Despite the widespread use of OCT in adult vitreoretinal diseases, its application to the pediatric population has been limited chiefly because of the limitations in the availability of machines in imaging the uncooperative child. Children have been imaged on OCT either in the supine position or under anesthesia in the operating room
1,2,8 –10
The utility in detecting subclinical pathologies has been established previously with the use of the TD-OCT
1,2 and more recently with the SD-OCT.
3,12,19,20 TD-OCT, for instance, has shown that ROP stage 4A with macular sparing was in fact 4B undetected by clinical examination. This would influence the surgical consideration and outcome of these infants. Chavala et al.
3 recently revealed that the handheld SD-OCT (Bioptigen Inc., Research Triangle Park, NC) showed preretinal structures, retinoschisis, and retinal detachments in three infants with severe ROP that were undetected by experts using standard examination techniques. The potential for detecting subclinical pathology using this new technology seems to be improving with the use of adaptive optics including Fourier-domain OCT
13 and by a more recently described protocol used to optimize the parameters of the handheld OCT device by Maldonado et al.
19
To a large extent, the availability of the handheld SD-OCT (Bioptigen Inc.), which provides two- and three-dimensional SD-OCT scans with a resolution of better than 10 μm, has helped image infants.
20 However, this device is also limited by the requirement of a very skilled operator, clear media, and a still infant. Moreover, reliable and reproducible images with the handheld SD-OCT can be technically challenging.
20
We recently described a method that converts a tabletop combined imaging device (Spectralis) into a handheld device to image the supine, nonanesthetized infant in the office setting. We showed that images could be obtained from the same area of interest serially and demonstrated its utility in mapping clinically missed flat neovascularization in cases of acute aggressive posterior ROP.
12 In this series, we show the utility of the same procedure in imaging the macula of infants with mild ROP.
It is noteworthy that none of the 40 control eyes or the 27 eyes with stage 1 ROP showed any foveal disruption or edema. More than a quarter (29.1%) of cases with stage 2 ROP revealed foveal changes. We chose to classify these changes into two types with no prejudice to grade them according to the severity. The pattern A changes, comprising loss of foveal depression and a dome-shaped serous elevation was noted in 12 (52.2%) of these eyes. Pattern B or cystic changes within the retinal layers with preservation of the normal foveal contour was noted in 47.8%. More interestingly, we noted that of the 19 eyes with abnormal OCT changes imaged at 52 weeks' PMA, all (100%) normalized at this visit. However, we could not obtain serial images this late (PMA) in the remaining eyes included in the study, and in those we did image at 52 weeks' PMA, we were not able to determine the exact week of normalization because weekly SD-OCT images were not possible.
The etiology of these macular changes, especially the “edema” that resembles adult cystoid macular edema, is currently unknown. Although we did not find any neonatal risk factor(s) to be significantly distributed between the three groups, it is possible that unique comorbid factors may influence the disease pattern in our setting, which may influence structural changes in the macula, especially in the heavier infants screened in our country. This may not be applicable to other countries where these heavier infants are not at similar risk.
21,22 In the absence of any obvious causes for the macular edema, we hypothesize that this “macular edema” seen in eyes (29.1%) with more severe ROP could either be (1) a response to biochemical modulators, including higher concentrations of vascular endothelial growth factors (VEGFs), which could play a role in increased vascular permeability leading to retinal edema, or (2) could be caused by mechanical traction exerted on the macula. The biologic plausibility and rationale of the hypothesis of the biochemical theory has been suggested in retinal vascular disorders including AMD, diabetic retinopathy, retinal artery or vein occlusion, and ROP.
23 This is further corroborated in cases of stage 4 ROP and more advanced disease wherein VEGF was observed in much higher concentrations in the vitreous compared to controls.
24 In a study of 27 cytokines in the vitreous of ROP cases, VEGF was found to have the strongest correlation with vascular activity in the disease.
25 Studies correlating early ROP (stages 1 and 2) with vitreous levels of these cytokines are not available because of the difficulty in obtaining samples from these early cases. However, it is probable that levels in “excess of normal” contributing to increased permeability in the retinal vessels begin at these early stages and could be responsible for the macular changes we have observed. However, it must be emphasized that this theory currently remains speculative, because there is insufficient evidence to suggest that VEGF is significantly different between cases of stage 1 and 2 ROP.
With regard to the mechanical theory, there is currently insufficient evidence to believe that the “ridge” (stage 2) may contribute to mechanical traction that may extend to the fovea being greater in volume than the “line of demarcation” (stage 1) to explain the foveal changes on OCT. However, this has been implicated in stage 4A ROP, where OCT has shown the separation or schisis-like change to extend more posteriorly, converting a diagnosis of 4A to stage 4B.
