Retinal Blood Flow Decreases after Treatment with Bevacizumab for Retinopathy of Prematurity

Abstract

Purpose

To compare total retinal blood flow (TRBF) rates before and after retinopathy of prematurity (ROP) treatment with intravitreal bevacizumab using laser speckle contrast imaging (LSCI).

Design

A prospective cohort study.

Participants

Twenty-five eyes from 14 premature infants in the neonatal intensive care unit receiving intravitreal bevacizumab for treatment-requiring ROP.

Methods

Total retinal blood flow was measured using LSCI longitudinally before and after bevacizumab treatment. Subject characteristics and clinical ROP features, including the need for ROP retreatment, were included in regression analysis using generalized estimating equations to account for 2 eyes per subject and longitudinal measures over time.

Main Outcome Measures

The main outcome measure was TRBF, which includes components of peak, mean, and dip over the cardiac cycle.

Results

Before ROP treatment, subjects had a peak TRBF of 11.1 ± 2.9 a.u. compared to 8.6 ± 1.8 a.u. after treatment (mean difference = 2.5 a.u., P < 0.0001). Among eyes that required ROP retreatment earlier (<10 weeks) after initial treatment, the posttreatment peak TRBF was 9.0 ± 1.5 a.u., compared to 7.3 ± 2.2 a.u. for eyes that did not require retreatment in the first 10 weeks after initial bevacizumab injection (mean difference = 1.7 a.u., P = 0.01). Peak TRBF decreased over time after bevacizumab treatment (β = −0.1 a.u./week, P = 0.004).

Conclusions

We observed lower TRBF after treatment with intravitreal bevacizumab.

Financial Disclosure(s)

Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Retinal vascular changes are widely accepted as a critical examination feature indicating the need for treatment of retinopathy of prematurity (ROP).^1,2 Dilation and tortuosity of retinal blood vessels observed in zone I is termed “plus disease,” or “pre-plus” when present in milder form. Recent literature has refined our understanding of vascular changes on a continuum, which can be scaled using reference images such as the “vascular severity score,”^3,4 developed based on artificial intelligence, or plus score (p-score),⁵ developed based on clinical assessment. Reversal of pathologic vascular changes after ROP treatment is well described.^6,7 In addition to retinal blood vessel monitoring, ROP staging also plays a critical role in monitoring disease progression; however, stage alone dictates treatment requirement only if stage 3 is present in zone I.²

In order to quantify vascular changes, clinically available quantitative blood flow measurement devices such as color Doppler ultrasound imaging and laser speckle contrast imaging (LSCI) have been considered for use during ROP screening examinations.^8,9 Intravenous fluorescein angiography can also quantify retinal blood flow, offering transit time as a single, albeit useful, quantitative measure.¹⁰ Additionally, OCT angiography and adaptive optics scanning laser ophthalmoscopy can assess vascular architecture, but not blood flow dynamics.^11,12 Laser speckle contrast imaging offers the advantage of noncontact, nonperturbative, optical mapping of the entire retinal vascular network at high resolution as well as assessment of local, regional, and blood flow velocities.

Previous literature suggests a significant decline in blood flow after intravitreal injections of anti-VEGF for treatment of ROP using both color Doppler ultrasound imaging¹³ and LSCI.¹⁴ However, past studies were limited by small sample size and lacked longitudinal follow-up after anti-VEGF treatment, even though anti-VEGF–treated eyes are at known risk for reactivation and persistent avascular retina, which may impact the course of disease.¹⁵

The purpose of this study was to assess total retinal blood flow (TRBF) using LSCI longitudinally before and after ROP treatment with intravitreal anti-VEGF bevacizumab. We hypothesized that TRBF will be associated with a decrease in TRBF. Furthermore, we hypothesized that eyes with reactivation will have higher TRBF at the time of retreatment compared with eyes without reactivation.

Methods

This study was conducted as part of an ongoing prospective study at the University of Maryland Neonatal Intensive Care Unit in Baltimore, Maryland. The study was approved by our accredited institutional review board (HP-00088705) in adherence with US Health Insurance Portability and Accountability Act guidelines and the Declaration of Helsinki. The study was conducted from January 2021 to December 2024.

