratio (Regression coefficient = − 0.212, P = 0.05). Eyes with a larger negative LC displacement were more likely to experience an HVF improvement of more than a 3 dB gain in mean deviation (P = 0.002).

Conclusion

A larger IOP reduction and younger age was correlated with a larger negative LC displacement and improving HVF. The correlation between lower SDOCT cup-disc ratio and postoperative negative LC displacement was borderline (P = 0.05). Corneal biomechanics did not predict LC displacement.

Introduction

Loss of visual function in glaucoma is secondary to axonal ganglion cell damage [1] initiated in and near the lamina cribrosa (LC) [2]. Although a causative link between high intraocular pressure and optic nerve damage is well-established in glaucoma, the exact mechanism remains only partially understood [3]. High intraocular pressure (IOP) and IOP fluctuations cause biomechanical stress and strain that compress, dislocate, stretch, and shear the LC [4]. This leads to a mechanical failure of the load-bearing connective tissues of the optic nerve head (ONH), to damage of nearby axons and compromise of the ONH blood supply [2, 4–6]. The mechanical failure is followed by a posterior bowing and compaction of the LC [4, 7]. Patient-specific properties of the LC may explain why some patients are more likely to develop glaucoma damage despite a similar IOP. The extracellular matrix that adds to the biomechanical properties of the LC is composed of collagens, elastins, and proteoglycans [8, 9]. Collagen fibers primarily resist tensile forces and determine tissue elasticity while proteoglycans resist compressive forces and confer viscosity [10].

The proteoglycans are similar in the cornea and LC despite different collagen types in these tissues [11]. Although an in vivo assessment of the biomechanical properties of the LC is not directly possible, the cornea can be readily analyzed with an ocular response analyzer (ORA, Reichert Instruments, Depew, NY). An examination of the biomechanical features of the cornea might serve as a substitute for that of the LC [12, 13]. The ORA [14] uses a metered air jet to displace the cornea and determine its hysteresis. This variable can be described as the delay between a cause and an effect or, in this case, the difference between the pressure at which the cornea bends posteriorly during an air jet applanation and the pressure at which it moves anteriorly again. A larger corneal hysteresis (CH) can be interpreted as increased viscoelastic damping of stress-strain forces.

While the importance of the LC in glaucoma pathogenesis has been shown in mathematical models and in ex vivo and in vivo studies [1, 5, 15, 16], high-resolution in vivo assessment has only become available recently [17, 18]. The association between corneal and ONH biomechanical properties has been explored in several studies in an attempt to propose easily measurable CH as a biomarker for LC biomechanical behavior [2, 12, 13, 19].

The results remained contradictory; while some studies suggested a higher CH is associated with a greater anterior lamina cribrosa depth change (ALDC) [12, 13], others observed an association with a low CH and thin corneal thickness [2, 20]. A limitation of these studies was that the ALDC was measured in healthy eyes following an increase [13] in IOP or only a modest IOP reduction in glaucomatous eyes after initiating medical treatment [12]. The biomechanical properties of the LC in healthy eyes are different from eyes exposed to the chronic stress-strain deformation seen in glaucoma that leads to optic nerve head remodeling [21, 22]. In addition, the IOP reduction achieved by glaucoma eye drops only alters the stress (force per unit area) and strain (proportional deformation) within a limited range [12] given that the Young’s modulus of the ONH is higher than other ocular tissues [23].

In this study, we evaluated the relationship of biomechanical cornea properties to ONH parameters in eyes undergoing a trabeculectomy. We hypothesized that a higher CH would predict a larger ALDC.

Methods

Study design

The protocol of this study was approved by the institutional review board of the Shahid Beheshti University of Medical Science (protocol number: IR.SBMU.ORC.REC.95267) and adhered to the tenets of the Declaration of Helsinki. Informed written consent was obtained from each participant. Patients with uncontrolled primary open-angle glaucoma (POAG) on maximum medical treatment and an age above 25 years were enrolled in this study. The exclusion criteria consisted of a history of prior ocular surgery, need for a combined cataract procedure, high myopia (≤ − 6) and hyperopia (≥ + 6), ocular or systemic comorbidities that could affect the corneal biomechanics measurement including corneal opacities and dystrophies, keratectasia, connective tissue disease and uncontrolled diabetes, and pregnancy or nursing. Only one eye of each eligible patient was included, chosen randomly by coin flip.

