Implementation of OCT in Retinopathy of Prematurity Screening

Abstract

Purpose of review:

In this review, we explore the investigational applications of optical coherence tomography (OCT) in retinopathy of prematurity (ROP), the insights they have delivered thus far, and key milestones for its integration into the standard of care.

Recent findings:

While OCT has been widely integrated into clinical management of common retinal diseases, its use in pediatric contexts has been undermined by limitations in ergonomics, image acquisition time, and field of view. Recently, investigational handheld OCT devices have been reported with advancements including ultra-widefield view, non-contact use, and high-speed image capture permitting real-time en face visualization. These developments are compelling for OCT as a more objective alternative with reduced neonatal stress compared to indirect ophthalmoscopy and/or fundus photography as a means of classifying and monitoring ROP.

Summary:

OCT may become a viable modality in management of ROP. Ongoing innovation surrounding handheld devices should aim to optimize patient comfort and image resolution in the retinal periphery. Future clinical investigations may seek to objectively characterize features of peripheral stage and explore novel biomarkers of disease activity.

Introduction

Retinopathy of prematurity (ROP) is a leading cause of childhood blindness. While nearly always avoidable with prompt diagnosis and treatment, preventable blindness due to ROP occurs in approximately 20,000–50,000 infants annually due to missed, delayed, or incorrect diagnoses.[1–6] The current standard of care for monitoring involves weekly or bi-weekly dilated binocular indirect ophthalmoscopy with scleral depression at the bedside, or review of ultra-widefield digital fundus imaging via telemedicine. Since 1984, the International Classification of Retinopathy of Prematurity (ICROP) has established guidelines for interpreting clinical findings from these exams to classify disease according to features including zone and stage of retinopathy, and presence of plus disease.[7] Despite longstanding acceptance of these recommendations, classification of ROP remains highly subjective and prone to interobserver variability, undermining standardization of management paradigms across providers and institutions.[8-12**] This observation has driven exploration of alternative approaches for ROP diagnosis. Of these, optical coherence tomography (OCT) has shown promise as a means of observing pathology and identifying novel biomarkers of disease with higher resolution and objectivity than ophthalmoscopic exam alone.[13-15*] While OCT has been adopted as standard of care in other leading causes of blindness including age-related macular degeneration and diabetic retinopathy, it has yet to be broadly implemented into clinical practice in ROP after over a decade of investigation into its utility within this disease model.[16] In this review, we will explore the use of OCT in ROP with regard to its clinical advantages, the insights it has delivered thus far, and the potential pathways to its adoption into the standard of care.

The Advantages of OCT in ROP Diagnosis

The current standard of care for in-person screening and monitoring of ROP remains bedside ophthalmoscopy in the neonatal intensive care unit (NICU), which traditionally involves topical anesthesia, pharmacologic pupillary dilation, insertion of an eyelid speculum and examination with the bright light of binocular indirect ophthalmoscopy aided by scleral depression.[17] By this means, ROP zone, stage and plus disease are subjectively assessed. An accepted alternative is the fundus photography-based telemedicine approach, which still requires the topical anesthesia, pupillary dilation, eyelid speculum and a contact based camera. These approaches are imperfect for two reasons: 1) the ROP examination is stressful, adding medical risks to fragile neonates, 2) both ophthalmoscopic examination and telemedicine interpretation are limited by subjectivity and inter-observer diagnostic variability. OCT-aided visualization has been shown to offer a viable alternative to overcome these challenges.

The ROP examination adds medical risks to fragile neonates

While necessary to identify potential signs of vision-threatening conditions at the bedside, ophthalmoscopic exams are physiologically stressful for neonates, producing changes in heart rate, and apnea during and after examination.[18, 19] Handheld WDFI devices used in bedside and telemedicine exams for ROP require similar preparation including bright light exposure which is physiologically stimulating.[20, 21] In addition to stress during the exam, cycloplegic medication for pharmacologic dilation may cause systemic complications like impaired gut motility, require time for dilation, and incur cost of additional medication.[22] Handheld retinal imaging devices capable of noncontact photography may obviate the need for topical anesthesia, but they have lower field of view.[23**, 24]