Participants

Informed consent was obtained from the parent of each subject after explanation of the nature and possible consequences of participating in the study. Subjects included 14 infants meeting the following inclusion criteria: (1) gestational age ≤30 weeks, or (2) birth weight ≤1500 g, and (3) treatment for ROP was indicated based on the presence of type 1 ROP, namely: (a) plus disease or stage 3 in zone I, (b) plus disease with stage 2 to 3 in zone II,² or (c) the presence of concerning features, namely pre-plus disease with thick or extensive stage 3 membranes with anteroposterior traction concerning for progression to stage 4 ROP.16, 17, 18, 19, 20 Infants with concurrent media opacity unrelated to ROP (i.e., cataract) or genetic syndromes were excluded. Prior approval by neonatology was obtained for all enrolled subjects. Data were collected from subjects before and after treatment with bevacizumab for comparison. There was no external control group.

ROP Imaging and Examinations

One hour before ROP examination, infants' pupils were dilated with topical mydriatic (cyclopentolate 0.2% and phenylephrine 1%) drops. One drop of 0.5% proparacaine was placed in each eye immediately before the eye examination. Positioning was conducted as previously described.⁸

Laser speckle contrast imaging was performed using the investigational XyCAM CRE prototype system (Vasoptic Medical, Inc). The XyCAM CRE is a custom-modified version of the US Food and Drug Administration-cleared XyCAM RI System (Vasoptic Medical, Inc) wherein the imaging unit is mounted on a mobile articulating arm to facilitate retinal imaging in supine infants in the neonatal intensive care unit setting. Details of the imaging protocol have been previously described.⁸ Imaging was performed by 2 experienced imagers (J.L.A. and K.W.) with 6 months of experience and training with LSCI before initiating recruitment for this study. Immediately after LSCI, a standard ROP screening examination was performed using binocular indirect ophthalmoscopy followed by retinal photography using the RetCam 3 Shuttle (Natus Medical Inc). After ophthalmoscopy and photography, the examining physician documented the zone, stage, and p-score. RetCam fundus photos were used to verify zone using a ruler, and p-score was verified using side-by-side image comparison to the standard 9-level p-score reference photo.⁵

Clinical Data

Clinical data were collected retrospectively from documentation in the electronic medical record and also separately graded in masked fashion. The terms “plus,” “pre-plus,” and “no plus” were analyzed as categorical variables and as continuous variables using p-scores. Plus scores, zone, and stage were assigned based on fundus photos and compared to clinical documentation. Plus scores ≥7 were classified as plus disease, p-scores of 4 to 6 were classified as pre-plus, and p-scores ≤3 were classified as no plus.⁵ For statistical analysis, zone IIp was entered as “1.5” and zone IIa was entered as “2.0.” All other zones were classified according to their numerical whole number. RetCam images were used to verify all clinical documentation, and no discrepancies were noted in zoning or p-score assignment between our 2 graders (J.L.A. and M.R.L.) (κ = 1.0).

Image Analysis

The XyCAM CRE LSCI outputs a time-stack of blood flow velocity index (BFVi) maps of the posterior retinal vasculature. Obtained data contains a time sequence of BFVi values at each pixel. Pseudo-color heatmaps (Fig 1) are used to visualize the spatial distribution of BFVi, while time plots reveal the trends of blood flow pulsations over multiple cardiac cycles. Within specific regions, BFVi-related metrics are also provided in tabular format for subsequent analysis.

Figure 1.

Illustration of the blood flow variables based on 2 methods of image analysis. A, Total retinal blood flow-vessels have a region of interest drawn based on fundus photo overlay to outline the optic nerve, followed by a freehand outline of the vessels within the encircled optic nerve. Peak and dip are averaged across many cardiac cycles. B, Total retinal blood flow-optic nerve head has region of interest as a standard ellipse within the border of the optic nerve. Peak is a single value; dip is the average of the 2 troughs adjacent to the peak. TRBF = total retinal blood flow; TTF = time to fall; TTR = time to rise; VFI = volumetric fall index; VRI = volumetric rise index.

