Introduction: The objective of this study was to report the clinicopathologic features of three cases of MYCN-amplified retinoblastoma identified genetically by aqueous humor sampling. Methods: Whole-genome sequencing was performed using isolated cell-free DNA (cfDNA) from aqueous humor of 3 retinoblastoma patients. We analyzed genomic copy number and mutational alterations, histologic and pathologic features, and clinical data. Results: The most common genetic alteration identified in these three retinoblastoma cases was a focal MYCN amplification on 2p. All tumors showed an early age of diagnosis with a median of 9 months. The tumor histopathologic features included neovascularization and subretinal seeding in case 1, diffuse nature with choroidal and prelaminar optic nerve invasion in case 2, and complete vitreous seeding in case 3. Case 1 expressed RB protein and had no RB1 mutation, case 2 did not express RB protein and had an RB1 mutation, and case 3 did not express RB protein and likely had an epigenetic effect on RB expression. Conclusions: Our report shows 3 cases of unilateral retinoblastomas diagnosed in patients ranging from 4 months to 18 months old. Genomic analysis from AH cfDNA revealed MYCN amplification with intact RB protein staining in case 1 and lack of RB staining in cases 2 and 3. RB1 mutational analysis in the AH confirmed a pathogenic variant in case 2. Clinical pathology showed features requiring aggressive treatment, specifically enucleation. Importance:MYCN-amplified retinoblastomas demonstrate unique pathogenesis and aggressive behavior, regardless if MYCN is a primary or secondary driver of disease. Genomic analysis from aqueous humor may be useful when deciding to enucleate as opposed to treating conservatively. Focal MYCN amplification on 2p might be relevant for tumor growth in this subset of the retinoblastoma population in terms of targeted therapeutics.
Retinoblastoma is the most common primary intraocular malignancy in childhood, encompassing 4% of all pediatric cancers . Ninety percent of retinoblastomas are diagnosed in patients younger than 3 years. The incidence of retinoblastoma is 1 in 20,000 live births , with about 200–300 new cases in the USA each year . The highest number of cases annually per country is found in India, China, and Africa .
Retinoblastoma arises from neuroectodermal blast cells from the inner layer of the optic cup. The tumor may grow toward the vitreous and/or subretinal space, forming a multiloculated white mass with associated avascular vitreous or subretinal seeds. In this setting, the tumor may grow around neovascular central vessels or existing retinal vasculature for nutrition. When the tumor outgrows its vascular supply, it undergoes ischemic necrosis which may result in dystrophic calcification, yielding a chalky appearance [4, 5]. Additionally, rosettes range from less differentiated Homer Wright rosettes, also present in neuroblastomas, a more differentiated Flexner-Wintersteiner rosettes, which are specific for retinoblastoma, to a photoreceptor differentiation called fleurettes which are characteristic of a retinocytoma or further differentiation .
Most retinoblastomas (∼98%) are initiated by a pathologic variant of the tumor suppressor gene RB1 and acquire other chromosomal alterations. These “typical” RB tumors have neuroendocrine-type nuclei with the previously described differentiation patterns of rosettes and/or fleurettes. Approximately 2% of unilateral retinoblastoma cases form a subgroup with no detectable mutation in the RB1 gene but instead are primarily driven by MYCN gene amplification. These tumors possess a distinctive bland morphology with cells that have a round nucleus and frequent large nucleoli and lack rosette or fleurette differentiation . While MYCN amplification is the primary driver in this small subset (RB1+/+/MYCNA), this alteration can also be seen in my “typical” RB tumors that have a mutated RB1 gene (RB1−/−), leading to loss of functional RB protein.