The use of OCT in monitoring of pediatric vitreoretinal disease has been historically challenging, particularly in awake NICU patients.[25] Recently, handheld spectral-domain OCT devices can provide high-resolution en face and cross-sectional retinal and optic nerve images in supine NICU infants that are inherently unable to position adequately into tabletop devices.[15*, 16, 26–30] The use of non-visible, infrared light in these devices eliminates the stress added by the bright, visible light of binocular indirect ophthalmoscopy. Further, some handheld devices have been designed to capture images without direct contact with the cornea or use of an eyelid speculum.[16, 23**] Using noncontact handheld OCT, Mangalesh et al. demonstrated lower behavioral and cardiovascular markers of stress during bedside ROP exam compared to binocular indirect ophthalmoscopy, despite longer examination time.[23**] Speed of acquisition is another challenge being addressed to optimize comfort in awake patients and quality in the setting of photographer hand tremor and subject motion during exams. Thus, the introduction of handheld OCT pediatric bedside exams has accelerated imaging speeds from tens of kilohertz (kH) to more than one megahertz (MHz), reducing examination time to less than that of conventional binocular indirect ophthalmoscopy, and even making OCT angiography (OCTA) possible.[31] Innovation in handheld OCT devices has introduced a modality with potential to reduce the duration and physiologic stress of conventional bedside exams. Looking forward, development of nonmydriatic image acquisition as an additional feature could eliminate the need for pharmacologic dilation and reduce physiologic risks of these topical medications.

ROP grading is subjective and prone to interobserver variability

The framework for current treatment guidelines in ROP is derived from the following classification schema provided by ICROP in 1984: the zone (location of border of non-vascularized retina in relation to pre-determined borders), stage (degree of vascular abnormality at the avascular border on a 0-5 scale), and plus disease (degree of arterial tortuosity and venous dilation in the posterior retina according to original standard photographs).[32, 33] In 2005, ICROP was updated with the introduction an intermediate classification of “pre-plus” disease, introducing the concept that plus disease manifests on a spectrum, as opposed to a binary (i.e. “plus” or “no plus”) distribution.[32] In its most recent update in 2021, the new ICROP formally acknowledged the manifestation of plus disease along a continuous spectrum.[7] While these increasingly nuanced recommendations for classification more accurately reflect the profile and natural history of disease, inter-observer diagnosis remains discordant. Clinicians have been shown to disagree on all ICROP classifications of zone, stage and plus with both binocular indirect ophthalmoscopy and widefield digital fundus imaging.[8, 9, 34–36] Additionally, clinicians disagree on the relative relevance of various clinical features to classification and prognosis.[37, 38] This observed disagreement undermines consistency in diagnostic standards and may lead to delayed or missed diagnosis of treatment-requiring disease.

OCT’s ability to capture ultra-widefield high resolution images of the retina in cross section could provide a solution to the subjectivity of clinical examinations in two ways: 1) quantification of established components of disease classification, and 2) identification of new biomarkers of disease not observable by ophthalmoscopy.

Quantification of novel components of disease classification

While plus disease has been the first disease trait to be formally recognized by ICROP to manifest along a continuous spectrum, we posit that all 3 components of the ICROP classification (zone, stage, and plus) reflect continuous findings that are grouped into ordinal categories based on the clinical examination. “Zone 1” ROP has a worse prognosis (higher risk of disease progression requiring treatment, and retinal detachment), and the ICROP recently recommended further subdividing zone II into posterior and anterior zone II for the same reason. However, clinician determination of zone is subjective and inconsistent, and fundamentally reflects a clinical variable (area of avascular retina) that is inherently continuous. Recently, a handheld ultra-widefield OCT device capable of real time, en face visualization of an up to 140° field of view was demonstrated to provide higher contrast of the vascular-avascular border than binocular indirect ophthalmoscopy or widefield digital fundus imaging, including visualization to the ora serrata without scleral depression (Figure 1).[14*, 15*, 39*] Longitudinal characterization of the vascular-avascular border with such a device could facilitate measurement of the area of vascularized retina, and permit objective assessment of this feature as a continuous variable.

Figure 1:

Representative ultra-widefield en face ROP fundus images obtained via prototype handheld OCT developed at Oregon Health & Science University [14, 65]. (A) Simultaneous visualization of the posterior pole and ora serrata is possible without scleral depression (yellow arrowheads). (B) Subtle features of stage at the vascular-avascular border are visible in 360 degrees.

In addition to en face assessment of zone, OCT B-scans permit the examiner to assess three-dimensional relationships to inform classification of stage.[40] The ability to reconstruct the topographic anatomy of the retina reveals fine details like popcorn neovascularization and the continuity of stage progression as extraretinal fibrovascular proliferation coalesces into characteristic stage 3 appearance.[15*] This modality not only allows examiners to monitor known features of staging with finer detail but could also reveal new biomarkers for diagnosis and progression. In a longitudinal study, Nguyen et al. found a strong association between OCT-measured retinal thickness measurements at the vascular-avascular junction and ordinal disease classification in patients with ROP, indicating that OCT-based quantification of peripheral stage in ROP may have future implications for disease classification and longitudinal monitoring as an objective and quantitative biomarker.[14*] With features enabling ultra-widefield visualization and quantification of three-dimensional changes, OCT could offer more precise and continuous monitoring of stage across the length of the vascular-avascular junction than “maximum stage”, as is the current convention. In advanced ROP, OCT has been used to differentiate retinal detachment from retinoschisis, which are often difficult to distinguish by indirect ophthalmoscopy alone. Chen et al. also found handheld OCT to be useful in determining Stages 4A from 4B, which each differ greatly in visual prognosis.[41]