Total retinal blood flow was determined using 2 methods. Both methods were centered on the optic disc region in order to capture the entire retinal blood flow entering and exiting the eye. The purpose of the dual analysis was to determine if a faster but possibly less precise approach, TRBF-optic nerve head (ONH), could provide a similar dataset to the more precise but time-consuming approach, TRBF-vessels (V). Image analysts were masked to the clinical status of the subject.

In the first method, after image registration for motion, the ONH border was outlined using the RetCam image overlay as a guide. Next, the ONH was further segmented to include only visually discriminable vessels within the ONH boundary. Vessels within the encircled nerve were freehand outlined as the region of interest (ROI). Blood flow in the ROI was calculated using the XyCAM CRE automated software.²¹ Data collected from this ROI were analyzed over multiple cardiac cycles up to 6 seconds in duration. Associated data were termed “TRBF-vessels (V),” and the dataset included peak, mean, dip, time to rise, time to fall, volumetric rise index (VRI), and volumetric fall index (VFI). The peak and dip were average values from the entire duration of image acquisition, which generally included >5 complete cardiac cycles (Fig 2A).

Figure 2.

Representative laser speckle contrast images from 4 subjects. Images were selected at peak flow during the cardiac cycle for 1 eye in each of 4 subjects with color flow mapping with matched scale on dynamic display setting for image analysis using a predetermined range (lower limit of 3 to upper limit of 9 BFVi [a.u.] range). Subject 1 (A) immediately prior to intravitreal bevacizumab and (B) 2 weeks after intravitreal bevacizumab. Subject 2 (C) immediately prior to intravitreal bevacizumab and (D) 4 weeks after intravitreal bevacizumab. Subject 3 (E) immediately prior to intravitreal bevacizumab and (F) 4 weeks after intravitreal bevacizumab. Subject 4 (G) immediately prior to intravitreal bevacizumab and (H) 6 weeks after intravitreal bevacizumab. The visual difference between the pre- and postimage pairs becomes more apparent as the time interval between pre- and posttreatment increases. BFVi = blood flow velocity index.

In the second method, all images used a reference standard elliptical ROI with semiaxis lengths of 162.71 and 172.41 pixels, respectively, which fully encircled the optic nerve in 100% of the cohort. After 6-second image acquisition, 1 single frame with optimal focus, centration, and no motion was selected at peak flow on the cardiac cycle tracing. The 2 adjacent dips for the peak were selected by default for this ROI. Associated data were termed “TRBF-optic nerve head (ONH)” and the dataset included peak, mean, and dip. The dip for TRBF-ONH was the average of the 2 dips on either side of the selected peak point (Fig 2B).

Statistical Analysis

Variables evaluated included age at imaging, age at bevacizumab injection, age at retreatment after bevacizumab injection, gestational age, current weight, birth weight, p-score, zone, stage, and plus disease status. The main outcome measure was TRBF (examined as both TRBF-V and TRBF-ONH), which includes components of peak, mean, and dip over the cardiac cycle. Total retinal blood flow-V also includes time to rise, time to fall, VRI, and VFI (Fig 2). Longitudinal TRBF was collected from both eyes, at multiple time points, both before and after treatment, for most subjects. Variables were compared before and after intravitreal bevacizumab using a nonparametric test of significance (Wilcoxon test) and regression models with adjustment for inclusion of 2 eyes per subject and longitudinal measures in the same eye over time (generalized estimating equations).

The time course and severity of reactivation were also analyzed as outcomes by specifically evaluating the stage of ROP at retreatment, and the time between bevacizumab initial treatment and retreatment.

Univariate analysis assessed participant demographics, characteristics, clinical features, and TRBF. Next, the association between treatment and TRBF was examined using the Wilcoxon nonparametric test to compare TRBF before and after treatment as a global comparison. Multivariable regression with generalized estimating equations was also used to compare TRBF before and after treatment. Resulting P values were calculated based on Wald tests from the mixed effects model. Subject characteristics and clinical ROP features including need for ROP retreatment were included in regression analysis. Generalized estimating equations were used to account for 2 eyes per subject and longitudinal measures over time to compare TRBF before and after treatments. SAS v9.4 (Statistical Analysis Software, North Carolina State University) was used to perform all statistical analysis. To determine statistical significance, 2-sided P < 0.05 was utilized in the analysis. For models with multiple variables demonstrating P < 0.05, exact P values were reported.