Clinical data acquisition

At baseline, all patients underwent a comprehensive ophthalmic examination including best-corrected visual acuity (BCVA), slit lamp examination, Goldmann applanation tonometry, gonioscopy, and fundus examination. Patients had at least two reliable (less than 33% fixation losses, false negatives, and false positives) on 24–2 standard automated perimetry visual fields (Humphrey visual field analyzer (HVF); model 750; Carl Zeiss Meditec, Dublin, CA, USA) before the operation. While 20% is considered a more reliable cutoff point for 24–2 Swedish interactive threshold algorithm, we opted for 33% as we included advanced glaucoma cases and in line with previous studies on this subject [2, 6, 24, 25]. We collected the mean deviation (MD) from the HVF. To detect any changes in visual function after surgery, we defined a 3 dB change in MD as the cutoff point for detecting a change in the visual field [26].

Baseline axial length and central corneal thickness were measured with a modified Michelson interferometer that uses infrared laser light (IOLMaster, Carl Zeiss Meditec, Dublin, CA, USA).

Before the surgery, corneal biomechanical properties including CH and corneal resistance factor (CRF) were obtained by ORA. Three measurements were obtained for each eye, and the average of these measurements was considered for final analysis. Measurements with a waveform score above 5 were included. We only measured ORA parameters preoperatively as it is shown that corneal biomechanics remain unchanged after trabeculectomy [27]. Moreover, we wanted to assess the association between preoperative corneal biomechanics and postoperative laminar behavior.

Spectral domain optical coherence tomography (SDOCT, Spectralis; Heidelberg Engineering, Heidelberg, Germany) measurements were performed after dilation with 1% tropicamide, using an enhanced depth mode by the same operator. The single-line scan width produced by the Spectralis OCT is 8.9-mm long, including 1024 A-scans per B-scan and 25 B-scans per second. Six high-resolution radial scans were centered on the ONH. Since the images had a fixed size but different magnification, all the exported images were rescaled to 1:1 pixel using a customized computer program developed in Matlab (R2010a, the MathWorks, Inc., Natick, MA). On each radial scan, Bruch’s membrane opening (BMO) was identified and connected by a line as a reference line [28] and multiple points were marked on this line with an interval of 50 μm. We applied a subjective LC visibility grading [29] that describes the anterior LC as “grade 0” if no part is visible, “grade 1” if less than 25% is visible, “grade 2” for 25 to 50%, “grade 3” for 50 to 75%, and “grade 4” for greater than 75% visibility. Only scans with Lamin Cribrosa Visibility Grading (LCVG) ≥ 3 were considered for inclusion. The mean of vertical lines from marked points on the reference line to the anterior surface of the LC was designated as the anterior lamina cribrosa depth (ALD). The difference between the pre- and postsurgical ALD was called LC displacement (ALDC). The prelaminar tissue thickness (PLTT) for each point was defined as the difference between the perpendicular distance from the reference line to the overlapping en face prelaminar tissue surface and anterior lamina cribrosa border. The LC thickness (ALD) was the difference between the perpendicular distance from the anterior and posterior LC border at each point.

In summary, ORA measurements were taken once before the operation, visual fields were performed at least two times before and at 3 and 6 months postoperatively. Changes ≥3 dB in MD at 3-month follow-up that were confirmed at 6 months were considered significant. Lamina cribrosa depth was measured once preoperatively and once at 1 month postoperatively.