Recent advances in local therapy for retinoblastoma have shown paracentesis with extraction of the aqueous humor (AH) to be safe [7‒9]. Liquid biopsies based on circulating tumor cells and cell-free DNA (cfDNA) in the blood or other fluids have been explored for other cancers as a means to further prognosticate therapeutic outcomes without the need for an invasive tissue biopsy . Berry et al.  first described the potential of AH as a surrogate tumor biopsy for retinoblastoma. Those authors analyzed retinoblastoma tumor DNA in 63 separate AH samples, showing that this is a valid source of tumor-derived cfDNA and is representative of the genomic state of the tumor. With access to tumor DNA in vivo, they were able to identify differences in the somatic chromosomal copy number alteration profiles from enucleated and salvaged eyes. They also found a potential benefit of longitudinal sampling of the AH as the overall amplitude of genomic alterations may provide a real-time measure of therapeutic response.
Herein we report 1 patient with unilateral retinoblastoma without an RB1 mutation, intact RB protein and with MYCN amplification (RB1+/+/MYCNA) (case 1), and 2 patients with focal MYCN amplification and lack of RB protein; a pathogenic variant in the RB1 gene was identified from the AH in one case (case 2), and one case possibly had an epigenetic effect causing lack of RB protein (case 3). In all cases, the MYCN amplifications were identified through AH sampling. We show that AH can be extracted at the time of diagnosis for these patients to identify this biomarker. This allows for more precise prognostication and treatment planning and underlies that as with other tumors, MYCN amplification is a poor prognostic biomarker for RB.
A 4-month-old female presented with a chief complaint of abnormal eye movements involving the right eye. Her right eye was deviated inward for 3 weeks. It also appeared to be darker than the left eye. Her father also noticed what appeared to be a translucence in her right pupil. This white pupillary reflex is otherwise known as leukocoria. Upon further examination, it was found that the right eye alignment was not maintained, there was no red reflex in the right eye, and there was a positive afferent pupillary defect in the right eye. The eye also showed evidence of neovascularization and buphthalmos, and thus was classified as group E American Joint Commission on Cancer (AJCC) stage cT3c. MRI was done within a week of the patient’s first visit, and enucleation was completed less than a month later. AH was collected at primary enucleation, shortly after diagnosis.
The tumor was found to have an exophytic growth pattern that filled the posterior compartment of the eye (Fig. 1a). There were numerous subretinal pigment epithelium tumor deposits. The tumor consisted of sheets of cells with round to oval nuclei and prominent nucleoli. There were occasional apoptotic cells within the tumor. There were no rosettes or fleurettes present. Immunohistochemical stains were positive for RB1 in the tumor (Fig. 2). The tumor did not invade the choroid nor was there optic nerve invasion; thus, there were no high-risk features.
Genomic analysis was done via an AH liquid biopsy approach. The surgical procedure and details of genomic analysis of the AH has been described extensively by Berry et al. [11, 12] In this case, genomic analysis demonstrated significant MYCN amplification in the absence of other alterations. Fresh tumor was sampled after eye removal in the operating room. Analysis of this tumor tissue, performed by Impact Genetics (Brampton, Ontario), confirmed MYCN amplification with a ratio to the median of 19.76. Copy number analysis of the AH showed 23 copies of the MYCN oncogene in the patient’s tumor. The patient’s blood showed normal 2 copies of the MYCN gene. In addition, both blood and tumor showed no mutations of the RB1 gene.
A 9-month-old male presented with chief complaint of abnormal eye movements involving the left eye and cloudiness of the left pupil for 4 months. The patient was otherwise healthy without a family history of ocular diseases or cancers. In-office examination was notable for a shallow left anterior chamber and retinal detachment in the left eye, which prevented view of posterior structures. Therefore, examination under anesthesia was performed; a large, exophytic, whitish vascular tumor was seen in the left eye (Fig. 1b). The presence of tumor in the left eye was confirmed with B-scan ultrasound, ultrasound biomicroscopy, and MRI. The tumor was classified as group D AJCC stage cT2b. The eye was primarily enucleated given the extent of disease, and AH was collected at the time of primary enucleation done the same day as the diagnostic EUA . Genetic analysis of the AH revealed focal MYCN amplification on chromosome 2 with a ratio to the median of 5.52. The patient’s serum was analyzed for a germline RB1 mutation which was not detected. Mutational analysis of the AH demonstrated a pathogenic variant in exon 17 of the RB1 gene, c.1666C>T, p. Arg556*, with a variant allele fraction of 52.2%.
