You are here

  • Home
  • Archive
  • Volume 97, Issue 1
  • Hand-held high-resolution spectral domain optical coherence tomography in retinoblastoma: clinical and morphologic considerations

Article Text

Article menu

Download PDFPDF

Clinical science
Hand-held high-resolution spectral domain optical coherence tomography in retinoblastoma: clinical and morphologic considerations
  1. Daniel Benson Rootman1,2,
  2. Efren Gonzalez1,3,
  3. Ashwin Mallipatna1,4,
  4. Cynthia VandenHoven1,
  5. Lindsay Hampton1,
  6. Helen Dimaras1,2,
  7. Helen S L Chan5,
  8. Brenda L Gallie1,2,6,
  9. Elise Heon1,2
  1. 1Department of Ophthalmology and Vision Sciences, The Hospital for Sick Children, Toronto, Canada
  2. 2Department of Ophthalmology and Vision Sciences, The University of Toronto, Toronto, Canada
  3. 3Surgery Department, Brigham and Women's Hospital, Boston, Massachusetts, USA
  4. 4Department of Pediatric Ophthalmology & Strabismology, Narayana Nethralaya, Bangalore, India
  5. 5Division of Hematology/Oncology, Department of Pediatrics, The University of Toronto, Toronto, Canada
  6. 6Cancer Informatics, Ontario Cancer Institute/Princess Margaret Hospital, Toronto, Canada
  1. Correspondence to Dr Elise Heon, Department of Ophthalmology and Vision Sciences, The Hospital for Sick Children, 555 University Ave, Toronto, Ontario, Canada M5G 2M9; eheon{at}attglobal.net

Abstract

Purpose Hand-held spectral domain optical coherence tomography (HHSD OCT) has greatly expanded the imaging/diagnostic capacity for clinicians managing children with intraocular retinoblastoma. We present our early experience with HHSD OCT and conventional spectral domain OCT imaging in these patients.

Methods In this retrospective cross-sectional observational study, infants were imaged during examination under anaesthesia with HHSD OCT in the supine position. Older cooperative retinoblastoma patients were additionally imaged with upright conventional OCT. Clinical data were derived from patient charts and from a prospectively maintained interinstitutional retinoblastoma database. Complementary imaging techniques, including RetCam™, fluorescein angiography and B-scan ultrasound, were assessed.

Results Twenty-two intraocular lesions in 16 patients were imaged. HHSD OCT was used exclusively in 19 lesions, while conventional OCT was also performed in three cases. Small lesions were imaged in five cases, all of which were localised to the middle retinal layers. Clinical uses for HHSD OCT imaging identified included: diagnosis of new lesions, monitoring response to laser therapy and the identification of edge recurrences.

Conclusions Although indirect ophthalmoscopy remains the gold standard for diagnosis and treatment of retinoblastoma, HHSD OCT is a valuable tool in better understanding and managing retinoblastoma.

  • Child health (paediatrics)
  • Diagnostic tests/Investigation
  • Neoplasia
  • Retina

http://dx.doi.org/10.1136/bjophthalmol-2012-302133

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Originally described in 1991 by Huang,1 optical coherence tomography (OCT) has become a powerful tool for imaging the retina in vivo. The first commercially available time-domain OCT machine was able to image a cross-section of the retina at a 10-μm resolution. The development of spectral domain technology has significantly increased the spatial resolution of these devices to upwards of 3 μm.2 These imaging systems have become an integral part of clinical ophthalmology practice across specialties.

    Most commercially available OCT systems are self-contained units in which a patient is required to sit upright and place their chin in a slit lamp-type headrest while fixating on a visual target. Imaging sequences takes only a matter of seconds, and for cooperative patients, obtaining images is relatively easy. However, cooperation and fixation are significant barriers to the use of these systems in infants and toddlers.