Results

Demographics

This cohort included 14 participants (25 eyes) with ROP who underwent ROP treatment with intravitreal bevacizumab between 2021 and 2024 and were followed for up to 36 months. Among the 14 participants, 6 were females (43%), 11 were non-White (79%), and 1 was of Hispanic ethnicity (7%). Birth weight in the cohort was 664 ± 193 (range 485–1219 g), and gestational age was 24.4 ± 1.2 (range 23.0–27.7 weeks). Subjects were imaged between the age of 8.7 and 43 weeks of life (33–67 postmenstrual weeks) (Table 1).

Table 1.

Subject Demographics, Birth Characteristics, Clinical Features n = 14 Subjects Total, 25 Eyes

Female sex	N = 6 (43%)
Non-White race∗	N = 11 (79%)
Hispanic ethnicity	N = 1 (7%)
BW (grams)	664 ± 193 (485–1219)
GA (wks)	24.4 ± 1.2 (23–27.7)
PMA at bevacizumab	35.0 ± 2.1 (31.3–39.8)
PMA at retreatment	45.0 ± 7.4 (37.7–66.9)
Indication for bevacizumab	Extensive and progressive stage 3 with pre-plus disease (n = 1) Zone 1 stage 3 or plus disease (n = 13)
Indication for retreatment after bevacizumab	PAR only with stage 1 at retreatment (n = 4) PAR with reactivated stage 2 (n = 6) PAR with reactivated stage 3 (n = 4)

BW = birth weight; GA = gestational age; PAR = persistent avascular retina; PMA = postmenstrual age.

^∗Due to the small proportion of the cohort identified as Asian, other, or >1.

Treatment Timing and Clinical Features

Subjects received intravitreal bevacizumab treatment between the age of 6.9 and 16.9 weeks of life (31.3–39.8 postmenstrual weeks). All subjects had stage 3 at the time of bevacizumab treatment. Thirteen (n = 13) subjects had plus disease (p-score >7), and 1 (n = 1) subject had pre-plus disease (p-score between 5 and 7). Subjects received retreatment postbevacizumab with either repeat bevacizumab (n = 1) or laser photocoagulation (n = 13) at the age of 13.7 to 43 weeks of life (37.7–66.9 postmenstrual weeks) (Table 1). Retreatment was provided for all subjects 9.2 ± 6.0 weeks after initial intravitreal bevacizumab treatment (range 4.6–32 weeks). All subjects had persistent avascular retina after intravitreal bevacizumab. At retreatment, 4 subjects had stage 1, 6 subjects had stage 2, and 4 subjects had stage 3 (Table 1).

TRBF after Treatment

Longitudinal TRBF was collected at multiple time points for most subjects, ranging from 2 weeks prior to intravitreal bevacizumab to 32 weeks after bevacizumab (mean imaging time point 5.4 ± 7.3 weeks after bevacizumab).

After bevacizumab intravitreal injections, zone increased (P < 0.0001), stage decreased (P < 0.0001), and p-score decreased (P < 0.0001). The peak, mean, and dip TRBF-ONH and TRBF-V also decreased significantly after treatment. After treatment, TRBF-V VFI also decreased significantly (P = 0.0003). Before ROP treatment, subjects had a peak TRBF-V of 11.1 ± 2.9 a.u. compared to 8.6 ± 1.8 a.u. after treatment (mean difference = 2.5 a.u., P < 0.0001) (Table 2). Granular clinical and TRBF results over specific time intervals after intravitreal bevacizumab and before retreatment are provided in Table S1 (available at www.ophthalmologyscience.org). Peak TRBF-V decreased over time after bevacizumab treatment (β = −0.1 a.u./week, P = 0.004).

Table 2.

