Detection and agreement of event-based OCT and OCTA analysis for glaucoma progression

Abstract

Objective

To examine event-based glaucoma progression using optical coherence tomography (OCT) and OCT angiography (OCTA).

Methods

In this retrospective study, glaucoma eyes with ≥2-year and 4-visits of OCT/OCTA imaging were included. Peripapillary capillary density (CD) and retinal nerve fibre layer thickness (RNFL) were obtained from 4.5 mm × 4.5 mm optic nerve head (ONH) scans. Event-based OCT/OCTA progression was defined as decreases in ONH measurements exceeding test-retest variability on ≥2 consecutive visits. Visual field (VF) progression was defined as significant VF mean deviation worsening rates on ≥2 consecutive visits. Inter-instrument agreement on progression detection was compared using kappa(κ) statistics.

Results

Among 147 eyes (89 participants), OCTA and OCT identified 33(22%) and 25(17%) progressors, respectively. They showed slight agreement (κ = 0.06), with 7(5%) eyes categorized as progressors by both. When incorporating both instruments, the rate of progressors identified increased to 34%. Similar agreement was observed in diagnosis- and severity-stratified analyses (κ < 0.10). Compared to progressors identified only by OCT, progressors identified only by OCTA tended to have thinner baseline RNFL and worse baseline VF. VF progression was identified in 11(7%) eyes. OCT and VF showed fair agreement (κ = 0.26), with 6(4%) eyes categorized as progressors by both. OCTA and VF showed slight agreement (κ = 0.08), with 4(3%) eyes categorized as progressors by both.

Conclusions

OCT and OCTA showed limited agreement on event-based progression detection, with OCT showing better agreement with VF. Both OCT and OCTA detected more progressors than VF. OCT and OCTA may provide valuable, yet different and complementary, information about glaucoma progression.

Introduction

Clinical monitoring of glaucoma requires periodic assessment [1]. While the visual field (VF) test is traditionally used to detect glaucoma progression, there has been increased evidence showing structural changes, such as retinal nerve fibre layer (RNFL) thinning measured by optical coherence tomography (OCT), may be more sensitive in detecting progression in the earlier stages [2–4]. With a high reproducibility [5–7], OCT is also considered more reliable in long-term follow-up, although its utility can be limited by floor effect in advanced glaucoma [8, 9].

Recently, OCT angiography (OCTA) has been evaluated for assessment of glaucoma. In addition to a robust diagnostic performance [10, 11], retinal vessel density (VD) measured by OCTA has demonstrated good correlation with optic nerve head (ONH) OCT and VF parameters [12–14]. Moreover, it can predict functional and structural progression [15–18]. Notably, OCTA has a wider dynamic range with slightly greater, although satisfactory, variability as compared to OCT [7, 19, 20]. Prior studies have shown a stronger functional correlation of ONH VD loss than that of retinal nerve fibre layer (RNFL) thinning as glaucoma worsens [21, 22]. This suggests that OCTA ONH VD may be more sensitive than RNFL to measurement changes as glaucoma progresses. Although the progression detection capabilities of OCT and VF have been examined [23–25], there has been very limited information about the value of OCTA for detecting glaucoma progression. Whether OCTA may be beneficial for progression detection and how it compares to OCT is unclear.

In clinical practice, whether the changes in clinical tests exceed the measurement variability limits derived from a stable patient population is crucial in judging possible glaucoma progression. Although trend-based analyses have been more widely used for progression detection, event-based analysis may help detect progressive glaucomatous changes within fewer visits [26–29], especially when assessing measurements with good reproducibility. The performances of structural parameters, particularly OCTA measurements, in event-based analysis remains largely unknown. In addition, to the best of our knowledge, no prior study has examined the agreement between OCT and OCTA on progression detection. Therefore, in this study, the capabilities and agreement of OCT and OCTA in glaucoma progression detection using event-based analysis were evaluated.

Materials and methods

This retrospective cohort study adhered to the tenets of the Declaration of Helsinki and the Health Insurance Portability and Accountability Act. The study protocol was approved by the University of California San Diego Human Research Protection Program (NCT00221897).