Surgical technique

After a peribulbar injection of 2-mL 2% lidocaine (Lignidic 2%, Caspian Tamin Pharmaceutical Co., Rasht, Iran), the culde-sac was irrigated with povidone-iodine and normal saline solution followed by sterile draping and insertion of the lid speculum. Then, 0.1 mL of mitomycin C 0.01% (MMC, Mitomycin C Kyowa, Kogyo Company, Tokyo, Japan) was injected into the subtenon space in the superior cul-de-sac and diffusely spread with a blunt spatula over the superior conjunctiva; a Weck-Cel was used to prevent anterior migration of the MMC, limiting it to the area covered by the upper lid. The conjunctival peritomy was performed at the superior to superonasal quadrant for 1.5–2 clock hours followed by blunt dissection of Tenon’s, 1 min after MMC injection. The operation field was copiously irrigated with balanced salt solution (BSS). A 3.0 × 4.0 mm trapezoidal half-thickness scleral flap was created using a crescent knife followed by lamellar dissection of the scleral flap 1 mm into clear cornea. After fashioning a side port, the anterior chamber was entered underneath the scleral flap with a keratome, and the incision was squared off with the side port knife. An anterior tissue block that included clear cornea was removed with a Kelly punch, and a peripheral iridectomy was performed using Vannas scissors. The scleral flap was tied down with two 10–0 nylon releasable sutures, the knots were buried, and the conjunctiva was closed with two 10–0 nylon wing sutures. At the conclusion of surgery, 50 mg cefazolin (Cefazolin 500, Exir pharmaceutical, Tehran, Iran) and 2 mg of betamethasone (Betazone, Caspian Tamin Pharmaceutical, Rasht, Iran) were given into the inferior fornix. Postoperatively, patients were seen on a weekly basis for 1 month and then monthly for up to 3 months. The comprehensive eye exam was performed at each postoperative visit. The EDI-OCT was repeated after 1 month and visual field 3 and 6 months after the surgery.

Statistical analysis

The primary outcome variable was changed in ALD. We explored the relationship between ALD and ALD change compared with parameters such as age, sex, corneal biomechanics, severity of visual field damage, glaucoma diagnosis, central corneal thickness (CCT), and baseline axial eye length. We used linear and multivariable regression analyses. Kolmogorov-Smirnov test was used to establish the normal distribution (p > 0.05). Paired t test was used to compare differences before and after treatment. Independent t test was used to compare differences between groups with more and less than three MD improvements. Pearson correlation coefficients were calculated to find associations between quantitative variables. Regression analysis was used to determine the factors associated with the change in the LC depth. P ≤ 0.05 was considered significant (IBM Corp. Released 2016. IBM SPSS Statistics for Windows, Version 24.0. Armonk, NY: IBM Corporation).

Results

A total of 100 patients were enrolled in this study and 81 were included. Fourteen were excluded for a low LCVG and five lost to follow-up. All patients were phakic from the beginning to the end of the study. The mean age of study participants was 61.0 ± 13.3 years, and 44 (54.3%) of the patients were male (P = 0.901). The demographic data are presented in Table 1. The mean preoperative IOP was 22.8 ± 5.1 mmHg on 3.1 ± 0.9 medications and decreased to 8.6 ± 2.5 mmHg at the 3-month follow-up (Table 2, P < 0.001). The baseline refractive error was 1.65 ± 0.75 diopter, visual acuity was 0.5 ± 0.5 logMAR and decreased to 0.6 ± 0.5 logMAR at final follow-up (P < 0.001), while the HVF MD changed from − 17 ± 5 dB at baseline to − 18 ± 6 dB at final follow-up (P = 0.23). Lamina cribrosa thickness (LCT) and BMO remained unchanged (Ps = 0.398 and 0.234, respectively), but PLTT increased from 47 ± 12 μm at baseline to 52 ± 12 μm (P < 0.001). Also, ALD decreased from 366 ± 167 μm at baseline to 324 ± 165 μm at the 1-month follow-up (P = < 0.001, Table 2). CH was strongly correlated with CRF (regression coefficient = 0.918, P < 0.001) as well as baseline HVF mean deviation (regression coefficient = − 0.234, P = 0.041). As CH was correlated with CRF and is the most studied corneal biomechanics parameter, and to avoid the effects of using collinear variables in the same mode, we only used CH as primary biomechanical property of the cornea in our multivariate regression analysis. There was no correlation between CH and other parameters, like the baseline LCT, visual acuity, axial length, changes in IOP, ALDC, and changes in MD.

Table 1.

Demographics and ocular characteristics

Parameter		Value
Age	Mean ± SD	61.98 ± 13.28
Sex	Male	44 (54.3%)
	Female	37 (45.7%)
Baseline medications	Mean ± SD	3.14 ± 0.85
	Median (range)	3 (1 to 4)
C/D ratio	Mean ± SD	0.81 ± 0.15
AL	Mean ± SD	23 ± 1
	Median (range)	23 (20 to 25)
CCT	Mean ± SD	539 ± 32
	Median (range)	542 (432 to 632)
CH	Mean ± SD	11 ± 2
	Median (range)	11 (6 to 14)
CRF	Mean ± SD	10 ± 1
	Median (range)	11 (7 to 14)

AL axial length, CCT central corneal thickness, C/D ratio cup-disk ratio, CH corneal hysteresis, CRF CRF

Table 2.

