One year structural and functional glaucoma progression after trabeculectomy

We evaluated the changes in visual field mean deviation (VF MD) and retinal nerve fibre layer (RNFL) thickness in glaucoma patients undergoing trabeculectomy. One hundred patients were examined with VF and spectral-domain optical coherence tomography (OCT) before trabeculectomy and 4 follow-up visits over one year. Linear mixed models were used to investigate factors associated with VF and RNFL. VF improved during the first 3 months of follow-up (2.55 ± 1.06 dB/year) and worsened at later visits (−1.14 ± 0.29 dB/year). RNFL thickness reduced by −4.21 ± 0.25 µm/year from 1st month of follow-up. Eyes with an absence of initial VF improvement (β = 0.64; 0.30–0.98), RNFL thinning (β = 0.15; 0.08–0.23), increasing intraocular pressure (IOP; β = −0.11; −0.18 to −0.03) and severe glaucoma (β = −10.82; −13.61 to −8.02) were associated with VF deterioration. Eyes with VF deterioration (β = 0.19; 0.08–0.29), increasing IOP (β = −0.09; −0.17 to −0.01), and moderate (β = −6.33; −12.17 to −0.49) or severe glaucoma (β = −19.58; −24.63 to −14.52) were associated with RNFL thinning. Changes in RNFL structure and function occur over a 1-year follow-up period after trabeculectomy. Early VF improvement is more likely to occur in patients with mild/moderate glaucoma, whereas those with severe glaucoma show greater decline over one year. Our findings indicate that progression is observable using OCT, even in late-stage glaucoma.

The demographic and clinical baseline characteristics of the patients are presented in Table 1. The mean (±standard deviation) age of the patients' age was 67.28 ± 8.92 years and 47% female. Baseline VF MD value was −14.68 ± 8.43 dB (range, −7.18 to −20.55 dB), and RNFL thickness was 53.38 ± 13.62 µm (range, 43 to 60 µm). Majority of these patients had exfoliation syndrome and severe glaucoma.