Participants

Participants from the Diagnostic Innovations in Glaucoma Study (DIGS) [30, 31] were included. Written informed consent was obtained. Inclusion criteria for DIGS were: (1) age ≥18 years; (2) open angles on gonioscopy; (3) refraction within ±5.0 dioptres spherical and within ±3.0-diopters cylinder at study entry; (4) best-corrected visual acuity (BCVA) of 20/40 or better. Relevant clinical information, including demographics and systemic medical history, was also collected. Routine examinations of DIGS included: (1) annual comprehensive ophthalmic examination with dilated fundus examination, slit-lamp biomicroscopy, BCVA, and stereoscopic optic disc photography; (2) semi-annual examination of Goldmann applanation tonometry (for intraocular pressure [IOP] measurement), VF, and OCT/OCTA imaging; (3) gonioscopy and ultrasound pachymetry at the first visit. Further exclusion criteria for this study included: (1) axial length ≥27 mm; (2) uveitis; (3) history of trauma; (4) non-glaucomatous optic neuropathy; (5) coexisting retinal disease including diabetic retinopathy; (6) history of Parkinson’s disease, clinical dementia, or stroke.

The primary analysis was performed on glaucoma eyes (baseline diagnosis of glaucoma suspect and primary open angle glaucoma [POAG]) from DIGS with at least 2-years, 4-visits of OCT/OCTA follow-up. An additional subset of healthy eyes that met the same criteria was used as supplementary control to determine the specificity of the progression criteria (Supplementary Table 1), given the possibility of false positive finding on healthy eyes with greater measurement variability and aging effects. The definition for POAG was having repeatable (≥2 consecutive tests) and reliable (fixation losses and false negatives ≤33% and false positives ≤15%) abnormal VF results using the 24-2 Swedish Interactive Thresholding Algorithm (SITA) with a pattern standard deviation (PSD) outside the 95% normal limits or a glaucoma hemifield test (GHT) result outside the 99% normal limit. The definition for glaucoma suspect was having a suspicious-appearing optic disc or an elevated IOP (≥22 mm Hg) without repeatable glaucomatous VF damage. Healthy eyes were defined as having: (1) Intraocular pressure (IOP) ≤21 mmHg and without a history of IOP elevation; (2) normal-appearing optic disc with intact neuroretinal rim and retinal nerve fibre layer; (3) visual field test result with a pattern standard deviation within the 95% normal limits and glaucoma hemifield test result within normal limits.

Analysis of OCTA and spectral domain (SD)-OCT progression

This study included 4.5 mm × 4.5 mm optic nerve head (ONH) OCT/OCTA scans (304-A scans in each B-scan and 304-B scans) acquired using the Avanti Angiovue system (Optovue, Inc. Fremont, CA, software version 2018.1.1.63) [32]. The OCT/OCTA images were acquired simultaneously and RNFL thickness were analyzed from the same scan. From the ONH scans, the circumpapillary capillary density (CD) and peripapillary RNFL thickness were measured from the 0.75 mm-wide elliptical annular area surrounding the optic disc boundary. Quality review of OCT/OCTA images was performed by trained graders. Images with any of the following feature were excluded: (1) scan quality<4; (2) residual motion artifacts visible as irregular vessel pattern on the en-face angiogram; (3) image cropping or local weak signal; (4) off-centred fovea; (5) poor clarity; (6) uncorrectable severe segmentation errors. Event-based progression by OCT/OCTA was defined as having repeatable (≥2 consecutive visits) decreases in OCT/OCTA measurements from baseline greater than the pre-set cut-offs (see below for definitions) during follow-up [19, 26].

Since there is no agreed upon gold standard for OCT/OCTA cut-offs to detect progression event, we used test-retest variability (TRV, calculated as within-subject test-retest standard deviation ×2.77) reported in earlier studies and considered the specificity in healthy eyes when deciding the cut-offs for main analysis, which were set at 4.1% and 4.02 μm for CD and RNFL, respectively [33, 34]. To further validate the generalizability of our results, different TRVs reported in other studies and the TRV derived from the current cohort were used as cut-offs in the sensitivity analyses [20, 26, 35, 36].