Changes in study parameters after intervention

	Pre	Post	Diff	95% CI		P value
	Mean ± SD	Mean ± SD		Lower	Upper
BCVA (logMAR)	0.49 ± 0.52	0.55 ± 0.53	−0.06	−0.08	−0.04	< 0.001
IOP (mmHg)	22.77 ± 5.14	8.56 ± 2.5	14.2	13	15.4	< 0.001
MD (dB)	−17 ± 5	−18 ± 6	−1	−1	0	0.23
LCT (microns)	141 ± 17	142 ± 18	−1	−4	1	0.39
PLTT (microns)	47 ± 12	52 ± 12	1	1	2	< 0.001
BMO (microns)	486 ± 57	485 ± 57	3	2	5	0.234
ALD (microns)	366 ± 167	324 ± 165	42	29	55	< 0.001

BCVA best-corrected visual acuity, IOP intraocular pressure, MD mean deviation, LCT lamina cribrosa thickness, PLTT prelaminar tissue thickness, BMO Bruch’s membrane opening, ALD anterior lamina cribrosa depth

Parameters associated with a deeper ALD were the cup-disk ratio (regression coefficient = 0.07, P = 0.04); baseline IOP (regression coefficient: − 0.276, P = 0.01); and CH (regression coefficient = − 0.523, P = 0.001). Scatter plots of relevant covariates’ significant correlations with ALDC are shown in Fig. 1. ALDC was correlated to a younger age (regression coefficient = − 0.224, P = 0.04) and a larger IOP reduction (regression coefficient = 0.251, P = 0.02), and to some extent, a lower baseline cup-disk ratio was correlated with larger LC displacement after the surgery (regression coefficient = − 0.212, P = 0.05).

Fig. 1.

Scatter plot and regression line of following outcomes. a Lamina cribrosa displacement (ALDC) resulted negatively correlated to age, regression coefficient = − 0.224, P = 0.04. b ALDC was positively correlated with the amount of IOP reduction, regression coefficient = 0.251, P = 0.02. c ALDC was large in eyes with lower baseline cup-disk (C/D) ratio, regression coefficient = − 0.212, P = 0.05. d ALDC was positively associated with improvement in MD, regression coefficient = − 0.18, P = 0.04

LC displacement was not associated with sex, corneal biomechanical properties, axial length, corneal thickness, or ALD (Fig. 2). Highly hyperopic or myopic eyes were not included in this study. The range was 20 to 25 mm with a median of 23 mm.

Fig. 2.

Scatter plot and regression line of following outcomes. a There was no correlation between ALDC and corneal hysteresis (CH), regression coefficient = 0.076, P = 0.58. b ALDC was not associated with baseline axial length of the eye, regression coefficient = − 0.049, P = 0.69. c There was no association between central corneal thickness (CCT) and ALDC, regression coefficient = 0.288, P = 0.09

Improvement of visual field was detected in eight (9.8%) eyes, 3 months following the surgery. MD improvement was correlated with a more extensive ALDC (regression coefficient = − 0.18, P = 0.04, Fig. 1). As is presented in Table 3, only ALDC was significantly associated with improvement of the visual field.

Table 3.

Study variable difference between two groups with and without 3 dB MD improvement

Parameter	Time	No MD improvement	MD improvement	Diff	95% CI		P value
		Mean ± SD	Mean ± SD		Lower	Upper
BCVA (logMAR)	Pre	0.51 ± 0.56	0.44 ± 0.35	−0.13	−0.42	0.16	0.375
	Post	0.6 ± 0.57	0.45 ± 0.34	−0.14	−0.44	0.15	0.332
MD (dB)	Pre	−17 ± 5	−16 ± 7	1.01	−1.9	3.93	0.491
	Post	−19 ± 5	−12 ± 7	5.95	1.96	7.95	0.002
IOP (mmHg)	Pre	21.88 ± 4.9	22.23 ± 4.3	− 1.35	−6.9	−1.71	0.542
	Post	10.56 ± 3.81	9.82 ± 1.5	0.74	−0.63	3.14	0.171
CCT (microns)	Pre	541 ± 34.1	528 ± 27.44	13.26	−4.7	31.22	0.146
CH	Pre	10.6 ± 1.59	10.09 ± 1.87	0.5	−0.4	1.4	0.27
LCT (microns)	Pre	139 ± 10	137 ± 11	−1.3	−14.76	4.15	0.267
	Post	142 ± 19	141 ± 12	−1.07	−10.21	8.82	0.885
BMO (microns)	Pre	488 ± 63	471 ± 34	−17.6	−49.1	14	0.271
	Post	485 ± 63	468 ± 33	−17.1	−48.6	14.4	0.284
ALDC (microns)		−23.2 ± 37.6	−119.7 ± 60.0	−96.5	−120.24	−72.75	< 0.001