Enucleation of the left eye revealed a moderately differentiated tumor with Flexner-Wintersteiner rosettes and exophytic growth pattern. Areas of calcification were present with minimal (10%) necrosis. There was also evidence of mild focal anaplasia. Multiple peripheral subretinal seeds formed plaques over the retinal pigment epithelium. There was massive choroidal invasion (7.0 mm in maximum diameter). The tumor minimally invaded the optic nerve head (prelaminar invasion 0.1 mm from Bruch’s membrane), though the optic margin and meninges were found to be free of tumor. Immunohistochemical staining for RB protein shows that the tumor had negative staining (meaning lack of a functional protein), while the vascular endothelium and the retina cells were positive for nuclear staining, as is expected (Fig. 3). There was also a total retinal detachment. Additionally, a small focus of extramedullary hematopoiesis (less than 0.1 mm) was found adjacent to the choroidal invasion. The choroidal invasion was a high-risk feature. Adjuvant chemotherapy was offered to the family to decrease the risk of metastatic disease; however, it was declined. The last clinical follow-up was 43 months after diagnosis, and he remained free of disease.
An 18-month-old male presented with a chief complaint of leukocoria of the right eye first noticed 3 months prior to presentation. The patient’s family also noted abnormal movement of the right eye 3 months prior to diagnosis. The patient was otherwise in good health, with no family history of eye diseases or cancers. On in-office examination, the patient had good fix and follow, with normal intraocular pressure to palpation bilaterally. Slit-lamp exam was within normal limits for both eyes. Posterior segment exam was limited due to patient cooperation, but a large, vascular, endophytic tumor was visualized in the right eye. Examination was done under anesthesia for better characterization of the tumor. Anterior segment examination showed that the tumor was visible behind the lens in the right eye. Posterior segment exam was notable for a large, endophytic, anterior nasal tumor blocking view of normal structures, with extensive vitreous seeding (Fig. 1c). Imaging including B-scan ultrasound, ultrasound biomicroscopy, and MRI all supported a diagnosis of unilateral retinoblastoma. MRI demonstrated a hypercellular mass along the medial posterior margin of the globe without abnormal enhancement of the optic nerve; additional features described by Jansen et al.  including anterior, peripheral tumors with plaque or pleomorphic shape, with irregular folding, tumor-retinal folding, and peritumoral hemorrhage were not identified. Clinically, however, there was retinal folding. During this diagnostic EUA, AH liquid biopsy sample was collected.
The tumor was found to be mostly regressed with large areas of calcification and well-differentiated retinocytoma-like features. It had an endophytic growth pattern with minimal (less than 10%) tumor necrosis admixed with regressed tumor. Residual retinoblastoma was present primarily as vitreous seeds (clouds, dust, and spheres) in the anterior vitreous. Immunohistochemical staining for RB protein showed that the tumor had lack of a functional RB protein as evidenced by the negative tumor staining while the vascular endothelium and the retina cells were positive for nuclear staining (Fig. 4). Since there was no RB mutation, this was interpreted to possibly be due to an epigenetic effect on RB1. There was focal retinal detachment and degenerated retina with gliosis. There was a gliovascular epiretinal membrane over anterior tumor. The tumor did not invade the choroid, though there was a small focus of extramedullary hematopoiesis in the choroid. The gliotic epiretinal membrane with mostly regressed tumor was present over the edge of the optic nerve. However, there was no tumor invasion of the optic nerve including the optic nerve head and surgical margin; thus, there were no histologic high-risk features.