Vascular changes in moderate to severe ROP present along the spectrum of plus disease, and are related to the zone, stage, and extent of disease.[7**, 40, 42] Ultra-widefield OCT would be capable of visualizing all of these relationships simultaneously, non-invasively, and in three dimensions. With this imaging modality, longitudinal observation and quantification of these metrics could be utilized through an infant’s NICU course, improving clinical care and offering researchers the novel opportunity of monitoring vascular growth patterns in vivo. En face imaging could also be utilized to generate AI-derived predictions of disease severity, as has been demonstrated previously with fundus photography.[42]

Prototype handheld OCTA devices have also been developed for pediatric use, recently described to be capable of high-speed, high-resolution, and ultra-widefield image capture.[26, 29, 31, 43] Ni et al. recently reported a prototype handheld OCTA device capable of visualization of peripheral retinal vasculature and neovascularization. This would conceivably provide greater resolution for measurement of area of vascularization and localizing neovascularization, as well as nonperfusion within the vascularized retina. Used in combination with peripheral OCT B-scans, these modalities would also enable more precise ROP staging by localizing neovascularization either confined beneath or broken through the ILM.[44] Conversely, observation of avascular regions may be informative in cases of spontaneous regression or longitudinal follow-up, and serve as a potential biomarker for risk of vision threatening events like retinal detachment in the future.[45]

Identification of novel biomarkers of disease

In addition to finer detail and quantification of known clinical findings that contribute to ordinal ICROP classifications, OCT may also reveal new biomarkers for prognosticating and monitoring disease. A myriad of potential markers have already been identified in the literature.[13] Vitreous opacities have been increasingly studied for their association with ROP and correlation with severity. Vitreous material shadowing was first described by Lee et al. using handheld OCT, and later described by Zepeda et al. as dense, band-like opacities either in contact with, or hovering parallel to the retinal surface.[46–48] Legocki et al. demonstrated that, when in contact with the retinal surface, these bands are associated with a diagnosis of plus disease.[47] Separately, Scoville et al. observed punctate hyperreflective vitreous opacities in newborns, noting them to appear with higher density in premature infants. Investigations by Legocki et al. and Scoville et al. have drawn associations between presence of these opacities and ROP severity according to zone, stage, plus disease, and Type I ROP.[47, 49]

Cystoid macular edema (CME) is a feature directly observable through OCT, and commonly seen in premature infants undergoing ROP screening.[50, 51] With some exceptions, many studies have shown prevalence of CME to increase with stage, and degree of CME to correlate with stage, presence of plus disease, response to laser treatment, and a host of systemic factors.[52–55] Other investigations have implicated CME in reduced visual acuity as early as 3 months of age.

Objectively measuring choroidal and retinal features in cross section has begun to elucidate these complex relationships between choroidal structure, prematurity, and ROP. Choroidal thinning has been reported in preterm infants and associated with higher ROP stage, plus disease, lower gestational age, and lower birth weight.[54, 56, 57] This may in part be explained by the association between choroidal and retinal development.[58] Indicators of retinal immaturity including shallow foveal pit, persistence of foveal inner retinal layers, attenuated ellipsoid zones, and retinal layer thinning have been extensively described in neonates.[51, 59–61] Additionally, vertical optic nerve cup diameter and vertical optic nerve diameter have been found to be larger, and may be associated with CNS pathology and future cognitive development in preterm infants.[62]

Alternative biomarkers for plus disease have additionally been explored using OCT. Maldonado et al. calculate a vascular abnormality (VASO) score through presence of one of the following OCT-derived features: retinal vessel elevation, scalloped retinal layers, hyporeflective vessels and retinal spaces.[63] The VASO score was found to be higher in patients with clinical plus disease and demonstrated the effects of vascular dilation and tortuosity on perivascular retinal tissue.

Pathway to adoption

Since its commercial introduction in 1996, OCT has become adopted as standard of care for diagnosis and monitoring of numerous disease models in ophthalmology, and an estimated 30 million ophthalmic OCT studies are captured annually worldwide.[64] However, the widespread acceptance of OCT as a diagnostic tool has been minimally adopted in neonatal contexts. The translation of OCT to clinical use in ROP would not only be beneficial, but also possible if three criteria are fulfilled: technological feasibility, clinical utility and industry support.