Analysis of VF progression

VF test was performed using SITA standard 24-2 threshold test (Humphrey Field Analyzer, Carl Zeiss Meditec, Dublin, California, USA). Early glaucoma was defined by a baseline 24-2 VF mean deviation (MD) ≥-6 dB, whereas moderate-advanced glaucoma was defined by a VF MD < −6 dB. VF analysis was performed on glaucoma eyes using serial VF tests obtained around the same period of OCT/OCTA visits. Considering the greater variability of VF than OCT/OCTA, 24-2 VF tests from up to 1-year before the baseline OCT/OCTA visit to within 6-months after the last visit were included to ensure at least 2 years and 5 visits of VF tests for the progression analysis. Linear regression analysis was used to estimate the rate of VF MD worsening, and the presence of VF progression was defined as having repeatable (≥2 consecutive visits) and significantly (P < 0.05) negative slopes of VF MD worsening during follow-up.

Statistical analysis

Demographic and clinical characteristic data were presented as count (%) for categorical variables and mean (95% confidence interval [CI]) for continuous variables. The count (%) of progressors in glaucoma eyes were calculated individually using OCT, OCTA, and VF criteria. The number of progressors in healthy eye subset (specificity) was also evaluated. The rates of changes in OCT/OCTA and VF measurements over time were calculated using linear regression analysis. Agreement among different instruments on glaucoma progression detection was compared using kappa (κ) statistics, and baseline diagnosis- and severity-stratified agreement analyses were performed. The strength of agreement was categorized as: 0 = poor agreement, 0–0.20 = slight agreement, 0.21–0.40 = fair agreement, 0.41–0.60 = moderate agreement, 0.61–0.80 = substantial agreement, 0.81–1.00 = almost perfect agreement [37]. Eye characteristics were compared between eyes categorized as progressors only by OCTA and only by OCT using linear mixed-effects models, which accounted for the within-participant variability. As aforementioned, sensitivity analyses using different cut-offs (TRVs) for event-based analysis were performed to confirm the generalizability of our results. In addition, a trend-based progression analysis (method as described in prior study) was performed supplementarily to further examine OCT/OCTA/VF inter-instrumental agreement [38]. Between-eye correlation was adjusted for during the analyses where applicable. Statistical analyses were performed using Stata version 17.0 (StataCorp, College Station, TX). A P value ≤ 0.05 was considered statistically significant.

Results

A total of 147 glaucoma eyes (56 glaucoma suspects, 91 POAG) from 89 patients were included. The demographic and clinical characteristics are shown in Table 1. The mean (95% CI) age at baseline was 68.5(66.2, 70.8) years, with a mean(95% CI) follow-up of 4.0(3.9, 4.2) years. The baseline and last-visit mean (95% CI) 24-2 VF MD was −3.9(−4.8, −3.0) dB and −4.6(−5.6, −3.7) dB, RNFL was 80.7(77.4, 82.8) µm and 78.3(75.6, 81.0) µm, and CD was 43.9(42.9, 44.8)% and 40.6(42.9, 41.8)%, respectively. The demographic and characteristics of the healthy eye subset are shown in Supplementary Table 1.

Table 1.

Demographic and clinical characteristics of glaucoma eyes.