The tumor was classified as group D AJCC stage cT2b. There was complete vitreous seeding at the time of diagnosis, predominantly classified as clouds (class 3) morphologically. Clinically, the patient was initially treated with intra-arterial chemotherapy for 1 cycle with a very poor response, so secondary enucleation of the right eye was done.
Genomic sequencing in the AH taken at diagnosis showed an amplification in a locus on chromosome 2 of MYCN with a ratio to the median of 4.42 (Fig. 5, which also shows results for cases 1 and 2), and AH mutational analysis was negative for an RB1 mutation. The patient’s serum tested negative for a germline RB1 mutation. Genomic analysis of tumor tissue confirmed the same MYCN amplification on chromosome 2.
MYCN is a known oncogene overexpressed in tumors of the central nervous system (such as retinoblastoma) and also in non-neuronal tumors [14, 15]. There was MYCN amplification in all three of our cases (Fig. 5). This gene encodes the MYCN protein, responsible for cell growth and metabolism, with dysregulation resulting in tumor proliferation [14, 15]. Rushlow et al.  analyzed 1,068 unilateral nonfamilial retinoblastoma tumors from two separate laboratories using different cohorts and technologies in Toronto and Amsterdam. They compared tumors with no evidence of RB1 mutations (RB1+/+) with tumors carrying a mutation in both alleles (RB1−/−). No RB1 mutations were reported in 29 of the unilateral tumors, and 15 of these 29 had high-level MYCN oncogene amplification (RB1+/+/MYCNA). These RB1+/+/MYCNA tumors were classified as retinoblastomas as they expressed embryonic cell markers consistent with a retinal origin. Importantly, these tumors expressed full-length functional RB1 protein, fewer overall genomic copy number changes in genes associated with retinoblastoma, and distinct aggressive histologic and clinical features compared to prototypical RB1−/− retinoblastoma tumors .
However, Ewens et al.  challenged Rushlow’s claim that Rb protein in RB1+/+/MYCNA is fully functional and able to bind E2F, its primary target that is necessary for driving the cell cycle . Those authors found that in retinoblastomas without coding sequence mutations in one or both alleles of RB1, Rb protein demonstrated phosphorylation at its S608 residue. Phosphorylation of the Rb protein at key residues, including S608/S612, has been shown to inhibit interaction with the transactivating domain of E2F . This is a common mechanism of Rb protein inactivation in most known cancers . Thus, Ewens et al.  proposed that MYCN amplification alone is not sufficient for retinoblastoma tumorigenesis but rather that Rb inactivation is still a necessary initiating event. It should be noted that mutations alone are not the only inactivating mechanisms of the RB1 gene; epigenetic alterations with promotor methylation are a well-described mechanism of RB1 silencing, without detectable mutations in the gene.
One study showed that following Rb protein loss, there was evidence for Myc-dependent E2f3 accumulation, resulting in unregulated cell proliferation . The hypothesis that MYCN is potentially recruited in the Rb-deficient cell cycles may explain the high amplification of MYCN in some retinoblastoma tumors, though more research must be done to confirm this.
Our case 1 RB1+/+/MYCNA tumor was classified by AJCC criteria as an advanced intraocular tumor. Gross appearance of the right eye showed leukocoria, neovascular vessels for nutrition, and enlargement of the eye due to high pressure or buphthalmos. The tumor grew outward into the vitreous and subretinal space, manifesting as a white mass with avascular tumor deposits. The aggressive nature of this tumor was consistent with RB1+/+/MYCNA tumors described by Rushlow et al.  Additionally, the early age of diagnosis at 4 months was consistent with the median age of 4.5 months for RB1+/+/MYCNA tumors analyzed by Rushlow et al. , compared to 24 months for RB1−/− tumors. Ewens et al.  also found an earlier age of diagnosis at a median of 9.5 months for MYCN-amplified tumors compared to MYCN-low tumors. Pathologic evaluation confirmed both the characteristic features and intact RB protein staining in the tumor.