Technological Feasibility

Historically, the speed and positioning limitations of tabletop OCT and OCTA devices have prohibited widespread use in the NICU. However, innovations in handheld OCT devices have made bedside NICU exams ergonomically feasible and increasingly capable of capturing scans of awake neonates. The ideal combination of features for OCT-based ROP screening remains an open question. The development of non-contact probes may reduce infection risk, at the expense of lower field of view. Contact-probes may provide UWF visualization and may work without pupillary dilation. Alternatively, contact-based probes with either disposable or cleanable lenses may provide the best combination of features and risk.

Clinical Utility

A key driver in the clinical adoption of OCT was the simultaneous introduction of pharmacologic treatment for macular disease which OCT could directly monitor. Establishing a similar link between diagnostics and therapy in the context of ROP is necessary may facilitate more rapid clinical uptake of this technology. If the next generation of handheld OCT device(s) can more objectively monitor progression and regression of the stage of ROP with less stress, a compelling case could be made for replacing the ophthalmoscopic examination. Simultaneously, investigation of alternative, OCT-derived biomarkers of disease is ongoing. While it remains to be seen whether these novel biomarkers may supplant known markers of disease, OCT lends itself to deeper innovation for solving this question. Tools for quantification and AI-based image acquisition may soon be implemented into these devices to enable validation of objective biomarkers of disease, both conventional and novel.

Industry Support

While technological advances and published research draw an increasingly compelling argument for the implementation of OCT in clinical management of ROP, buy-in from device manufacturers is essential for widespread distribution of NICU-compatible devices. There is now a network of innovation in OCT with products including handheld, home-use, and intraoperative devices, possibly lowering the threshold for device manufacturers to adopt new iterations to their line. Currently, innovation in handheld devices for ROP examination has been reported by few groups with investigational models touting such features as ultra-widefield views and non-contact image capture.[26, 29, 65*] Once optimized for clinical use, these prototypes may be brought to market by means of either a new company formed by the developers, or through licensing to a larger OCT device manufacturer. Published research demonstrating the clinical indications and technical capabilities of such devices for ROP screening is key for manufacturers to consider adding them to their product line, especially if devices can be shown to be easier to use than widefield digital fundus imaging and without the need for pupillary dilation.

The growing demand for a product that can deliver more efficient and objective means of ROP screening is evident. Advancements in neonatal care and higher preterm infant survival rates have paradoxically elevated ROP burden, heralding what may be considered a modern ROP epidemic.[66, 67] In a recent study, Bhatnagar et al. estimated an 86% increased incidence in ROP between 2003 and 2019, disproportionately affecting newborns from lower-income households and of minority demographics.[68**] To meet this growing demand, pilot tele-screening initiatives such as the Stanford University Network for Diagnosis of Retinopathy of Prematurity (SUNDROP) program have been established to remotely diagnose treatment-requiring ROP from fundus photographs obtained by trained nurses and interpreted at a single location.[17, 69–71] Implementation of OCT as an alternative to color fundus photography in a growing landscape of telescreening programs could broaden its acceptance as a diagnostic tool not only for ROP, but also for other conditions. The ongoing Newborn Eye Screen Test Study and other reports of universal newborn eye screening protocols have explored the incidence of ocular pathology in otherwise healthy infants, reporting abnormal findings in up to 23% of cases.[72–77] As conditions in which early recognition and intervention can improve visual outcomes are identified, OCT may be increasingly utilized as both a screening and diagnostic tool.

Conclusion

OCT may offer a promising solution to the subjectivity and physiologic stress associated with current methods of ROP screening. Research with investigational OCT devices has validated both non-contact and Ultra-widefield approaches, demonstrating potential for feasible use in the NICU setting. Peripheral visibility, high resolution, and cross-sectional tissue analysis could objectively measure known pathologic features and identify novel biomarkers of disease. In time, OCT may become a preferred modality in the diagnosis and monitoring of ROP.

Key Points.

• Retinopathy of prematurity is traditionally diagnosed and monitored by clinical examination, which has been shown to be highly subjective and prone to inter-observer variation.

• Though widely adopted as a standard of care in adult retinal conditions, use of optical coherence tomography (OCT) in neonatal contexts has historically been limited due to technical limitations including positioning difficulties and speed of image capture.

• UWF-OCT could be a promising alternative to bedside indirect ophthalmoscopy for its potential to reduce infant stress during examinations, reduce examination times and objectively quantify pathology.

• Recent technical advancements including ultra-widefield visualization and non-contact, high-speed image acquisition in investigational OCT devices have enabled visualization of ROP stage and revealed potential novel biomarkers for disease activity and prognostication.

• Continued refinement of technologic advancements, demonstration of clinical utility in published research and partnership with industry for manufacturing NICU-compatible OCT devices are key checkpoints in the translation of OCT to clinical management of ROP.