Cellular Origins of Retinoblastoma

It is widely believed that the molecular and cellular features of a tumor reflect its cell of origin and can thus provide clues about treatment targets. Since 1897 when Wintersteiner first proposed that rosettes in retinoblastoma are reminiscent of photoreceptor differentiation, the cell of origin and the cellular features of retinoblastoma have been debated. Initially the focus was on the cellular features that could be identified by histologic analyses, such as the presence of rosettes. With advances in molecular biology and biochemistry, the debate about the retinoblastoma cell of origin has been revisited, with a strong predilection toward photoreceptors being the cell of origin. In addition to photoreceptors, a number of independent studies suggested that retinoblastomas may also have features of progenitor cells or interneurons such as amacrine cells. The possibility of tumor cells with a hybrid molecular/cellular phenotype was not considered in those previous studies.

In a recent study of the molecular, cellular, neuroanatomic, and neurochemical features of mouse and human retinoblastoma, it was shown that they have features of multiple cell classes, principally amacrine/horizontal interneurons, retinal progenitor cells, and photoreceptors. Indeed, single-cell gene expression array analysis showed that these multiple cell type-specific developmental programs are coexpressed in individual retinoblastoma cells, which creates a progenitor/neuronal hybrid cell. It is widely believed that the initiating genetic lesion in the RB1 gene occurs in proliferating retinal progenitor cells during fetal development. Retinal progenitor cells may be the cell of origin for retinoblastoma, and the multiple differentiation programs that are found in human and mouse retinoblastomas may reflect the multipotent competence of retinal progenitor cells. Indeed in single-cell gene expression array analyses of mouse retinal progenitor cells, photoreceptor and interneuron genes were often expressed in proliferating retinal progenitor cells.

The finding that retinoblastoma tumor cells express multiple neuronal differentiation programs that are normally incompatible in development suggests that the pathways that control retinal development and establish distinct cell types are perturbed during tumorigenesis. This discovery underscores the challenges of assigning cell-of-origin status without first establishing the full molecular and cellular signature of a tumor. It is important to emphasize that these studies do not necessarily indicate that a photoreceptor, interneuron, or progenitor cell is the cell of origin for retinoblastoma. Detailed cell-lineage studies combined with live imaging approaches will be required to definitively identify the cell of origin for retinoblastoma. Nonetheless, this detailed characterization of the neuronal pathways that are deregulated in retinoblastoma may provide novel targets for therapeutic intervention. This example highlights the importance of comprehensive molecular, cellular, and physiologic characterization of human cancers with single-cell resolution as we incorporate molecular targeted therapy into treatment regimens.

Retinoblastoma Tumor Cell Differentiation

Many of the previous studies on retinoblastoma tumor cell differentiation have focused on photoreceptor features, including a recent study linking the origin of retinoblastoma to cone photoreceptors. As previously described, a detailed cross-species gene expression array analysis indicated that retinoblastomas express molecular features of photoreceptors and that the rod signature is more robust than the cone signature. However mor­phometric and neuroanatomic studies of mouse and human retinoblastomas showed few cellular features of photoreceptors, which raises the possibility that the gene regulatory network that controls photoreceptors is deregulated in human retinoblastomas. Interestingly a large number of photoreceptor genes are also expressed in medulloblastomas despite their distinct cellular origins, which may indicate that the transcriptional pathways involved in photoreceptor differentiation are nonspecifically deregulated in a variety of tumors of the central nervous system (CNS). It is not known if this “photoreceptor signature” has any prognostic or therapeutic significance in retinoblastoma or other tumors of the CNS.

Human and mouse retinoblastomas have molecular, morphometric, neuroanatomic, and neurochemical features of interneurons, including amacrine and horizontal cells. Indeed, the cell-type specific features that are most common across human and mouse retinoblastomas are those related to amacrine cell differentiation. As previously discussed, the tumor’s cell of origin may be completely independent of amacrine interneurons, and because tumor suppressor or oncogenic pathways are deregulated during tumorigenesis, the amacrine differentiation program may be aberrantly activated during retinoblastoma tumorigenesis.

