Retinoblastoma is the most common primary intraocular malignancy in children. In recent years, advances in treatment options such as highly selective intra-arterial and intravitreous chemotherapy have reduced the use of systemic chemotherapy, external beam radiation therapy, and enucleation while maintaining rates of survival.1 Despite these advances in therapy, ophthalmoscopy remains the gold standard for retinoblastoma detection and evaluation for treatment response. In recent years, handheld and portable floor-mounted spectral-domain optical coherence tomography (OCT) has shown clinical utility in the detection of ophthalmoscopically invisible retinoblastoma tumors, monitoring treatment response, and evaluation for recurrence at the edge of treated tumors.2–4 Description of the retinal vasculature in retinoblastoma through fluorescein angiography (FA) is limited in the literature.5–8 Reports describing changes in FA features following tumor regression such as a diminution of the intrinsic capillary network and regression of feeder vessels show that angiography may be a useful ancillary in monitoring treatment response.5–8 FA is not widely used at each repeat evaluation of retinoblastoma patients, however, due to its indeterminate clinical utility.
OCT angiography (OCTA) is a noninvasive imaging modality that allows for imaging of the retinal and choroidal vasculature while circumventing many of the drawbacks of conventional angiography. We have previously described the OCTA features of a small untreated retinoblastoma tumor, which revealed large superficial feeder vessels that circumscribed the tumor and branched into a dense intrinsic network of vessels.9 There are no reports, however, on the clinical features or utility of OCTA in the evaluation of treated retinoblastoma lesions. In this study, we report on the OCTA features of treated and clinically regressed retinoblastoma lesions in a series of five children imaged with an investigational portable OCTA system.
Patients and Methods
Following approval of the Duke Institutional Review Board, children with ongoing evaluation for retinoblastoma were recruited after obtaining written parental consent. This exploratory study adhered to the tenets of the declaration of Helsinki and the Health Insurance Portability and Accountability Act. Recruited subjects were those with a known diagnosis of retinoblastoma who had recently completed chemoreduction with systemic chemotherapy or intra-arterial chemotherapy (IAC) and were undergoing a regularly scheduled examination under anesthesia (EUA). All EUAs and local treatment were performed by the same physician (MAM). Eyes were graded using the international classification of retinoblastoma. During the EUA, the regressed tumors were imaged with an investigational portable OCTA system (Spectralis with Flex and OCTA module; Heidelberg Engineering, Heidelberg, Germany), and imaging was performed as previously described.10 Automated and manual segmentation of the retinal layers were difficult or not possible in areas of tumor due to disorganization and obliteration of normal cellular boundaries.
A summary of the treatment history and OCTA features of imaged retinoblastoma lesions is summarized in Table 1.
Summary of the Treatment History and OCTA Features of Imaged Retinoblastoma Lesions
Case 1 was a 17-month-old girl with bilateral group B retinoblastoma treated with six cycles of systemic chemotherapy in addition to multiple sessions of transpupillary thermotherapy (TTT) and cryotherapy in both eyes. At the time of OCTA imaging, multiple regressed retinoblastoma lesions were noted in both eyes. OCTA imaging of a lesion with type 2 regression (noncalcified remnant) in the right eye highlighted a kink in a large caliber retinal venule as it crossed the tumor (Figure 1). There was a network of moderate caliber vessels feeding and draining the lesion circumferentially and a rather dense network of fine intrinsic vessels (at various depths on cross-sectional OCTA images) not seen on FA. OCTA imaging of a portion of a lesion with type 4 regression (atrophic scar) in the left eye revealed an absence of a capillary network within both the retina and choriocapillaris. The lesion in the right eye underwent indocyanine green-enhanced TTT and the lesion in the left eye was observed.
Multimodal imaging of Case 1. Fundus photo of a retinoblastoma lesion with type 2 regression in the right eye (A). Mid-phase fluorescein angiography of this lesion (B) did not reveal any obvious feeder vessels or intrinsic vascularity. (C) En face 20 ° × 20° optical coherence tomography angiography (OCTA) of the lesion revealed a circumferential network of feeder and draining vessels (arrows) around the noncalcified remnant of the tumor and prominent intrinsic vascularity to this remnant. (D) OCT B-scan with flow overlay of the lesion. This lesion was treated with transpupillary thermotherapy. Fundus photo of the left eye (E) showing a portion of a retinoblastoma lesion with type 4 regression. A horizontal line scan with OCTA flow overlay (F) demonstrates lack of flow signal in the regressed tumor. En face 20° × 20° OCTA of this lesion showed minimal vascularity in the superficial vascular complex (G), deep vascular complex (H), and choriocapillaris (I). This lesion was observed.