    Recently, the availability of a hand-held spectral domain (HHSD) OCT device (Bioptigen, Durham, North Carolina, USA) has minimised barriers for patients with cooperation challenges by allowing for directional user-defined imaging. With this device, imaging can be performed under anaesthesia in the supine position, which has proven to be useful in elucidating retinal morphologic features of children with non-accidental trauma,3 late-stage retinopathy of prematurity4 ,5 and ocular albinism.6

    As the vast majority of patients with retinoblastoma present before 1 year of age and are seen regularly under anaesthesia in the first 3 years of life, this high-resolution imaging tool allows clinicians and researchers to optimise the imaging capacity of this disease in vivo.

    Others have demonstrated the complementary use of conventional spectral domain OCT in cooperative children affected by retinoblastoma for defining the disease status in the posterior pole, and for managing retinal complications such as cystoid macular oedema (CMO) and epiretinal membranes (ERM).7

    Here, we present our experience with the use of OCT in young children with retinoblastoma. HHSD OCT imaging was used in children examined under anaesthesia, while conventional spectral domain OCT was additionally used in older, more cooperative patients.

    Methods

    Study design

    In this retrospective observational case series, all retinoblastoma patients imaged with HHSD OCT at a tertiary referral centre over a 2-year period were included. Patients were excluded if their guardians did not consent to their participation, or if they did not have imaging of sufficient quality for analysis.

    Clinical and imaging information was reviewed for each participant. Clinical information was derived from patient charts. Treatment history was reviewed from a prospectively maintained multi-institutional retinoblastoma database. All imaging modalities, including OCT, B-scan ultrasound, RetCam photography and fluorescein angiography, were additionally examined.

    Demographic data including gestational age at birth, gender and family history of retinoblastoma were collected. Disease characterisation included age of diagnosis, laterality and International Intraocular Retinoblastoma Classification group8 at presentation. The type and timing of systemic treatment including chemotherapy, radiation therapy and local periocular chemotherapy were recorded. Focal treatments were defined relative to the lesion of interest, and the type, number and interval of focal treatments were obtained. Images were extracted from institutional databases. Data and images were presented in descriptive form. No inferential statistics were performed.

    This study was approved by the institutional research ethics board, and consent for all procedures was obtained from the patients’ guardians in accordance with the 1964 Declaration of Helsinki.

    OCT procedure

    In each case of HHSD OCT imaging, infants were examined under anaesthesia as per our standard protocol. A full ophthalmic examination including hand-held slit lamp tests, biomicroscopy of the anterior segment, measurement of the intraocular pressure, cycloplegic refraction and indirect ophthalmoscopy with scleral indentation was performed for each patient visit. Additionally, RetCam imaging was acquired. If clinically applicable, B-scan ultrasound, fluorescein angiography and anterior segment imaging were performed.

    The Bioptigen (Durham, North Carolina, USA) HHSD OCT imaging system was used to acquire images in this study. This device is capable of imaging a 10×10 mm area of retina in a series of 100 horizontal spectral domain OCT images, each containing 1000 A-scans, in under 6 s. The distance between scans is approximately 80–100 μm, and the estimated axial resolution is <5 μm.6 ,9

    Patients imaged with the HHSD OCT system were examined under anaesthesia in the supine position with a lid speculum. Frequent instillation of a balanced salt solution maintained corneal clarity. The non-contact hand piece was held in the 12 o’clock position approximately 2 cm above the cornea and was directed towards the areas of interest. Manual rotation of the reference arm and retinal bore to match the patient's refraction and eye length optimised the OCT images’ clarity and brightness.10 Retinal area parameters ranged from a maximal 10×10 mm volume scan when a broad retinal area was of interest to a narrow 5×5 mm area when maximal detail of small lesions was sought.

    Conventional spectral domain OCT imaging was performed with the Cirrus OCT device (Zeiss Meditec, Oberkochen, Germany). This system has an axial resolution of 5 μm and, in the raster setting, captures 4096 A-scans for every five parallel 6-mm B-scan images.9 This is a similar resolution to the Bioptigen HHSD OCT device. Images were acquired with this system in the standard fashion, with older and more cooperative patients seated in the upright position and fixating to the best of their ability.