Nonetheless, the amacrine differentiation of human and mouse retinoblastoma may be functionally significant for tumor survival and progression. The most striking evidence for amacrine differentiation at the cellular level came from the identification of dense core vesicles in human and mouse retinoblastomas in electron micrographs. Retinoblastomas produce catecholamines and other monoamines, and several of the receptors for monoamines are expressed, as are the genes involved in the biosynthesis of monoamines. When monoamine receptors were blocked with pharmacologic agents, retinoblastoma growth was reduced in culture and in vivo. In contrast, agents that block the other major neurotransmitter signaling pathways (γ-aminobutyric acid, glycine, and glutamate) had no effect on retinoblastoma, nor did the inhibition of selective reuptake of catecholamines with pharmacologic agents. One interpretation of these data is that retinoblastoma cells are engaged in an autocrine induction of the catecholamine-signaling pathway. All of the major catecholamine/monoamine receptors are G-protein–coupled receptors and signal through adenyl cyclase–mediated cyclic adenosine monophosphate induction. This pathway can directly or indirectly stimulate the mitogen-activated protein kinase pathway and provide an important mitogenic signal for retinoblastoma. It may also explain why retinoblastomas form in the retina. That is, this tissue is uniquely susceptible to tumorigenesis because the amacrine-differentiation pathway can be co-opted for mitogenic signaling after RB1 gene inactivation. Additional studies will be required to determine whether catecholamine signaling influences survival and growth through the mitogen-activated protein kinase pathway or if it is required for some other aspect of tumorigenesis.

Retinoblastoma Genomics and Epigenomics

Most retinoblastomas initiate with biallelic loss of the RB1 gene. RB1 inactivation confers limitless replicative potential to retinoblasts, and these preneoplastic cells can progress to retinoblastoma by acquiring additional somatic mutations that contribute to the acquisition of new cellular properties, including evasion of cell death and senescence, sustained angiogenesis, and activation growth–signaling pathways. Several different mechanisms have been proposed to explain the rapid progression of retinoblastoma after RB1 inactivation. In a series of studies using genetically engineered murine cells and immortalized human cells, it was shown that RB1 may play an important role in maintaining genomic stability. Thus in some cellular contexts, inactivation of the RB1 gene could lead to chromosome instability, allowing secondary and tertiary mutations in key cancer pathways to be rapidly acquired. Alternatively RB1 has also been implicated in a variety of epigenetic processes, so it is also possible that perturbations in the epigenetic landscape may contribute to tumorigenesis in the retina. In support of an epigenetic mechanism, recent whole-genome sequencing and integrated epigenetic analysis of human retinoblastoma revealed that the tumors have relatively stable genomes, and several cancer genes were epigenetically deregulated. At least one of those epigenetically deregulated genes (spleen tyrosine kinase; SYK ) is required for retinoblastoma tumor cell survival in vivo (discussed later). These two alternative mechanisms (genome instability and epigenetic deregulation) of retinoblastoma progression are not necessarily mutually exclusive, and some tumors may show evidence of both chromosomal instability and epigenetic deregulation.

Although epigenetic deregulation of key cancer pathways has been shown to be directly relevant for retinoblastoma tumor progression, the functional significance of recurrent secondary genetic lesions is still poorly understood. During the 27 years since the RB1 gene was cloned, researchers have focused on identifying genetic lesions in retinoblastoma that contribute to tumor progression after RB1 inactivation. Cytogenetic and array comparative genome hybridization studies have led to the identification of regions of the genome that are gained or lost in retinoblastomas and may contribute to tumorigenesis. Those studies have led to the identification of candidate oncogenes and tumor suppressor genes whereby copy number variations (CNVs) correlate with changes in gene expression. However the major limitation of those studies is their relatively small number of tumors analyzed and the modest effects on gene expression for tumors with CNVs versus those without the corresponding lesion. Even for the most common recurrently mutated gene in retinoblastoma, BCL6 corepressor (BCOR) , there have been no direct in vivo studies showing that it contributes to tumorigenesis.

Carefully designed in vivo experiments to explore the role of putative oncogenes and tumor suppressor genes in retinoblastoma tumorigenesis are important for advancing our understanding of the molecular mechanisms of retinoblastoma progression. For example, a recent study showed that in a small proportion (1.5%) of human unilateral nonfamilial retinoblastomas, no RB1 gene mutations are present and they have v-myc avian myelocytomatosis viral oncogene neuroblastoma derived homolog (MYCN) amplification (≥10 copies). The authors suggested that in those patients, MYCN amplification may initiate tumorigenesis. This observation was recently validated and extended in an independent study with a separate cohort of patients.