Case 2 was an 18-month-old girl with unilateral group C retinoblastoma. She had previously received four cycles of IAC (most recently 7 months prior) and multiple sessions of TTT (most recently 4 months prior). Examination revealed type 1 regression (calcified remnant) of a macular retinoblastoma lesion. OCTA revealed a dense network of capillaries in the portion of retina superficial to the areas of calcification (Figure 2). OCTA analyses within the areas of calcification, although limited by shadowing, revealed a less dense but more dilated and tortuous network of intrinsic vessels. This lesion was observed.
Multimodal imaging of Cases 2 and 3. Color fundus photo (A) of a retinoblastoma lesion from Case 2 shows type 1 regression. En face 20° × 20° optical coherence tomography angiography (OCTA) through the superficial aspect of this lesion (B) shows a dense intrinsic vascular network with fine vessel caliber, whereas OCTA through a deeper aspect of the lesion (C) shows a more dilated and tortuous network of intrinsic vessels and shadowing from the calcifications. (D) OCT B-scan with flow overlay of the lesion. This lesion was observed. (E) OCT B-scan with flow overlay of the retinoblastoma lesion from Case 3. Color fundus photo from Case 3 showing a retinoblastoma lesion with type 3 regression (F). The noncalcified remnant (red arrow) can be appreciated. There is an area with a pearly appearance concerning for tumor reactivation (green arrow). OCTA through the lesion (G) shows a prominent well demarcated vascular network with a dilated draining vessel in the area concerning for tumor reactivation (green arrow). OCTA also reveals finer lacey vascularity with diminutive feeder/draining vessels in the inactive noncalcified remnant (red arrow). The area concerning for tumor reactivation was treated with transpupillary thermotherapy.
Case 3 was a 23-month-old boy with group B retinoblastoma in the right eye and group D retinoblastoma in the left eye. He previously underwent six cycles of systemic chemotherapy and multiple sessions of TTT and cryotherapy in both eyes. Funduscopic examination of the right eye revealed an area concerning for tumor reactivation at the edge of a tumor previously noted to have type 3 regression (both calcified and noncalcified remnants). OCTA of this lesion revealed a distinct area of intrinsic vascularity with telangiectatic vessels and a dilated draining vessel corresponding to the area of suspected tumor reactivation (Figure 2). The area of the lesion corresponding to the clinically inactive noncalcified remnant revealed sparser and lacy vascularity and diminutive feeder and draining vessels. The area of reactivation was treated with TTT. Examination also revealed a large macular lesion in the left eye with type 1 regression. OCTA features of this lesion were similar to the type 1 regression pattern described in Case 2, in that vascularity was noted in the superficial retina overlying the area of calcification. This lesion differed from Case 2 in that the height of the lesion and extent of calcification was greater. Thus, we were unable to accurately detect flow deeper in the lesion due to significant shadowing. This lesion was observed.
Case 4 was a 3-year-old girl with 13q deletion syndrome and bilateral retinoblastoma previously treated with six cycles of systemic chemotherapy and enucleation of the right eye, and multiple sessions of TTT and cryotherapy in the left eye. She was referred for tumor reactivation in the left eye. She received two rounds of IAC as well as multiple sessions of TTT and cryotherapy, most recently 3 months prior. Examination revealed an exophytic tumor in the macula with type 3 regression. OCTA through the noncalcified portion revealed a complex network of vessels in the superficial aspect of the lesion with prominent feeder vessels (Figure 3). Evaluation of the deeper aspect of this noncalcified portion was limited by projection artifact and shadowing but did reveal a fine network of vessels as well. OCTA features of the calcified portion of the tumor were similar to that seen in type 1 regression. The noncalcified portion of this lesion was treated with TTT. Additionally, there was a lesion with type 4 regression which had OCTA features similar to the lesion with type 4 regression described in case 1.
Multimodal imaging of Case 4. Fundus photo (A) showing a retinoblastoma lesion in the macula with type 3 regression. The noncalcified portion appears somewhat pearly (red arrowhead). En face 20° × 20° optical coherence tomography angiography (OCTA) through a portion of this lesion (outlined in green) shows dense intrinsic vascularity (B) in the superficial aspect of the noncalcified portion with dilated feeder/draining vessels (red arrowhead). The OCTA image also shows a deeper aspect of the calcified portion of this lesion, which reveals more dilated intrinsic vasculature (green arrow) and shadowing from the calcifications. (C) OCT B-scan with flow overlay of the lesion. The noncalcified portion of this lesion was treated with transpupillary thermotherapy.