	Glaucoma eyes (N = 147 eyes from 89 patients)
Patient-level characteristics
Age at baseline (years)	68.5 (66.2, 70.8)
Sex (Female, n)	45 (51%)
Race (n, eyes)
Black or African American	21 (24%)
White	54 (61%)
Asian	12 (13%)
Others	2 (2%)
Hypertension (Hypertensive, n)	42 (47%)
Diabetes (Diabetic, n)	9 (10%)
Eye-level characteristics
Baseline diagnosis (n, eyes)
Glaucoma suspect	56 (38%)
POAG	91 (62%)
Mean IOP during follow up (mmHg)	15.1 (14.5, 15.7)
CCT (µm)	534.5 (527.4, 541.6)
Axial length (mm)	24.4 (24.2, 24.6)
Baseline 24-2 VF MD (dB)	−3.9 (−4.8, −3.0)
Last-visit 24-2 VF MD (dB)	−4.6 (−5.6, −3.7)
Baseline RNFL (µm)	80.7 (77.4, 82.8)
Last-visit RNFL (µm)	78.3 (75.6, 81.0)
Baseline CD (%)	43.9 (42.9, 44.8)
Last-visit CD (%)	40.6 (42.9, 41.8)
Signal strength index	62.9 (61.4, 64.3)
Maximum number of OCT/OCTA follow-up visits	6.0 (5.7, 6.3)
Maximum number of VF follow-up visits	7.4 (7.0, 7.7)
OCT/OCTA Follow-up time (years)	4.0 (3.9, 4.2)
VF Follow-up time (years)	4.4 (4.3, 4.6)

Values are shown in mean (95% confidence interval), unless otherwise indicated.

CCT central corneal thickness, IOP intraocular pressure, MD mean deviation, RNFL peripapillary RNFL, CD peripapillary capillary density, POAG primary open angle glaucoma, VF visual field.

The distribution of progressors and non-progressors identified by OCT, OCTA, and VF among glaucoma eyes and healthy subset are summarized in Supplementary Table 2. OCTA progression was detected in 33 (22%) out of 147 glaucoma eyes and 2 (6%) out of 34 healthy eyes. OCT progression was detected in 25 (17%) glaucoma eyes and 1 (3%) healthy eye. VF progression was detected in 11 (7%) glaucoma eyes only, with none of the healthy eyes identified as progressing. For all instruments, the percentage of progressors were higher in POAG (OCTA:25%, OCT:19%, VF:10%), as compared to in glaucoma suspects (OCTA:18%, OCT:14%, VF:4%). And for all diagnosis, OCTA detected the most progressors, followed by OCT and VF. Supplementary Fig. 1 illustrates the percentages of progressors identified by each instrument in glaucoma eyes by diagnosis.

Table 2 shows the agreement of OCT and OCTA on event-based progression detection. In general, the agreement between OCT and OCTA was slight (kappa value [κ] < 0.10 for all). Among the 147 glaucoma eyes, 7 (5%) were categorized as progressors by both OCT and OCTA, 26 (17%) were categorized by only OCTA, and 18 (12%) were categorized by only OCT, leading to a κ of 0.06, which indicates slight agreement (P = 0.232). The overall rate of eyes identified with progressive changes when incorporating both instruments was 34% (OCT and OCTA), in comparison to 17% and 12% by stand-alone OCTA and stand-alone OCT, respectively. The specificity remained >90% in the healthy eye subset when using either instrument alone or in combination. Similarly slight agreement between OCT and OCTA was observed when evaluating glaucoma suspect eyes (n = 56; κ = 0.08, P = 0.284) and POAG eyes (n = 91; κ = 0.04, P = 0.331) separately. A severity-stratified analysis was also performed. OCT and OCTA again showed slight agreement on early glaucoma eyes (n = 114; κ = 0.06, P = 0.257) and moderate-advanced glaucoma eyes (n = 33; κ = 0.05, P = 0.382).

Table 2.

Agreement of OCT and OCTA on progression detection.

All glaucoma eyes (N = 147)
		OCT			κ (P value)
		Progressor	Non-progressor	Total
OCTA	Progressor	7 (5%)	26 (17%)	33	0.06 (0.232)
	Non-progressor	18 (12%)	96 (65%)	114
	Total	25	122	147
Glaucoma suspect (N = 56)
OCTA	Progressor	2 (4%)	8 (14%)	10	0.08 (0.284)
	Non-progressor	6 (11%)	40 (71%)	46
	Total	8	48	56
POAG (N = 91)
OCTA	Progressor	5 (5%)	18 (20%)	23	0.04 (0.331)
	Non-progressor	12 (13%)	56 (62%)	68
	Total	17	74	91
Early glaucoma (N = 114)
OCTA	Progressor	5 (4%)	20 (18%)	25	0.06 (0.257)
	Non-progressor	13 (11%)	76 (67%)	89
	Total	18	96	114
Moderate-advanced glaucoma (N = 33)
OCTA	Progressor	2 (6%)	6 (18%)	8	0.05 (0.382)
	Non-progressor	5 (15%)	20 (61%)	25
	Total	7	26	33

OCT optical coherence tomography, OCTA optical coherence tomography angiography, POAG primary open angle glaucoma.