    Results

    Sample

    Images of 22 lesions in 16 patients were included in the study. HHSD OCT alone was used for 19 of these lesions, conventional upright spectral domain OCT images were additionally obtained from three patients. No patients were excluded due to insufficient quality of images. The median age (range) at the time of OCT imaging was 1.5 years (2 weeks–3 years).

    The median age (range) at diagnosis of patients in the study was 13 months (neonate to 6.7 years) and 68% were female. All patients in this study had bilateral disease. Systemic chemotherapy was required in 62.0% of patients, while periocular chemotherapy11 was used in five cases. Focal laser therapy was applied to 17 of the lesions at some point in the course of treatment.

    The most common pathologies imaged were scars (n=11), with or without CMO or ERM, active lesions (n=13) and edge recurrences (n=3). Longitudinal HHSD OCT imaging sequences were acquired for 10 lesions, with four of these lesions having been imaged three or more times in successive examination under anaesthesia sessions (table 1).

    View this table:
    Table 1

    Summary of patient demographics and imaging characteristics

    OCT appearance of active retinoblastoma tumours

    Imaging features of retinoblastoma tumours varied with the lesion size. Medium-sized new tumours (figure 1A and B) usually appeared as hyperdense elevations involving the middle and outer retinal layers, with shadowing of the underlying structures. The tumours were relatively homogeneous (figure 1A).

    Figure 1
    Figure 1

    Hand-held spectral domain optical coherence tomography (HHSD OCT) of larger retinoblastoma tumours shows a well-defined isodense mass. (A) left: RetCam photograph of a medium-sized macular retinoblastoma tumour in left eye of a 3-week-old patient with inherited retinoblastoma; (A) right: HHSD OCT images at Positions i and ii. Position i demonstrates relative preservation of the superficial retinal architecture in the peripheral tumour, with an isodense lesion below inducing shadowing of deeper structures. At Position ii, the retinoblastoma lesion involves the full thickness of the retina with a patch of calcification and shadowing. (B) left: RetCam photograph of another medium-sized retinoblastoma tumour along the superior arcade; (B) right: HHSD OCT images taken at Positions i and ii, also demonstrating a large isodense intraretinal lesion confined to the inner retinal layers, with relative preservation of the overlying inner retina and mild shadowing. This figure is only reproduced in colour in the online version.

    The five small lesions characterised with OCT (figure 2A–E) showed a distinctive intraretinal appearance. These spherical lesions were isodense, with smooth, distinct borders centred on the inner nuclear layer (INL) and appeared to consume the middle layers. The nerve fibre layer  remained distinct in some images. Some of the small lesions (figures 2A,C and E) allowed penetration of the OCT signals to tissues below and had a rounded appearance posteriorly, sparing the outer retinal layers. Other tumours demonstrated a slight shadow artefact from the tumour.

    Figure 2
    Figure 2

    Hand-held spectral domain optical coherence tomography (HHSD OCT) of small posterior pole lesions in young patients. Small, well-defined isodense masses with rounded edges were confined mostly to the middle retinal layers (Panels A-E). Panels A and B show some preservation of the inner retinal layers. The more visible tumour on Panel C is also denser and almost full-thickness on HHSD OCT. The well-defined rounded posterior border of the tumour suggests an early sparing of the photoreceptor layers. This figure is only reproduced in colour in the online version.

    Preretinal seeds

    Preretinal seeds in the posterior pole were imaged with HHSD OCT. Seeds had a smooth, rounded and isointense appearance, similar to intraretinal lesions. However, the seeds appeared to sit superficially on the retina, did not demonstrate an infiltrative pattern and completely shadowed the underlying retinal structures (figure 3).