Case 5 was a 2-year-old boy with bilateral retinoblastoma. He had previously undergone six cycles of systemic chemotherapy elsewhere. Neuroimaging had revealed concern for possible optic nerve extension in the left eye, for which IAC was recommended. Owing to denial by his insurance company, he underwent external beam radiation to the left eye. He was subsequently referred to our clinic and received multiple TTT treatments in both eyes, most recently 12 months prior. Examination of the left eye revealed a nasal lesion with type 1 regression. OCTA revealed findings similar to other lesions with type 1 regression in this report (Figure 4).
Multimodal imaging of Case 5. Fundus photo (A) showing a retinoblastoma lesion with type 1 regression. En face 20° × 20° optical coherence tomography angiography (OCTA) through a deeper aspect of this lesion (B) reveals a dilated, tortuous vascular network similar to that seen in other lesions with type 1 regression. (C) OCT B-scan with flow overlay of the lesion. This lesion was observed.
In this single-center, exploratory, consecutive case series, we provide the first description of OCTA features in treated retinoblastoma lesions demonstrating a variety of regression patterns. We found that tumor vascular features corresponded with the pattern of regression and clinical assessment of lesion activity.
In lesions with type 1 regression, OCTA revealed significant vascularity in the superficial retina and a more dilated network in the deeper layers. A single lesion with type 2 regression with some residual clinical activity showed prominent vascularity on OCTA in the superficial and deeper layers and somewhat prominent feeder vessels. Lesions with type 3 regression showed combined OCTA features of lesions with types 1 and 2 regression. Thus, inactive lesions with types 1, 2 and 3 regression all had persistent intrinsic vascularity on OCTA Shields et al reported that tumor regression is accompanied by diminution of the caliber of feeder vessels and reduced intrinsic vascularity on FA.5 Given that we did not have pretreatment OCTA data available, it is unclear whether the intrinsic vascular density declined followed treatment and if so to what extent. Lesions with type 4 regression showed a diffuse flow void on OCTA at the level of the retina and choriocapillaris. Areas with tumor reactivation or persistent activity showed a distinct vascular network with prominent feeder and/or draining vessels.
The OCTA pattern of lesions with type 4 regression and tumor reactivation validate FA findings of similar lesions.5–7 OCTA also revealed distinct microvascular features of lesions with types 1, 2, and 3 regression, which has not been reported previously with FA. Additionally, OCTA revealed intrinsic vascularity and feeder vessels to a lesion, which were not seen on the FA we performed.
Histopathologic studies have shown that greater tumor vascularity in retinoblastoma is associated with greater risk of local invasive growth, thus highlighting the potential importance of monitoring vascular density in such lesions over time.10 OCTA offers a noninvasive method of monitoring for such vascular changes, but normative data and longitudinal imaging are required to better understand the clinical implication of these changes. It is reasonable to hypothesize that persistently engorged feeder vessels or an increase in vascular density over time on OCTA could be a marker of persistent disease activity or tumor reactivation.
The limitations of this study include a small sample size of five patients and an unavailability of pretreatment OCTA data. Furthermore, segmentation of the retinal vasculature into the standard superficial and deep vascular complexes was not possible due to obliteration of discernable retinal layers. Full visualization of deeper vascular beds was also hindered due to shadowing from calcifications and thickened areas of tumor. Although projection artifact was not noted from larger feeder vessels, it was challenging to discern whether there was projection artifact from the lacy fine intrinsic vasculature. Additional limitations pertain to the technical aspects of portable OCTA image acquisition. Owing to the limited field of view, OCTA acquisition was confined to small, posterior tumors or only portions of larger or more peripheral tumors. Lubrication of the corneal surface and elimination of any movements was needed for image acquisition.
Overall, OCTA evaluation of eyes with retinoblastoma may help better understand the intrinsic microvascular response to chemoreduction and consolidation and may become a useful adjunct in monitoring disease activity.