The characteristics of eyes categorized as progressors only by OCTA (OCTA-only progressors, n = 26) and eyes categorized as progressors only by OCT (OCT-only progressors, n = 18) were additionally compared (Supplementary Table 3). OCTA-only progressors tended to have worse baseline VF MD (mean[95% CI] = −5.6[−8.2, −2.9] dB), as compared to OCT-only progressors (mean[95% CI] = −3.0[−5.3, −0.6] dB) (P = 0.077). A thinner baseline RNFL was also found for OCTA-only progressors (mean[95% CI] = 74.0[67.8, 80.2] µm), as compared to OCT-only progressors (mean[95% CI] = 83.4[76.2, 90.6] µm, P = 0.028). Other characteristics did not differ between the two groups.

The rates of structural and VF change over time of progressors and non-progressors categorized by OCT and OCTA are summarized in Table 3. When comparing OCT progressors to OCT non-progressors, faster rates of mean (95% CI) RNFL worsening (progressors: −1.49[−2.04, −0.93] µm/year; non-progressors: −0.17[−0.30, −0.05] µm/year) and VF MD worsening (progressors: −0.41[−0.63, −0.20] dB/year; non-progressors: 0.17[−0.25, −0.10] dB/year) were observed (P < 0.05 for both). When comparing OCTA progressors to OCTA non-progressors, a faster mean (95% CI) CD worsening rate (progressors: −1.73[−2.04, −1.43] %/year; non-progressors: −0.54[−0.66, −0.41] dB/year; P = 0.017) was observed. RNFL worsening rate also tended to be faster in OCTA progressors than non-progressors, but the difference did not reach statistical significance (P = 0.099). There was no difference in the rates of VF MD worsening (P = 0.324) among OCTA progressors and non-progressors.

Table 3.

Rates of structural and VF change based on progression status by OCT and OCTA.

		RNFL worsening rate (µm/year)	P value	CD worsening rate (%/year)	P value	VF MD worsening rate (dB/year)	P value
OCT	Progressors (N = 25)	−1.49 (−2.04, −0.93)	0.004	−0.90 (−1.35, −0.44)	0.628	−0.41 (−0.63, −0.20)	0.017
	Non−progressors (N = 122)	−0.17 (−0.30, −0.05)		−0.79 (−0.93, −0.64)		−0.17 (−0.25, −0.10)
OCTA	Progressors (N = 33)	−0.63 (−1.00, −0.26)	0.099	−1.73 (−2.04, −1.43)	<0.001	−0.28 (−0.44, −0.12)	0.324
	Non-progressors (N = 114)	−0.33 (−0.51, −0.16)		−0.54 (−0.66, −0.41)		−0.20 (−0.27, −0.11)

Significant P values were bolded. A significant P value indicates significant difference in the rates of change between progressors and non-progressors.

MD mean deviation, OCT optical coherence tomography, OCTA optical coherence tomography angiography, RNFL peripapillary RNFL, CD peripapillary capillary density, VF visual field.

Table 4 shows the agreement of VF with OCT and OCTA on progression detection among the glaucoma eyes. When assessing the agreement between OCT and VF, 6 (4%) eyes were categorized as progressors by both OCT and VF, 19 (13%) eyes were categorized by only OCT, and 5 (3%) eyes were categorized by only VF, leading to a fair agreement with κ = 0.26 (P < 0.001). For the agreement between OCTA and VF, 4 (3%) eyes were categorized as progressors by both OCTA and VF, 29 (20%) eyes were categorized by only OCTA, and 7 (5%) eyes were categorized by only VF, showing slight agreement with κ = 0.08 (P = 0.125). The numbers of OCT, OCTA, and VF progressors and their distribution across different instruments are presented as Venn diagrams in Fig. 1.