    Figure 3
    Figure 3

    Hand-held spectral domain optical coherence tomography (HHSD OCT) of a preretinal tumour in a 2-year-old child who then showed recurrent disease including vitreous seeding. (A) Fundus photo and (B) HHSD OCT of the preretinal seed located at the macula. HHSD OCT demonstrates a smooth, isodense lesion superficial to the retina, without significant infiltration but with shadowing of the underlying retina. (C) Fundus photography and (D) HHSD OCT documented almost complete resolution of the seed after 3 monthly periocular topotecan depositions. This figure is only reproduced in colour in the online version.

    Early identification of very small retinoblastoma tumours

    Very early retinoblastoma may be very difficult to detect in the poorly pigmented retina of young children. (figure 4A). OCT imaging allowed us to detect tumours in their earliest stages of development and distinguish these from chorioretinal hypopigmentation. On HHSD OCT imaging, areas of active retinoblastoma growth were identified by the presence of an isodense mass replacing the normal middle retinal architecture. As the case in figure 4 was our first where the lesion was barely detectable by indirect ophthalmoscopy but was identified with HHSD OCT, no immediate laser treatment was performed. One month later, the tumour had grown, confirming the activity of the tumour (figure 4B). This lesion was then successfully treated with the 532-nm laser.

    Figure 4
    Figure 4

    Fundus photos and hand-held spectral domain optical coherence tomography (HHSD OCT) images of a subtle tumour in the left eye of a child with inherited retinoblastoma delivered at 37 weeks’ gestational age. (A) RetCam photography (top) at 1 month of age demonstrated a relatively normal fundus in the region of interest (line), but HHSD OCT (bottom) revealed a suspicious intraretinal lesion (arrow). (B) RetCam photography (top) of the same region 1 month later (line) showed a clear tumour in the same position, confirmed with HHSD OCT (bottom). This figure is only reproduced in colour in the online version.

    Monitoring of treatment response

    Response to focal therapy was monitored with HHSD OCT. Laser-treated lesions demonstrated sequential changes from a localised, isodense intraretinal mass to a progressively more variably dense, flat, full thickness chorioretinal scar. The edges of the lesions showed advanced scarring earlier in the sequence, as would be expected from laser application to normal retinal pigment epithelium  and retina surrounding the tumour (figure 5).

    Figure 5
    Figure 5

    Monitoring of the response to focal laser coagulation therapy, using hand-held spectral domain optical coherence tomography (HHSD OCT). Panels demonstrate the sequence of treatment. New lesions were identified inferior to the disk were treated with a 532-nm laser: (A) HHSD OCT image at identification; (B) 1 month after first 532-nm laser application, showing the lesion regressing and (C) the quiescent lesion 1 month after the second laser treatment. This figure is only reproduced in colour in the online version.

    Monitoring therapy and recurrence

    HHSD OCT was used to identify recurrence of an active tumour at the edge of a focal therapy scar (figure 6). Edge recurrence appeared as an isodense intraretinal mass extending from the scarred area into the external layers of the retina (figure 6B), in contrast to the primary small tumours that appeared to arise in the INL (figures 2 and 4). Treated scarred areas appeared as variably hyperdense areas of irregular retina, typically with complete loss of retinal architecture. The edges of such lesions were sharply demarcated by the physical extent of focal therapy (figure 6C).

    Figure 6
    Figure 6

    Hand-held spectral domain optical coherence tomography (HHSD OCT) to monitor potential edge recurrence. (A) RetCam and HHSD OCT images of a previously treated retinoblastoma lesion; incidental imaging of the treated edge showed a flat scar. (B) Imaging 3 months later demonstrated a small isointense tumour in the external retinal layers and under the retina, at the edge of previous scar. (C) Follow-up imaging 4 months later after two treatments with a 532-nm laser showed an expanded scar with no intraretinal tumour. This figure is only reproduced in colour in the online version.