- Abramson DH, Shields CL, Munier FL, Chantada GL. Treatment of Retinoblastoma in 2015: Agreement and Disagreement. JAMA Ophthalmol. 2015;133(11):1341–1347. https://doi.org/10.1001/jamaophthalmol.2015.3108 PMID: doi:10.1001/jamaophthalmol.2015.3108 [CrossRef]26378747
- Rootman DB, Gonzalez E, Mallipatna A, et al. Hand-held high-resolution spectral domain optical coherence tomography in retinoblastoma: clinical and morphologic considerations. Br J Ophthalmol. 2013;97(1):59–65. https://doi.org/10.1136/bjophthalmol-2012-302133 PMID: doi:10.1136/bjophthalmol-2012-302133 [CrossRef]
- Seider MI, Grewal DS, Mruthyunjaya P. Portable Optical Coherence Tomography Detection or Confirmation of Ophthalmoscopically Invisible or Indeterminate Active Retinoblastoma. Ophthalmic Surg Lasers Imaging Retina. 2016;47(10):965–968. https://doi.org/10.3928/23258160-20161004-12 PMID: doi:10.3928/23258160-20161004-12 [CrossRef]27759865
- Finn AP, House RJ, Hsu ST, et al. Hyperreflective Vitreous Opacities on Optical Coherence Tomography in a Patient With Bilateral Retinoblastoma. Ophthalmic Surg Lasers Imaging Retina. 2019;50(1):50–52. https://doi.org/10.3928/23258160-20181212-08 PMID: doi:10.3928/23258160-20181212-08 [CrossRef]30640396
- Shields JA, Sanborn GE, Augsburger JJ, Orlock D, Donoso LA. Fluorescein angiography of retinoblastoma. Trans Am Ophthalmol Soc. 1982;80:98–112. PMID:6892130
- Ohnishi Y, Yamana Y, Minei M, Ibayashi H. Application of fluorescein angiography in retinoblastoma. Am J Ophthalmol. 1982;93(5):578–588. https://doi.org/10.1016/S0002-9394(14)77372-6 PMID: doi:10.1016/S0002-9394(14)77372-6 [CrossRef]6177245
- Bianciotto C, Shields CL, Iturralde JC, Sarici A, Jabbour P, Shields JA. Fluorescein angiographic findings after intra-arterial chemotherapy for retinoblastoma. Ophthalmology. 2012;119(4):843–849. https://doi.org/10.1016/j.ophtha.2011.09.040 PMID: doi:10.1016/j.ophtha.2011.09.040 [CrossRef]
- Rössler J, Dietrich T, Pavlakovic H, et al. Higher vessel densities in retinoblastoma with local invasive growth and metastasis. Am J Pathol. 2004;164(2):391–394. https://doi.org/10.1016/S0002-9440(10)63129-X PMID: doi:10.1016/S0002-9440(10)63129-X [CrossRef]14742245
- House RJ, Hsu ST, Thomas AS, et al. Vascular Findings in a Small Retinoblastoma Tumor Using OCT Angiography. Ophthalmol Retina. 2019;3(2):194–195. https://doi.org/10.1016/j.oret.2018.09.018 PMID: doi:10.1016/j.oret.2018.09.018 [CrossRef]31014771
- Hsu ST, Chen X, Ngo HT, et al. Imaging Infant Retinal Vasculature with OCT Angiography. Ophthalmol Retina. 2019;3(1):95–96. https://doi.org/10.1016/j.oret.2018.06.017 PMID: doi:10.1016/j.oret.2018.06.017 [CrossRef]30935662
Summary of the Treatment History and OCTA Features of Imaged Retinoblastoma Lesions
|Case||Tumor No.||Prior Treatment||FA Performed?||Clinical Regression Pattern||Superficial Vasculature Present?||Dilated Feeder/Draining Vessels?|
|1||1||Systemic chemotherapy (6 cycles), TTT (5 sessions)||No||2 (Residual activity)||Yes||Yes|
|2||Systemic chemotherapy (6 cycles), TTT (4 sessions)||No||4||No||No|
|2||1||IAC (4 rounds), TTT (4 sessions)||No||1||Yes||No|
|3||1||Systemic chemotherapy (6 cycles), TTT (5 sessions)||Yes||Tumor reactivation||Yes||Yes|
|2||Systemic chemotherapy (6 cycles), TTT (3 sessions)||Yes||1||Yes||No|
|4||1||Systemic chemotherapy (6 cycles), IAC (2 rounds), TTT (5 sessions)||No||3 (Residual activity)||Yes||Yes (at edge with residual activity)|
|2||Systemic chemotherapy (6 cycles), IAC (2 rounds), TTT (3 sessions)||No||4||No||No|
|5||1||Systemic chemotherapy (6 cycles), EBRT, TTT (4 sessions)||No||1||Yes||No|