MYCN amplification is the primary initiating event in a small number of tumors and is also a known recurrent genomic alteration in retinoblastoma where it is a poor prognostic biomarker . These genomic events are thought to drive further tumorigenesis in RB. MYCN is in no way unique to retinoblastoma and is in fact amplified in other adult and pediatric malignancies. An example with similar features to RB is pediatric neuroblastoma. Children with neuroblastoma with MYCN amplification are known to have high-risk disease with worse outcomes . Recent in vivo work using the AH for RB similarly suggests this to be a poor prognostic biomarker, regardless of the status of the RB1 gene.
Our case 2 tumor was similarly large and invasive, as with case 1, and was diagnosed somewhat early at 9 months of age. Due to the diffuse nature of this tumor (>3 mm) and extensive subretinal seeding, the tumor was staged unfavorably by the International Intraocular Retinoblastoma Classification (IIRC) system. There was no RB1 gene variant identified in peripheral blood. RB1 analysis of the AH demonstrated a pathogenic variant. This is consistent with the histopathologic analysis with RB protein staining, confirmed that this is a tumor with lack of a functional RB protein. Additionally, the tumor was high-risk with massive choroidal invasion and preliminary invasion of the optic nerve. However, it had a very small amount of necrotic tumor and was RB protein negative in the tumor, indicating an abnormal RB protein.
Our case 3 patient had the latest age of diagnosis at 18 months of all our patients, though this was still reduced compared to reported tumors without MYCN amplification [6, 16]. The tumor was staged the same as our case 2 patient, group D stage cT2b, due to complete seeding. However, in contrast to our case 2 patient, this tumor did not show any choroidal or optic nerve invasion. The vitreous seeds were morphologically described as primarily class 3 clouds . Clouds characteristically occur with unilateral disease in relatively older children at a median age of 32 months, possibly due to the architecture of vitreous in older patients or prolonged duration of active vitreous disease . This tumor also had very small amount of necrotic tumor and was RB protein negative by immunohistochemistry, consistent with a tumor with an inactive RB1 gene and lack of RB protein. The patient did not demonstrate a germline RB1 mutation. Diagnostic AH testing for somatic variants was negative for RB1 mutation. As discussed, it is possible that the gene was silenced by epigenetic mechanisms.
The most common single copy number alteration (SCNA) among the cases was a focal MYCN amplification on chromosome 2p as indicated in Figure 5. More specifically, our case 1 tumor showed a high-level amplification of 2p24.3 on MYCN, which is an SCNA that has been identified in other reported MYCN-amplified tumors that lacked mutations in RB1 [6, 24, 25]. This provides supporting evidence for a recurrent genetic alteration in MYCN that may drive tumor progression in a subset of the retinoblastoma population. Targeted disruption of these gain-of-function MYCN alterations that may be crucial for retinoblastoma cells to survive may be a useful therapeutic target. Other SCNAs are thought to be prognostic for poor response to therapy and likelihood of intraocular relapse; specifically, 6p gain has been identified as a poor prognostic biomarker [26, 27]. Aside from focal MYCN amplification, case 3 also demonstrated 1q gain, and no case demonstrated 6p gain or other chromosomal alterations.
Management of retinoblastoma includes a combination of modalities: enucleation, local/systemic chemotherapy, laser therapy, cryotherapy, plaque brachytherapy, and external beam radiotherapy . It is only in the last several years that a discussion has emerged regarding identification of molecular tumor markers and an approach to therapy that considers these features as well as clinical aspects such as tumor size and the extent of globe destruction.