    Discussion

    Over the past 10 years, OCT technology has revolutionised ophthalmologic practice. Ocular oncology has benefited from this trend with a number of reports describing applications of OCT to various intraocular tumours and secondary pathologies.12–14 Although it is a non-invasive test that is relatively quick and easy to perform on cooperative adults, the requirements for upright positioning and voluntary fixation have challenged its use in young children and neonates. Thus previous reports have mostly been confined to the adult population.

    The advent of HHSD OCT has provided an opportunity to image infants in the supine position. This has facilitated imaging of the pathology related to advanced-stage retinopathy of prematurity, revealing retinoschisis and macular oedema that may not have been detectable using standard examination techniques.5 Additionally, imaging retinal pathology related to non-accidental trauma with HHSD OCT3 ,15–17 and precise characterisation of foveal morphology in infants with ocular albinism6 have assisted clinicians to better understand these conditions.

    Although there is at least one report related to the use of OCT in retinoblastoma,7 this was mostly confined to upright OCT in older cooperative patients. We confirmed the findings of Shields et al,7 namely that macular pathologies such as CMO and ERM can be accurately characterised with conventional OCT in cooperative patients as young as 3.5 years of age (data not presented).

    However, in the current report, we have expanded the use of OCT imaging from older cooperative patients to the anaesthetised infant with active retinoblastoma undergoing therapy. Through the use of HHSD OCT, we have been able to characterise the morphologic features of active retinoblastoma and have demonstrated a number of avenues in which HHSD OCT can be useful in the clinical management of this condition.

    Concerning the former, our experience with HHSD OCT in active retinoblastoma sheds some light on a longstanding debate concerning the cell of origin of retinoblastoma tumours.18 All examples of the tiny, early, untreated tumours identified in this paper indicate that retinoblastoma originates in the INL of the retina (figures 2 and 4A)19 rather than the photoreceptor layer, as others have suggested based on molecular features.20 Although true characterisation of the cellular morphology and biochemistry was beyond the scope of this paper, these findings tend to support the contention that the middle retinal cells may be critical in the early stages of tumour development.

    In addition to enhancing our understanding of retinoblastoma, HHSD OCT has been useful in the management of young patients with this condition. A range of clinical applications for OCT imaging in retinoblastoma were identified in this paper. Initially, HHSD OCT helped to clarify whether a suspected retinal tumour area was active or not, showing an isodense consistency with smooth, rounded, well-defined edges in such cases.

    Serial HHSD OCT imaging found additional utility in monitoring treatment response. Retinoblastoma lesions demonstrated a characteristic series of morphologic changes on HHSD OCT from active lesions to treated scars as they progressively responded to focal therapy (figure 5). These features provide a guide and an endpoint for effective focal therapy.

    Treated inactive lesions were also monitored with HHSD OCT. We demonstrated that lesions treated with focal therapy had characteristic patterns of scarring and atrophy in the retina and retinal pigment epithelium. In contrast, recurrences at the edge of scars showed isodense expanding (convex) tumours underneath the adjacent retina (figure 6). This distinction was easily seen with HHSD OCT, thus demonstrating its utility in monitoring inactive tumours for signs of recurrence. We have continued to use HHSD OCT to identify edge recurrences and to guide therapeutic laser application until scarring is achieved.

    Finally, later in the course of disease, OCT was also useful in defining secondary pathology such as macular oedema and ERM, which is important in the estimation of visual capacity.

    A few limitations of this study should be noted. Initially, this is an observational study, and thus no data are presented as to whether outcomes are objectively affected by the clinical use of HHSD OCT. In the past, OCT imaging was not available for all patients, and thus some pathologies and clinical uses may not have been described due to selection bias. Moving forward, HHSD OCT is now performed on all posterior pole lesions in our institution.