Discussion
The current study investigated the rates of progression of VF MD and RNFL thickness and factors related to changes in the VF MD and RNFL thickness in a clinical sample of glaucoma patients undergoing trabeculectomy. After surgical reduction in IOP, we identified an initial improvement in VF followed by a subsequent decline over 9 months. We also found that the later decline in VF MD correlated with the initial changes in VF MD, degree of IOP lowering and RNFL thinning. Finally, individuals with severe glaucoma had a worse decline in both VF and RNFL thickness. Understanding VF/RNFL thickness changes over time and their related factors may be relevant when managing patients with severe glaucoma who are at greatest risk of lifetime blindness.
A critical issue is whether changes in VF over the short term as observed in the present study are real or related to fluctuations. We have demonstrated an initial VF gain after surgical intervention that could also reflect a regression to the mean (RTM) rather than an actual VF change. There are reasons to believe that the initial VF improvement is real and not simply due to variability. Firstly, our participants underwent to two VF tests at baseline which would mitigate VF measurement inaccuracy. In addition, we selected the second VF report for analysis, which should hopefully reduce learning effects and reduce measurement variability. We further ran a separate analysis using the first VF report and significance of the measured effect size and directions remained similar with the model reported in Table 3. Secondly, those individuals were measured with worse baseline VF MD did not demonstrate a more distinct improvement in MD at three-month follow-up, which is suggestive of RTM. Instead, the VF improvement was seen markedly in eyes with mild or moderate glaucoma. Finally, we perceived significant correlations between VF improvement with IOP control and RNFL thickness, which would not be likely from random fluctuation associated with RTM phenomenon The VF improvement over the 3 months after surgery was significantly related to RNFL thickness measurement, an objective and reproducible OCT biomarker, strongly indicates that the early VF improvement observed in this study is neither related to RTM nor with VF fluctuations.
Others have corroborated that VF improvement after glaucoma surgery as a real likelihood [6][7][8][9] . Spaeth was the first to report the effect of IOP lowering and the initial gain in VF following either trabeculectomy or argon laser trabeculoplasty 9 . Recent studies have reported short-and long-term VF improvement after glaucoma surgical intervention in individuals with mild to moderate glaucoma 6,7 .
We too observed an initial enhancement in VF in the first 3 months followed by a significant decline in the subsequent 9 months. This was seen even when IOP control remained good over the entire study period. However, it was important to state that higher IOP remained a risk factor for VF progression ( www.nature.com/scientificreports www.nature.com/scientificreports/ important risk factor was, however, advanced glaucoma damage with some patients losing as much as 7 dB per year. The association of this VF loss with structural progression indicates that this is not due to cataract formation after surgery. Those patients having an initial improvement in VF were less likely to experience a VF decline  www.nature.com/scientificreports www.nature.com/scientificreports/ during the following 9 months. When comparing the VF changes over one year in the present study with those obtained in other trials 1,[18][19][20][21][22] , the more advanced glaucomatous damage in our population needs to be considered.
An important finding is that RNFL thickness significantly declined over the one-year study period. This was seen in spite of the mostly severe glaucoma cases and the very thin RNFL thickness at baseline. Cross-sectional studies have reported the "floor effect" of RNFL thickness at approximately 50 µm with Spectralis OCT, the level at which point no additional thinning can be distinguished [23][24][25] . Longitudinal studies have, however, shown that structural progression can still be seen in late stage disease and that the floor effects only occur at thinner RNFL values, which is in agreement with the present study 26,27 . Also, postmortem studies suggested that a few surviving ganglion cells remain detected in cases of advanced glaucoma 28 . Hence, our data indicate that RNFL measurements can still be used to monitor VF progression in late stage glaucoma after trabeculectomy.
We 29 and others 30,31 have previously reported that the decline in RNFL thickness is dependent on the post-operative positioning of the lamina cribrosa. Also, we found that changes in VF were neither associated with the position or the shape of the lamina cribrosa. This makes lamina cribrosa parameters an unlikely candidate for sufficient structural progression analysis after trabeculectomy. This is in contrary to a study by Ha and co-workers, where they showed that the baseline mean anterior laminar cribrosa insertion depth was correlated with the rate of VF deterioration 32 . One plausible explanation may be related to the severity of glaucomatous damage. Ha et al. mostly had eyes with early glaucoma whereas ours were composed of eyes with late glaucoma. Hence, the behavior of the lamina cribrosa on VF worsening may be confounded by the glaucoma severity.
Study strengths and limitations. Strengths of the study include the prospective study design, large number of patients with severe glaucoma and spectral-domain OCT imaging. Compared with earlier studies, spectral-domain OCT is objective and reproducible and thus less prone to operator induced variability 33,34 . Limitations of the study include the relatively short observation period, small number of patients with mild or moderate disease and the lack of cataract staging after surgery. Approximately 48% of the eyes had mild cataract, 6% were pseudophakia, whereas the remaining 46% of the eyes did not have any diagnosis of lens disorders. Even though the severity of cataract was not graded, we would like to reiterate that we only included participants with good quality OCT scans. Those with poor quality OCT scans were excluded. Moreover, most of the patients had pseudoexfoliative glaucoma and it is unclear to which degree the results can be applicable to other study populations. Last, non-standard VF reliability criteria were used (e.g. 33% false positive errors instead of 15%). We selected a wider criteria of VF reliability in order to include more patients for analysis and because many patients  www.nature.com/scientificreports www.nature.com/scientificreports/ had late stage disease and therefore were unable to provide more reliable perimetry data. In spite of this limitation, we still observed VF changes over a period of time that were correlated to structural changes as assessed with OCT and therefore are likely to represent true changes rather than fluctuations.
In summary, we have shown that measurable changes occur in both VF MD and RNFL thickness after trabeculectomy. In the early phase, we observed an improvement in VF that was more likely to occur in patients with mild-moderate disease and less likely to occur in patients with RNFL thinning. From 3-12 months, a decline in VF was observed. Risk factors for the VF deterioration were increasing post-operatives IOP, late stage disease, RNFL thinning and absence of VF improvement in the first three months. Our data also indicate that even in late stage glaucoma, structural progression can be monitored with RNFL thickness measurement using spectral-domain OCT.

Methods
This was a longitudinal study of consecutive patients, who were undergoing trabeculectomy at the Vilnius University Hospital Santaros Klinikos (2014-2017) 29 . Approval for conducting the study was obtained from the Vilnius Regional Biomedical Research Ethics Committee and all study procedures adhered to recommendations of the Declaration of Helsinki. Written informed consents were obtained from participants. Inclusion criteria were defined as (1) clinical diagnosis of primary or secondary glaucoma, (2) trabeculectomy indicated because of progressing glaucoma or high risk of glaucoma progression due to high IOP (defined as IOP that is higher than patient's target IOP), (3) best corrected Snellen visual acuity of ≥ 0.1, (4) refractive error from −6.0 D to +6.0 D of sphere and ± 3.0 D of cylinder. Patients were excluded if they had any prior intraocular surgery, except phacoemulsification with intraocular lens implantation, other ophthalmological or neurological diseases affecting the VF, or poor image quality because of opaque ocular media. Glaucoma was defined based on the presence of glaucomatous optic neuropathy (neuroretinal rim thinning, notching or RNFL defects) with or without associated glaucomatous VF defect. A glaucomatous VF defect was defined as glaucoma hemifield test of standard automated perimetry outside normal limits and/or a cluster of at least three contiguous points on the pattern deviation plot with P < 5% and one with P < 1% probability of being normal or pattern standard deviation of <5%. The VF test was considered reliable if false positive and false negative errors were <33% and fixation losses <20%.    www.nature.com/scientificreports www.nature.com/scientificreports/ using Goldmann applanation tonometry (GAT, Haag-Streit AG, Switzerland), spaced one minute apart was calculated. If two measurements differed by more than 2 mmHg, we took a third reading, and averaged the two closest values. One ophthalmologist examined the patients.
Automated perimetry using 30-2 Swedish Interactive Threshold Algorithm Standard strategy (Humphrey visual field analyzer, Carl Zeiss Meditec, Dublin, CA, USA) were performed at five visits (before the trabeculectomy and postoperatively 3 months, 6 months, 9 months and twelve months). Baseline VF was done twice to eliminate the known learning effect and the second VF report was used for current analysis if it was reliable. Glaucoma severity was staged based on the standard automated perimetry using the Hodapp-Parrish -Anderson criteria 35 : mild glaucoma (mean deviation (MD) >−6 dB, less than 25% of points are depressed <5% and less than 10 points are depressed <1% on a pattern deviation plot, all points in the central 5° with sensitivity ≥15 dB), Figure 3. Scatterplot of the change in visual field mean deviation (VF MD) from baseline to 3 rd month follow-up versus the change in VF MD from 1 year after surgery compared with before, stratified by glaucoma severity. Eyes having mild glaucoma were indicated with dark filled circle, those having moderate glaucoma in green hollow diamond and severe glaucoma in black hollow triangle. The VF before trabeculectomy and the VF postoperatively (3 months after surgery) were used for the calculation of VF worsening (r = 0.32; P = 0.005). www.nature.com/scientificreports www.nature.com/scientificreports/ moderate glaucoma (MD > −12dB, less than 50% of points are depressed <5% and less than 20 points are depressed <1% on the pattern deviation plot, only one hemifield having a point in the central 5° with sensitivity <15 dB, no points within 5° of fixation with sensitivity of 0 dB), severe glaucoma (MD < −12dB, more than 50% of points are depressed < 5% and more than 20 points are depressed <1% on the pattern deviation plot, points within the central 5° with sensitivity <15 dB in both hemifields, at least one point with sensitivity of 0 dB within 5° of fixation).