Clinical and Blood Flow Difference Comparing Immediately before and after Intravitreal Bevacizumab (Granular Data Are Available in Table S3, available at www.ophthalmologyscience.org)

	Before Bevacizumab	After Bevacizumab	P Value∗	P Value†
Zone	1.50 ± 0.44 (1.0–2.0)	1.95 ± 0.15 (1.5–2.0)	0.004	<0.0001
Stage	2.90 ± 0.30 (2.0–3.0)	0.86 ± 0.86 (0–3.0)	<0.0001	<0.0001
Plus score	7.5 ± 1.00 (6.0–9.0)	3.68 ± 1.40	<0.0001	<0.0001
TRBF-ONH peak	8.49 ± 2.87	6.26 ± 1.39	0.02	<0.0001
TRBF-ONH mean	7.98 ± 2.84	5.52 ± 1.39	0.04	<0.0001
TRBF-ONH dip	6.97 ± 2.76	4.79 ± 1.45	0.03	<0.0001
TRBF-V peak	11.10 ± 2.87	8.64 ± 1.79	0.02	<0.0001
TRBF-V mean	9.64 ± 2.90	7.59 ± 1.85	0.05	0.0004
TRBF-V dip	8.19 ± 2.98	6.56 ± 1.98	0.2	0.004
TRBF-V TTR	241.28 ± 102.04	234.43 ± 111.11	0.9	0.9
TRBF-V TTF	279.33 ± 106.10	228.56 ± 110.53	0.01	0.2
TRBF-V VRI	2.30 ± 1.39	1.76 ± 0.97	0.2	0.08
TRBF-V VFI	2.76 ± 1.37	1.73 ± 0.81	0.03	0.0003

ONH = optic nerve head; TRBF = total retinal blood flow; TTF = time to fall; TTR = time to rise; V = vessels; VFI = volumetric fall index; VRI = volumetric rise index.

^∗Based on the nonparametric Wilcoxon test.

^†Based on regression with generalized estimating equations.

Among eyes that required ROP retreatment earlier (<10 weeks) after initial treatment, the posttreatment peak TRBF-V was 9.0 ± 1.5 a.u., compared to 7.3 ± 2.2 a.u. for eyes that did not require retreatment in the first 10 weeks after initial bevacizumab injection (mean difference = 1.7 a.u., P = 0.01) (Table 3).

Table 3.

Clinical and Blood Flow Parameters after Bevacizumab Treatment Comparing Mean ± SD between Subjects Requiring Retreatment within 10 Wks after Bevacizumab and Those Retreated >10 Wks after Bevacizumab

Clinical and Blood Flow Parameters	Retreatment <10 Wks	Retreatment >10 Wks	P Value∗
	Mean ± SD	Mean ± SD
Zone	1.94 ± 0.16	2.0 ± 0	0.6
Stage	0.81 ± 0.87	1.10 ± 0.84	0.8
Plus score	3.52 ± 1.31	5.25 ± 1.71	0.5
TRBF-ONH peak	6.34 ± 1.36	5.75 ± 1.58	0.4
TRBF-ONH mean	5.63 ± 1.38	4.82 ± 1.39	0.2
TRBF-ONH dip	4.93 ± 1.43	3.88 ± 1.38	0.1
TRBF-V peak	8.95 ± 1.54	7.28 ± 2.23	0.01
TRBF-V mean	7.96 ± 1.60	5.99 ± 2.09	0.006
TRBF-V dip	6.98 ± 1.72	4.71 ± 2.06	0.003
TRBF-V TTR	238.48 ± 122.52	216.60 ± 23.37	0.6
TRBF-V TTF	236.90 ± 120.43	191.88 ± 28.70	0.3
TRBF-V VRI	1.87 ± 1.01	1.20 ± 0.41	0.07
TRBF-V VFI	1.86 ± 0.81	1.15 ± 0.51	0.02

ONH = optic nerve head; SD = standard deviation; TRBF = total retinal blood flow; TTF = time to fall; TTR = time to rise; V = vessels; VFI = volumetric fall index; VRI = volumetric rise index.

^∗Regression with generalized estimating equations.