    Although it does have proven efficacy, HHSD OCT in its current form is not a perfect technology. Imaging capacity is limited geographically to the posterior pole and morphologically to smaller lesion sizes. Although it would be interesting and clinically useful to characterise peripheral lesions and the deep architecture of larger tumours, this is not currently possible. Additionally, imaging is somewhat user-dependent and requires skill in its application. As a corollary, there is a learning curve to overcome in developing an HHSD OCT programme, and skill is required to capture high quality images consistently. Finally, as with any OCT imaging device, highly dense materials completely shadow the structures below. This is of particular concern in retinoblastoma lesions, which often contain calcium. Although we did not find this to be a hindrance in the use of HHSD OCT in our present series, caution is required when interpreting the HHSD OCT images of calcium-dense lesions.

    One other significant drawback of HHSD OCT, relative to modern conventional OCT devices is the lack of an active eye tracking system. Such tracking systems, in combination with very short acquisition times, provide for precise intrascan and interscan ocular alignments. The software tools included in some of these devices allow for quantitative assessment of changes in lesion morphology over time. However, patient position in HHSD OCT is variable relative to the orientation of the imaging sequence and no automated eye tracking system is available. Thus, it is challenging with HHSD OCT to characterise changes in a lesion of interest qualitatively over time. It is also challenging to qualitatively image a single lesion of interest over successive sessions, especially when the lesion is changing (scarring or resolving). In order to minimise this effect, we have found it useful to maintain meticulous notes and to review images from previous sessions in real time while acquiring HHSD OCT.

    Overall, although indirect ophthalmoscopy remains the gold standard for the diagnosis and treatment of active retinoblastoma, HHSD OCT imaging has dramatically improved the sensitivity to detect early tumours, recurrences and complications in focal therapy management of retinoblastoma patients. As we have demonstrated here, this technology is extremely useful in the diagnosis and surveillance of retinoblastoma of the posterior pole.