Rushlow et al.  discussed that these unilateral aggressive RB1+/+/MYCNA retinoblastomas are typically characterized by poor outcomes; prompt removal of the unilateral affected eye proved more successful compared to attempts to salvage the eye. However, until recently, information regarding MYCN amplification was not available to the surgeon to help make this diagnosis as tissue biopsy is strictly avoided in RB. The use of the aqueous as a liquid biopsy overcomes this contraindication to biopsy and facilitates detection of molecular biomarkers which can aid the clinician in these decisions. As shown with this case series, as with other solid tumors, MYCN amplification appears to impact prognosis for globe salvage whether the RB1 gene and its protein byproduct are functional or not. These tumors tend to be larger and more invasive at an earlier age of diagnosis, poorly responsive to therapy, and with extensive necrosis. Enucleation was conducted in all 3 patients described in this case series, including one secondarily after poor clinical response to therapy. With the advent of promising therapeutics for MYCN-amplified neuroblastomas including proteasome inhibitors , similar therapeutics may also be explored in MYCN-amplified retinoblastomas.
Genetic counseling with a specialist is recommended for all families affected by retinoblastoma. RB1 mutational analysis of the serum is paramount as it impacts the child’s risk for second tumors and to other family members. Somatic mutations may be detected from the tumor or the AH. There can be a lack of pathogenic variants in the RB1 gene and still loss of a functional protein, including from methylation of the promotor region , this is not heritable. As discussed herein, primary MYCN amplification is another mechanism of tumorigenesis associated with nonhereditary disease, with normal population risks for retinoblastomas in the unaffected eye, other cancers, and familial risk.
In summary, primary MYCN oncogene amplification mutations account for ∼2% of unilateral retinoblastomas, which may have significant relevance to patients. The high-risk RB1+/+/MYCNA retinoblastoma case described here (case 1) differs from classic retinoblastoma; however, the MYCNA seen in cases 2 and 3 were in the context of a nonfunctional RB protein yet also exhibited aggressive behavior, underlying the importance of detecting this alteration. Importantly, cfDNA from the AH may be used as a companion diagnostic liquid biopsy to identify MYCN-amplified RB cases. The aggressive nature of retinoblastoma with high MYCN amplification suggests that MYCN may be an important driver of malignancy, regardless of RB1 status.
We thank the families of these 3 patients for their participation in research into new ways to diagnose this rare pediatric ocular cancer.
Statement of Ethics
This article complies with internationally accepted standards for research practice and reporting. This study was approved by IRB CHLA-17-00248 Retinoblastoma Patient Clinical Database and Tissue Biorepository and the Emory IRB 00000964 Retinoblastoma Tissue Biorepository with written and informed consent obtained from the patients’ parents/legal guardians which includes publication of the details of their medical case including any accompanying images.
Conflict of Interest Statement
H.E.G., S.P., and PP have no conflicts of interest to declare. J.L.B. and L.X. have applied for patents in Aqueous Humor Cell-Free DNA for Diagnostic and Prognostic Evaluation of Ophthalmic Disease 62/654,160.
This study was funded by NIH P30EY06360 (HEG), an unrestricted departmental grant from Research to Prevent Blindness (H.E.G), NIH K08CA232344 (J.L.B.), the Wright Foundation (J.L.B.), Children’s Oncology Group/St. Baldrick’s Foundation (J.L.B.), Danhakl Family Foundation (J.L.B.), the Knights Templar Eye Foundation (J.L.B., L.X.), Hyundai Hope on Wheels (J.L.B.), A. Linn Murphree, Chair in Ocular Oncology (J.L.B.), the Berle & Lucy Adams Chair in Cancer Research (J.L.B.), the Larry and Celia Moh Foundation (J.L.B.), an unrestricted departmental grant from Research to Prevent Blindness (J.L.B., L.X.).
Conception or design of the work; the acquisition, analysis, or interpretation of data for the work; drafting the work or reviewing it critically for important intellectual content; final approval of the version to be published, Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved: Sarah Joseph, Sarah Pike, Chen-Ching Peng, Brianne Brown, Liya Xu, Jesse L. Berry, Patricia Chévez-Barrios, G. Baker Hubbard, and Hans E. Grossniklaus.
Data Availability Statement
The data for this study are available from Drs. Berry and Grossniklaus. Data are not publicly available due to ethical reasons. Further inquiries can be directed to the corresponding author.