Eyes that had reactivation to stage 3 at the time of retreatment had higher blood flow (9.7) than those with stage 2 or less (7.5 ± 2.0), but that difference was not significant (P = 0.4) (Table S2, available at www.ophthalmologyscience.org). Granular TRBF results by stage at retreatment are provided in Table S3 (available at www.ophthalmologyscience.org).

Discussion

We found that TRBF decreased after intravitreal bevacizumab. The significance of this finding is particularly meaningful given existing literature demonstrating that TRBF generally increases over time in an ROP cohort that does not require treatment.⁸

We also found that infants who required retreatment after intravitreal bevacizumab within the first 10 weeks after initial treatment had higher TRBF than subjects who were retreated >10 weeks after initial treatment. Eyes with stage 3 at retreatment had higher blood flow than those with stage 2 or less, but that difference was not significant. These data consistently show a pattern of more severe ROP and higher TRBF. Our results are consistent with previous findings that intravitreal anti-VEGF lowers retinal blood flow velocities in eyes with treatment-requiring ROP.

Peak TRBF-V was our primary variable of interest. We found results were very consistent using a similar, but much faster, image analysis approach, TRBF-ONH. While peak, mean, and dip TRBF are associated metrics, they have the potential to highlight subtle differences in vascular physiology between the systolic and diastolic phases of microvascular flow and are therefore reported separately. Time-to-rise and time to fall are associated with heart rate but not flow so these variables were not significantly associated with any clinical circumstances. Volumetric fall index, but not VRI, was significantly lower after treatment.

Volumetric rise index and VFI are assessed as the area under the rising and falling portions of the blood flow pulse waveform for an ROI of the retina. Volumetric rise index and VFI represent the amount of blood flowing through the retinal region analyzed during the rising and falling time intervals of the heartbeat. Thus, unlike TRBF, these metrics capture the combined effect of blood flow, heart rate, and the ability of the vasculature in the analyzed region to passively constrict and dilate when the blood flow pulse passes through it. In the context of ROP, one might expect VFI to be more significantly impacted by treatment than VRI because while vessel compliance in the rising portion is caused by blood pressure, the falling portion is passive and depends upon the features of the local vasculature (e.g., vessel caliber or compliance). VEGF causes vessel dilation when VEGF binds to VEGFR2 receptors on endothelial cells, triggering a signaling cascade that results in release of nitric oxide, a potent vasodilator.^22,23 Therefore, high VEGF levels contribute to vessel relaxation, with venous caliber changes more prominent than arterial changes in plus disease. Bevacizumab neutralizes VEGF and blocks signal transduction through both VEGFR1 and VEGFR2 receptors, inhibiting many downstream effects, including nitric oxide production.^24,25 Venules also have distensible potential, while arteries do not. These reasons suggest benefits of assessing VRI and VFI separately, rather than collectively, in ROP.

Previous studies in smaller cohorts used various technologies to measure blood flow.^13,14,26 Using LSCI, Matsumoto et al found a 46% decrease in blood flow after treatment with bevacizumab (n = 8 eyes, P = 0.0007).¹⁴ Our findings, based on a larger cohort (n = 24 eyes), are consistent with Matsumoto et al's. We found a decrease of 23% after treatment with bevacizumab (P < 0.0001) (Table 4). While the trends were similar between our findings and Matsumoto et al's, direct numeric comparison is not possible because LSCI data offers an index for blood flow rather than an absolute value. Previous work using standard Doppler imaging demonstrated a 32% decrease in blood flow after bevacizumab (P = 0.001).¹³ Similarly, Sukgen et al found a 42% decrease in blood flow after ROP treatment with aflibercept (P < 0.001).²⁶ Overall evidence suggests that ROP treatment with anti-VEGF is associated with significant ocular blood flow reduction. Animal studies have likewise demonstrated a significant association between retinal blood flow, retinal VEGF,²⁷ and avascular retina.

Table 4.