    References

    1. ↵
      1. Huang D,
      2. Swanson EA,
      3. Lin CP,
      4. et al
      . Optical coherence tomography. Science 1991;254:1178–81.
      OpenUrlAbstract/FREE Full Text
    2. ↵
      1. Drexler W,
      2. Sattmann H,
      3. Hermann B,
      4. et al
      . Enhanced visualization of macular pathology with the use of ultrahigh-resolution optical coherence tomography. Arch Ophthalmol 2003;121:695–706.
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      1. Muni RH,
      2. Kohly RP,
      3. Sohn EH,
      4. et al
      . Hand-held spectral domain optical coherence tomography finding in shaken-baby syndrome. Retina 2010;30:S45–50.
      OpenUrlCrossRefPubMed
    4. ↵
      1. Chavala SH,
      2. Farsiu S,
      3. Maldonado R,
      4. et al
      . Insights into advanced retinopathy of prematurity using handheld spectral domain optical coherence tomography imaging. Ophthalmol 2009;116:2448–56.
      OpenUrlCrossRef
    5. ↵
      1. Muni RH,
      2. Kohly RP,
      3. Charonis AC,
      4. et al
      . Retinoschisis detected with handheld spectral-domain optical coherence tomography in neonates with advanced retinopathy of prematurity. Arch Ophthalmol 2010;128:57–62.
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Chong GT,
      2. Farsiu S,
      3. Freedman SF,
      4. et al
      . Abnormal foveal morphology in ocular albinism imaged with spectral-domain optical coherence tomography. Arch Ophthalmol 2009;127:37–44.
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      1. Shields CL,
      2. Mashayekhi A,
      3. Luo CK,
      4. et al
      . Optical coherence tomography in children: analysis of 44 eyes with intraocular tumors and simulating conditions. J Pediatr Ophthalmol Strabismus 2004;41:338–44.
      OpenUrlCrossRefPubMedWeb of Science
    8. ↵
      1. Linn MA
      . Intraocular retinoblastoma: the case for a new group classification. Ophthalmol Clin North Am 2005;18:41–53, viii.
      OpenUrlCrossRefPubMed
    9. ↵
      1. Kiernan DF,
      2. Mieler WF,
      3. Hariprasad SM
      . Spectral-domain optical coherence tomography: a comparison of modern high-resolution retinal imaging systems. Am J Ophthalmol 2010;149:18–31.
      OpenUrlCrossRefPubMedWeb of Science
    10. ↵
      1. Maldonado RS,
      2. Izatt JA,
      3. Sarin N,
      4. et al
      . Optimizing hand-held spectral domain optical coherence tomography imaging for neonates, infants, and children. Invest Ophthalmol Vis Sci 2010;51:2678–85.
      OpenUrlAbstract/FREE Full Text
    11. ↵
      1. Mallipatna AC,
      2. Dimaras H,
      3. Chan HS,
      4. et al
      . Periocular topotecan for intraocular retinoblastoma. Arch Ophthalmol 2011;129:738–45.
      OpenUrlCrossRefPubMed
    12. ↵
      1. Sayanagi K,
      2. Pelayes DE,
      3. Kaiser PK,
      4. et al
      . 3D Spectral domain optical coherence tomography findings in choroidal tumors. Eur J Ophthalmol 2011;21:271–5.
      OpenUrlCrossRefPubMed
    13. ↵
      1. Shields CL,
      2. Mashayekhi A,
      3. Materin MA,
      4. et al
      . Optical coherence tomography of choroidal nevus in 120 patients. Retin-J Retin Vitr Dis 2005;25:243–52.
      OpenUrl
    14. ↵
      1. Torres VLL,
      2. Brugnoni N,
      3. Kaiser PK,
      4. et al
      . Optical coherence tomography enhanced depth imaging of choroidal tumors. Am J Ophthalmol 2011;151:586–93.
      OpenUrlCrossRefPubMed
    15. ↵
      1. Scott AW,
      2. Farsiu S,
      3. Enyedi LB,
      4. et al
      . Imaging the infant retina with a hand-held spectral-domain optical coherence tomography device. Am J Ophthalmol 2009;147:364–73.
      OpenUrlCrossRefPubMedWeb of Science
    16. ↵
      1. Sturm V,
      2. Landau K,
      3. Menke MN
      . Optical coherence tomography findings in Shaken Baby syndrome. Am J Ophthalmol 2008;146:363–8.
      OpenUrlCrossRefPubMedWeb of Science
    17. ↵
      1. Forbes BJ
      . Clues as to the pathophysiology of retinal hemorrhages in Shaken Baby syndrome determined with optical coherence tomography. Am J Ophthalmol 2008;146:344–5.
      OpenUrlCrossRefPubMed
    18. ↵
      1. Dyer MA,
      2. Bremner R
      . The search for the retinoblastoma cell of origin. Nat Rev Cancer 2005;5:91–101.
      OpenUrlPubMedWeb of Science
    19. ↵
      1. Bremner R
      . Retinoblastoma, an inside job. Cell 2009;137:992–4.
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      1. Xu XL,
      2. Fang Y,
      3. Lee TC,
      4. et al
      . Retinoblastoma has properties of a cone precursor tumor and depends upon cone-specific MDM2 signaling. Cell 2009;137:1018–31.
      OpenUrlCrossRefPubMedWeb of Science

    Footnotes

    • ▸ Additional supplementary files are published online only. To view these files please visit the journal online (http://dx.doi.org/10.1136/bjophthalmol-2012-302133).

    • Contributors All authors made substantial contributions to (1) conception and design (DBR, HD, HSLC, BLG, EH), data acquisition (DRB, EG, AM, CVH, LH, BLG,) and/or data analysis and interpretation (DBR, HD, BLG); (2) drafting the article (DBR) and/or revising it critically for important intellectual content (EG, AM, CVH, LH, HD, HSLC, BLG, EH); (3) final approval of the version to be published (DBR, EG, AM, CVH, LH, HD, HSLC, BLG, EH).

    • Competing interests None.

    • Ethics approval The Hospital for Sick Children, Toronto, Canada.

    • Provenance and peer review Not commissioned; externally peer reviewed.