Table 4.

Agreement of VF with OCT and OCTA on progression detection.

OCT vs VF
		OCT			κ (P value)
		Progressor	Non-progressor	Total
VF	Progressor	6 (4%)	5 (3%)	11	0.26 (<0.001)
	Non-progressor	19 (13%)	117 (80%)	136
	Total	25	122	147
OCTA vs VF
		OCTA			κ (P value)
		Progressor	Non-progressor	Total
VF	Progressor	4 (3%)	7 (5%)	11	0.08 (0.125)
	Non-progressor	29 (20%)	107 (73%)	136
	Total	33	114	147

Significant P-values were bolded. MD mean deviation, OCT optical coherence tomography, OCTA optical coherence tomography angiography, VF visual field.

Fig. 1. Venn diagrams demonstrating the counts and distribution of OCT, OCTA, and VF progressors.

Venn diagrams demonstrating the counts and distribution of OCT, OCTA, and VF progressors. (Abbreviation: OCT optical coherence tomography, OCTA optical coherence tomography angiography, VF visual field).

Results of the sensitivity analyses using different cut-offs for event-based analysis are shown in Supplementary Table 4. Consistent findings of a slight OCT/OCTA agreement (κ ≤ 0.20) were found. Specifically, using TRV of longitudinal Angiovue/Optovue imaging (prior studies and current cohort) [20, 26], the OCT/OCTA agreement remained slight, and there were more OCTA progressors than OCT progressors, with the specificity remaining >90% in healthy eyes. In the Supplementary analysis based on trend-based method (Supplementary Table 5), the agreement between OCT and OCTA was slight (κ = 0.16), between OCT and VF was fair (κ = 0.22), and between OCTA and VF was slight (κ = 0.11), similar to that observed from the main analysis.

Discussion

This study examined the detection and agreement of event-based glaucoma progression analysis using both OCT and OCTA. Overall, more glaucoma eyes were found with progressive CD loss by OCTA than with progressive RNFL loss by OCT, and there only was slight agreement between these two tests. Both instruments detected more progressors and agreed weakly with VF, although the agreement between VF and OCT was slightly better. Consistent results were found in Supplementary analyses using different cut-offs for event-based analysis and trend-based method for progression detection. Our results provided novel insight into not only event-based progression analysis by structural parameters, but also the potential performance and agreement of OCTA in assessing progression in comparison to other clinical tests.

With the limited OCT/OCTA agreement and a trend toward more progressors identified by OCTA, our results support prior related studies [21, 26, 39]. In both cross-sectional and longitudinal studies, a more pronounced/faster loss of VD compared with RNFL thickness loss has been found [21, 39]. In an advanced glaucoma cohort, progressive OCTA changes were also found in more eyes than OCT changes, although the OCT floor effect might have played a role [26]. While no previous work has assessed OCT and OCTA agreement on progression detection, some have implied the differing roles of vascular/perfusion loss and retinal thinning in the pathophysiology of glaucoma [14, 40, 41]. A doppler OCT study found no correlation between reduced perfusion and RNFL loss, although both showed association with VF loss [40]. Another study also found the association between VD loss and VF defect was independent of RNFL thinning [14]. These suggest that VD and thickness changes may provide valuable, yet different and complementary information, about glaucomatous change.

The potential complementary role of OCTA in assessing glaucomatous change is also supported by the increased detection of progression when OCTA assessment is incorporated to OCT assessment in this study (17% vs 34%). Notably, OCTA changes were found to correlate better with VF progression than were OCT changes in prior studies [22, 42]. Furthermore, progressive VD loss during the initial evaluation period was found associated with faster subsequent VF loss [15]. Still, it remains to be determined if progressive changes detected by OCTA are true glaucomatous progression. Nevertheless, the aforementioned results indicate this finding is likely clinically relevant, and that OCTA may provide complementary information to VF and OCT in monitoring glaucomatous progression. Since VF and OCT are traditionally used for progression evaluation, more studies with longer follow-up are needed to confirm the relationship between VD loss and glaucomatous functional impairment.