Comparison to Previously Published Data on Ocular or Retinal Blood Flow after Intravitreal Anti-VEGF (Bevacizumab or Aflibercept^∗)

	N (Eyes)	Parameter	Before Anti-VEGF	After Anti-VEGF	% Decrease	P Value
LSCI, current study	24	TRBFpeak (a.u.)	11.1 ± 2.9	8.6 ± 1.8	23%	<0.0001
LSCI, Matsumoto et al14	8	MBR-A (a.u.)	14.0 ± 3.5	7.5 ± 2.3	46%	0.0007
Doppler, Mohr et al13	21	PASV (cm/s)	13.6 ± 6.2	9.3 ± 3.3	32%	0.002
Doppler, Sukgen et al∗26	29	DASV (cm/s)	4.8 ± 1.5	2.8 ± 0.8	42%	<0.001

DASV = dip arterial systolic velocity; LSCI = laser speckle contrast imaging, peak; MBR-A = mean blur rate of all vessels; PASV = peak arterial systolic velocity; TRBF_peak = total retinal blood flow.

^∗The use of the drug aflibercept in the publication by Sukgen et al. All other papers were based on use of the drug bevacizumab.

In the current study, subgroup analysis was underpowered. In order to frame our findings, we performed a post hoc power analysis. For TRBF, a normally distributed continuous response variable, among different subgroups of ROP (e.g., patients with zone I vs. zone IIp) or different subgroups of treatment response (e.g., patients with early reactivation after bevacizumab vs. more durable response to bevacizumab), we based our power analysis on the current study's standard deviation of 2.0, difference between groups of 2.3 (Table S2), power of 0.80, and type I error probability associated with test of the null hypothesis of 0.05. We found that we will need to enroll 14 subjects in each group (or 28 subjects total) to be able to reject the null hypothesis that the population means of the 2 groups are equal. Future studies are needed to offer additional insight among subgroups and to evaluate the predictive ability of multiple LSCI metrics. A key controversy in ROP care is the management of reactivation and the timing of retreatment after intravitreal anti-VEGF. In this study, eyes that required ROP retreatment earlier (<10 weeks) after initial treatment had 1.7 a.u. higher peak TRBF-V than eyes that did not require retreatment in the first 10 weeks after initial bevacizumab injection. Further study is needed to determine if higher peak TRBF-V can predict the need for earlier retreatment. Likewise, further study may also help understand if increased TRBF-V can predict need for initial treatment.

Increased blood flow in plus disease is caused by a combination of factors. Neovascularization increases the total length of the vascular network, while local vessel dilation increases the volume carried per vessel. Evolution of plus disease likely therefore has 2 phases, including immediate local effects of vessel dilation and delayed effects of vessel propagation and remodeling. There may be synergistic effects whereby increased flow amplifies VEGF signaling, which may explain the dramatic and often rapid progression of plus disease. For the purpose of ROP monitoring, the causal relationship is likely inconsequential, but insights offered by real-time blood flow monitoring may help identify clinically relevant features before blood vessel appearance changes.

Strengths of this study include a larger cohort compared with prior studies, with longitudinal follow up to include the interval of reactivation risk after intravitreal bevacizumab. All subjects were followed until postbevacizumab retreatment, with some subjects followed for years after treatment. We used statistical approaches to control for 2 eyes per subject and longitudinal measures over time.

Limitations

This study had a limited but sufficient sample size. Subjects were imaged at variable time points ranging from 2 weeks prior to intravitreal bevacizumab to 32 weeks after intravitreal bevacizumab. Some infants were enrolled in the study after their bevacizumab treatment, based on the availability of parents to provide consent. Other infants were discharged or transferred from our hospital before longitudinal follow-up was complete. Fortunately, statistical tools (generalized estimating equations) allowed us to account for the different follow-up schedules and different clinical features of each eye, including accounting for within-subject variation. None of the reasons for missing time points are expected to significantly bias our results.

Conclusions

We demonstrate the feasibility of blood flow measurement alongside validated metrics like stage and p-score to monitor the progression of ROP. Using LSCI-based blood flow measurements, we observed a significant reduction in TRBF after ROP treatment with intravitreal bevacizumab, with a corresponding reduction in ROP stage and p-score. Blood flow reduction was significantly less pronounced among eyes that required early retreatment for reactivation. Blood flow may offer an objective biomarker for ROP treatment response.

Manuscript no. XOPS-D-25-00157.