2 To summarize, we are unsure at this time of the exact etiology of the foveal edema in some and not in others. Our theories remain speculative and at best may serve to encourage additional research on this subject.
The findings of this study also allow us to extend our understanding of the development of the premature fovea to hypothesize the possible clinical and long-term effects of these abnormal changes. Anatomic studies on the development of the premature fovea have revealed that the formation of a parafoveal avascular zone occurs by midgestation, a period that coincides with the birth of subjects with ROP.
26,27 Later in gestation, there is a widening of the foveal pit with elongation of the cone inner and outer segments, and closer cone–cone packing that occurs and continues after birth and into early childhood.
14 –16 In formation of the ROP fovea, the centrifugal forces that lead to pit widening may not be intimately linked to the centripetal forces that lead to cone packing.
13 In addition, growth factors, including VEGF, neuropilin, and semaphorin, have been shown to have a role in the ROP fovea.
28,29
Adaptive optics Fourier-domain OCT (AO-FDOCT) on older individuals with historically mild ROP was reported by Hammer et al.
13 They observed neurovascular abnormalities in seven of nine cases (77.8%), and the authors opined that mild ROP may not universally contribute to long-term changes. Years after suffering from mild ROP, these cases presented with degraded best corrected acuity attributed to mild optical aberration or metabolic effects on neural cells that are sensitive to contrast. It was also suggested that the loss of foveal cones or increased cone–cone spacing were also responsible for these visual changes. Interestingly, it was noted that there was no paucity of cones years later despite the fact that cone packing may have been affected. We hypothesize that the transient edema in 29% of our cases may contribute to abnormalities in cone packing without affecting the actual number of cones by causing increased physical separation between adjacent cones that may hinder their tight packing during this critical period of immaturity—namely, between 37 and 52 weeks' PMA.
However, it must be emphasized that correlating vision in adults with historical ROP with macular changes is currently speculative in the absence of long-term follow-up studies. Clinically, the vision of young adults with historical ROP may be less than
13 or better than
30 predicted from the appearance of the macula. Studies of adults also suggest that vision may be altered by changes described as a vestige of prematurity, including changes in foveal depression, hyporeflectivity, preserved retinal layers, increased central foveal thickness, and total macular volume measured on TD- OCT.
31 Cellular level changes and histologic evidence suggest that intraretinal separation in the inner retinal layers account for increased retinal thickness in patients with macular heterotopias, which correlates with reduced visual acuity.
32 In addition, OCT in adults with historical ROP presenting with relatively normal maculae clinically has revealed a loss of foveal depression, increased macular thickness, and continuation of inner retinal layers within the fovea.
33 A small or absent foveal avascular zone
34 and attenuated central retinal ERG responses to multifocal stimulation in children
35 with historic ROP helps us hypothesize about the long-term effects of early macular changes described in the study. However, only long-term studies can conclusively help us understand the relevance of these early changes. Although the changes resolve by 52 weeks' PMA, it is uncertain if permanent effects may have set in by this time.
There are several other limitations of this study. First, the retrospective nature prevents serially documented OCT changes week after week in the acute period. This would give us a better understanding on the linear progression or regression of the macular changes. Second, we noted that these patterns of foveal edema resolved by 52 weeks' PMA. It is possible that the fovea normalized earlier, but because we schedule our infants at the third month of corrected age (approximately 52 weeks' PMA), this may bias the timing of our finding. Third, Raster scans to determine the volumetric distribution was not possible because this system is incapable of retinal tracking and because the image quality of these scans is poor. This limitation has also been noted in young adults with ROP even with advanced AO-FDOCT.
13 Line or section scans were possible with our method, even though the infant was not anesthetized. Other important limitations include the laterally inverted images because of the position of the operator in relation to the position of the infant. Motion artifacts were common, but were reduced with increasing practice. The procedure does, however, require a skilled operator who must adjust the mobile camera unit to reduce motion artifacts. The lack of a mount for image stabilization is a limitation. It must be noted that the Spectralis OCT machine was probably not intended for pediatric use, and our modification must be viewed in that light.
To the best of our knowledge, this is the first study that has reported subclinical changes on SD-OCT in the fovea in the acute ROP period in ROP that did not require treatment. We propose that these macular changes in the early premature period, although transient, could be the basis of future macular and foveal architectural changes reported in older ROP survivors,
13 contributing to the unexplained poor vision in these patients. Additional research is required to confirm this.