Characteristics
All the patients underwent a limbal-based trabeculectomy, with or without adjunctive 5-fluorouracil, following the same surgical protocol by one of four surgeons. Subsequently, needling with 5-fluorouracil was performed if failure of the filtrating bleb occurred. Only patients with reduced postsurgical IOP continued the study.
Spectral-domain optical coherence tomography. We evaluated the RNFL thickness and lamina cribrosa features using the spectral-domain OCT (Heidelberg Spectralis, Heidelberg Engineering, Dossenheim, Germany) at seven visits (before the trabeculectomy and postoperatively 3-10 days, one month, three months, six months, nine months and twelve months). A 15 × 10° rectangle scan was centered on the optic nerve head. Each OCT volume consisted of 49 serial horizontal B scans (4.5 mm long lines, 40 images averaged) spaced at approximately 63 µm intervals. At least two OCT scans were taken and the one with the best quality was chosen. Images with a quality score ≤15 were excluded. The baseline OCT scan was set as a reference and all subsequent scans were done adherent to it. Potential magnification error was avoided by entering the corneal curvature and refraction of the eye before the OCT scanning. The RNFL thickness was measured automatically from the circumferential SD-OCT scan of 3.4 mm diameter centered at the ONH (single circle B scan of 12°, 100 images averaged).
To visualize the features of the lamina cribrosa, OCT images were enhanced using adaptive compensation (Reflectivity software, version 3.4, Ophthalmic Engineering & Innovation Laboratory, National University of Singapore, Singapore) and the measurements of lamina cribrosa were extracted using Morphology 1.0 software (Ophthalmic Engineering & Innovation Laboratory, National University of Singapore, Singapore) [36][37][38] . Curvature was expressed as the values in mm −1 , negative values describing posteriorly curved lamina cribrosa and positive values indicating anteriorly curved lamina cribrosa 39 . Global shape index ranged between −1 and 1 and corresponds to a transition from spherical cup (GSI = −1) through a symmetric saddle-shaped LC (GSI = 0) to spherical cap (anteriorly curved LC; GSI = 1). The mean LCD was calculated as the mean depth of all points on the surface of the lamina cribrosa.

Statistical analyses.
Baseline values were defined as those before trabeculectomy. The primary outcome variables were the postoperatively changes in VF MD and RNFL thickness. Rates of progression were calculated from longitudinal data using linear mixed models adjusting for baseline age, sex, follow-up duration, and scan quality of OCT at each visit and accounting for correlation between eyes. Associations between clinical factors (independent variables) and changes in VF MD and RNFL thickness (dependent variables) were assessed by  Table 6. Factors related to changes in retinal nerve fibre layer thickness (Late; Postoperative 3 months to 12 months). IOP = intraocular pressure; MD = mean deviation; RNFL = retinal nerve fiber layer; VF = visual field. Linear mixed model was used to adjust for patient cluster. * Multivariate model -adjusted for age, gender, quality of OCT, change in VF MD, change in IOP, baseline IOP, glaucoma severity and lamina cribrosa features (curved, GSI and depth) in the same model. ‡ Changes in late phase of VF MD (dB) or IOP (in mmHg) or curvatures (mm −1 ) or GSI or depth (µm) for 1 µm increase in retinal nerve fibre layer thickness at late phase.