In the present study, OCTA-only progressors tended to have worse VF and thinner RNFL than OCT-only progressors. This implies that OCTA may better reflect progressive changes as compared to OCT as glaucoma worsens, which agrees with prior observations [21, 22, 41]. In one DIGS study, significant worsening of VD dropout rate, but not ganglion cell complex thinning rate, was found associated with VF MD worsening [21]. Furthermore, faster CD loss was found associated with VF progression in moderate-advanced glaucoma, while faster RNFL loss was not [22]. Interestingly, a cross sectional study showed that greater loss in macular perfusion density, instead of ganglion cell layer thickness, was associated with worse scan quality and female gender [43], which was not observed for ONH OCTA-only progressors in our study. In addition to the different study designs and methods, the number of OCT and OCTA-only progressors in the current study might not provide sufficient power to detect all differing characteristics. Future larger studies are needed to further investigate this.

Considering event-based analysis might allow earlier detection of progression with shorter follow-up data [20], it is clinically relevant to examine the progression detection capabilities of OCT and OCTA using this method, for which prior evidence is scarce. In this study, the percentage of OCT RNFL progressors identified was similar to that previously reported for trend-based analysis (15–20%) in a few studies, indicating a potentially comparable sensitivity of our current event-based approach [23, 44]. Meanwhile, a higher percentage of progressors (22–28%) has been reported for OCT guided progression analysis (GPA) [25, 45], although its specificity in healthy eyes was lower as it defines progression based on both event- and trend-based methods. The only other study examining progression detection by OCTA reported a lower rate of progression (13–15%), likely due to the inclusion of only advanced glaucoma eyes [26].

The selection of cut-offs is essential when applying the event-based method. However, due to the novelty of OCTA clinical application and event-based analysis, there is yet a gold standard cut-off. Thus, following prior studies, we chose the OCT/OCTA TRV as the cut-off, since a change exceeding TRV indicates possible glaucoma progression [26]. A strength of our study was the inclusion of a sensitivity analysis to evaluate how the different TRV cut-offs used to define progression influenced the results [20, 26, 35, 36]. Overall, consistent findings suggesting limited OCT-OCTA agreement were found in the sensitivity analyses, supporting the main results that OCT and OCTA may provide different information on progression detection. Of note, while all but one criterion found more eyes with progressive OCTA change than with progressive OCT change, the difference in the numbers of OCT and OCTA progressors varied across different cut-offs. This indicates the capabilities and sensitivity of OCT and OCTA to detect progression by event-based method may still be influenced by the cut-offs used, and a gold standard should be established to facilitate its clinical application.

With many more progressors identified by either OCT or OCTA than by VF, our results using event-based method are consistent with prior works suggesting OCT and OCTA may detect glaucomatous changes more sensitively and timely than VF in early glaucoma [3, 4, 24–26]. Besides the limited overall sample size, the smaller number of VF progressors identified might also be explained by the particular analysis that was conducted, the later presentation of functional change than structural change in some cases, and the relatively short follow-up period examined. In addition, compared to OCTA, OCT was slightly more strongly correlated with VF in both agreement and rate analyses. An overall slight-to-fair agreement between OCT and VF on progression detection has been shown previously [23–26], similar to our results. Although one study reported a stronger agreement of VF with OCTA than with OCT on progression detection [26], it was likely attributed to the OCT floor effect in advanced glaucoma [19, 41]. Notably, a better VF-OCT agreement than VF-OCTA agreement was also observed in our Supplementary analysis using trend-based method. Including mostly mild glaucoma eyes in this study, we provide additional evidence supporting the potentially complementary use of OCT and OCTA for functional association in the earlier stages of glaucoma [39, 46, 47].