BCVA best-corrected visual acuity, MD mean deviation, IOP intraocular pressure, CCT central corneal thickness, AL axial length, CH corneal hysteresis, LCT lamina cribrosa thickness, BMO Bruch’s membrane opening, ALDC anterior lamina cribrosa depth change

Discussion

Our understanding of glaucomatous optic neuropathy has evolved to include complex biomechanical stress-strain cycle modeling [4]. The examination of IOP-induced lamina cribrosa movements is receiving increasing interest because they reveal important aspects of biomechanical LC properties [4, 30] and can be measured in vivo [17]. CH is associated with the development and progression of glaucoma independent of corneal thickness and intraocular pressure [31]. The cornea and the sclera contain similar extracellular matrix proteoglycans with comparable viscosities although both are derived from different tissues during ocular development with different collagen fibers [32]. Eyes with a higher CH have been observed to experience a larger posterior LC dislocation with increasing IOP [13], a larger ALDC following a medical IOP reduction [12], and more ONH deformation during experimental IOP elevations [20]. We hypothesized that a higher CH would be representative of LC hysteresis and correlated to a larger post-trabeculectomy ALDC [20]. Surprisingly, we found this was not the case in our patients. Sigal discussed that the geometry and mechanical properties of the optic nerve head are highly complex and variably influence each other during IOP-induced stresses and strains [33]. The inclusion of normal participants with healthy loading-unloading cycles and Young’s modulus [13] or a smaller, medical IOP reduction [12] may explain these differences. In fact, Gizzi et al. found anterior displacement of the lamina cribrosa in advanced glaucoma compared to posterior displacement in milder cases following acute rise in IOP. Their finding underscores the complexity of posterior sclera and different behaviors of lamina cribrosa to chronic versus acute IOP changes [34]. A lower CH has been reported to correlate with glaucoma progression [35], possibly due to a reduced viscoelastic dissipation of mechanical forces that could harm axons passing through the LC. On the other hand, larger acute ALDCs have been observed with a large CH [20]. Glaucoma is a chronic condition that subjects the LC to the forces of IOP over a long time with an increasing deformation of the LC that becomes permanent. LC creep and tissue remodeling may contribute to the pathogenesis of glaucomatous optic neuropathy. The residual plastic biomechanical deformation affects the LC stress-strain curve and its displacement after IOP reduction [36], which also corroborate with less displacement in higher baseline cup-disc ratio, as observed in our study. As seen before, we observed that patients with a larger IOP reduction had a greater ALDC [37–41] which inversely correlated with age [37, 42, 43].

We saw an MD improvement in eight (9.8%) patients with a greater ALDC, comparable to previous reports [26, 44, 45]. Previous studies showed an improvement in HVF indices between first and second visual fields but not subsequent ones [46, 47]. The extent of LCDC reflects the amount of IOP reduction [37–41], but its clinical importance was not clear. Our study suggests that a negative ALDC may remove some of the strain on the ONH.

There are several limitations to our study. The baseline mean deviation and cup-disk ratio were − 17 ± 5 dB and 0.81 ± 0.15, respectively, consistent with mostly advanced glaucoma. In advanced disease, it becomes difficult to delineate the border [48] but by using several reference line points as well as the LCVG, we only had to discard 15% of images compared to other studies [12, 49, 50].

In summary, age, magnitude of IOP reduction, and to some extent, lower baseline cup-disc ratio were correlated to the extent of anterior lamina cribrosa displacement after trabeculectomy, but corneal biomechanical properties were not. The visual field mean deviation improved more commonly after a larger anterior lamina cribrosa displacement.