There were several study limitations. First, a limited number of moderate-advanced glaucoma eyes were included, and whether our findings can be generalized to later-stage glaucoma is uncertain. However, we performed severity-stratified analyses and found similar results. Similarly, the generalizability of the results to population with different racial composition is uncertain. Second, to ensure a comparable OCT/OCTA specificity, the cut-offs used in the main analysis were selected from two earlier studies. Nonetheless, as mentioned earlier, consistent results were obtained in sensitivity analyses using other cut-offs, suggesting the reliability of the current findings. Third, glaucoma progression may also occur in a localized manner, which was not evaluated in our study using global parameters. The average follow-up of 4 years may also be insufficient to detect progression in some cases. Fourth, although OCT/OCTA demonstrate good reproducibility [5, 6, 21] and all images have gone through quality review to eliminate the effects of noises, presence of outlier measurements might still affect event-based analysis. To address this, serial OCT/OCTA measurements of progressors were double-checked to ensure no such outliers were included. Last, due to the relatively small sample size and retrospective nature of the study, our findings should be confirmed in future larger, prospective investigations.

In conclusion, there was limited agreement on the specific eyes identified as progressing by OCT and OCTA using event-based analysis, and OCT agreed more strongly with VF than does OCTA. Furthermore, both imaging tests also detected more progressors than VF. Our results suggest that OCT and OCTA provide valuable, yet different and complementary, information about glaucoma progression. Future studies with longer follow-up are needed to further confirm the relationship of OCTA changes with glaucomatous functional impairment.

Summary

What was known before

• Information about the capabilities of optical coherence tomography angiography (OCT-A) in glaucoma progression detection and how it compares to or agree with other instruments is scarce.

• The performance of structural parameters on event-based progression analysis remains largely unknown, although it may allow earlier detection of glaucoma progression.

What this study adds

• Using event-based analysis, OCT and OCTA showed limited agreement on progression detection, suggesting they may provide different yet complementary information about glaucomatous change.

• Both OCT and OCTA detected more progressors than VF, with OCT showing a slightly better agreement with VF.

Author contributions

JHW: study design, data analysis, results interpretation, production of tables and figures, and drafting and critical revision of the paper. SM: study design, results interpretation, drafting and critical revision of the paper, and providing research resources and fundings. TN: study design, data analysis, results interpretation, and critical revision of the paper. GM: results interpretation and critical revision of the paper. LZ: subject recruitment, study design, results interpretation, critical revision of the paper, and providing research resources and fundings. RNW: subject recruitment, results interpretation, critical revision of the paper, providing research resources and fundings, and taking full responsibility for the study as the guarantor.

Funding

This work is supported by National Institutes of Health/National Eye Institute Grants (R01EY034148, R01EY029058, R01EY011008, R01EY019869, R01EY027510, R01EY026574, R01EY018926, P30EY022589); University of California Tobacco Related Disease Research Program (T31IP1511), and an unrestricted grant from Research to Prevent Blindness (New York, NY). The sponsor or funding organization had no role in the design or conduct of this research.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Competing interests

SM reported grants from the National Eye Institute. TN is a consultant of Topcon. LZ reported grants from the National Eye Institute; grants from Heidelberg Engineering and nonfinancial support from Carl Zeiss Meditec, Optovue, Heidelberg Engineering, and Topcon. Consultant of Abbvie, AISight Health and Topcon and patents from Carl Zeiss Meditec. RNW is a consultant of Abbvie, Aerie Pharmaceuticals, Allergan, Amydis, Equinox, Eyenovia, Iantrek, IOPtic, Implandata, Nicox, and Topcon. RNW reported nonfinancial support from Heidelberg Engineering, Carl Zeiss Meditec, Konan Medical, Optovue, Centervue, and Topcon; grants from the National Eye Institute and National Institute of Minority Health Disparities, patents from Toromedes, Carl Zeiss Meditec to UCSD; all outside the submitted work. No other disclosures were reported.

Supplementary information

The online version contains supplementary material available at 10.1038/s41433-023-02817-0.