Recent whole-genome sequencing of human retinoblastomas and their matched germline DNA demonstrated that at least some retinoblastomas have relatively stable diploid genomes with few CNVs or somatic nucleotide variations, and thus genome instability may not be required for retinoblastoma progression. One of the most common misconceptions related to cancer genomics is that tumors with few CNVs have stable genomes and tumors with more CNVs have unstable genomes. Chromosome instability is a dynamic process that cannot be accurately measured at a single time point because it involves the acquisition of sequential chromosomal lesions over time. It is important to distinguish between such dynamic processes that reflect the continuous accumulation of genetic lesions versus more acute genomic events such as multiple chromosome trisomies as seen in hyperdiploid acute lymphoblastic leukemia (ALL), multiple chromosome loss as seen in hypodiploid ALL, and the more recently described process of chromothripsis. The best way to directly analyze chromosome instability is to sample the same tumor at multiple time points and perform a comprehensive analysis of the genetic landscape at each time point. Orthotopic xenografts of human retinoblastoma in immunocompromised mouse eyes have proved to be very useful for analysis of chromosome instability in retinoblastoma because the genomic landscape can be analyzed sequentially over time. Clearly this does not preclude the possibility that some retinoblastomas have unstable genomes, but genome instability is not a universal hallmark of RB1 -deficient retinoblastomas, nor is it required for rapid tumor progression.

In the most recent genomic analysis of retinoblastoma, single nucleotide polymorphism 6.0 analysis was performed on 94 human retinoblastomas and matched normal germline DNAs. These data were used to identify regions of loss of heterozygosity and copy number changes, including whole chromosomes, regional changes, and focal changes (<3 Mb). Specifically, the majority (70%) of retinoblastomas have relatively few (≤10 per tumor) chromosomal, regional, or focal CNVs. No correlation was found between the number of CNVs and the heritable or sporadic form of disease, nor was there any relationship between the type of RB1 mutation and the number of lesions. A much larger study will be required to determine if more subtle associations exist between the clinicopathologic features of retinoblastoma and the rate of CNVs.

The overall low rate of CNVs was consistent with the paucity of focal recurrent lesions in genes. As reported previously, inactivation of the RB1 gene and the BCOR gene were the most common deletion events, and amplification of the MYCN gene was the most common focal recurrent gain. In this most recent study, a new recurrent focal amplification of orthodenticle homeobox 2 (OTX2) was discovered in 3% of retinoblastomas. OTX2 is a homeobox gene that is involved in photoreceptor and retinal pigment epithelium development. Retinoblastomas express a variety of rod and cone photoreceptor genes, and in future studies, it will be important to determine if OTX2 plays a role in modulating the photoreceptor differentiation program in retinoblastoma.

One of the most important aspects of the study was analysis of the relationship between mutations in MYCN, BCOR, and OTX2 and the mechanism of RB1 inactivation by sequencing all 27 exons of RB1 in the cohort. In total, 10 tumors were identified that had no single nucleotide variations or insertions or deletions in the coding region of RB1 . Whole-genome sequencing on those 10 tumors and their matched normal germline DNA led to the identification of one tumor with a wild-type RB1 gene and MYCN amplification. That tumor sample showed robust RB1 nuclear protein expression, suggesting that retinoblastoma can initiate in the absence of RB1 mutation. Gene expression array analysis showed that the tumor with wild-type RB1 was indistinguishable from all other retinoblastomas with RB1 mutations. Those data confirm a previous report from the Gallie laboratory that a small subset of retinoblastomas (~1% to 2%) have wild-type RB1 . More importantly this larger more recent genomic analysis also identified three tumors that had focal chromothripsis, disrupting the RB1 gene and leading to loss of RB1 protein expression. The chromothriptic lesions are not detected by routine genetic analyses because all the exons are present in the genome. This is the first example of chromothripsis in the RB1 gene in human cancer and the first clear example of chromothripsis as a mechanism for tumor initiation in a well-defined developmental cancer—retinoblastoma. Whole-genome sequencing combined with break-apart fluorescence in situ hybridization analysis of the RB1 locus and RB1 immunohistochemistry (IHC) can be used to identify this unique subset of retinoblastoma tumors.

Retinoblastoma Preclinical Models

Mouse models of cancer have become increasingly important in the fields of cancer genetics and translational research. The importance of mouse models is particularly true for pediatric cancers because the patient population is relatively small and preclinical models are essential for validating the efficacy of new combinations of chemotherapy before initiating clinical trials in children. In addition, animal models of cancer can provide important new insight into the molecular, cellular, and genetic mechanisms underlying tumor initiation and progression.

One challenge of modeling pediatric cancer in mice is that these tumors initiate in the context of developing tissues that change rapidly, and the cells that give rise to the tumors can display an extraordinary degree of plasticity and heterogeneity. This situation is further complicated by the fact that many tumor suppressor genes and protooncogenes play essential roles in regulating cell-fate specification and differentiation during development. Specifically, the genetic lesions that contribute to tumor initiation and progression may also alter the intrinsic cell-fate specification and differentiation programs in the tumor cells, thereby making it very difficult to infer the cell of origin for that tumor.

Retinoblastoma is a relatively simple tumor that initiates with a common genetic lesion ( RB1 inactivation) and progresses rapidly in children. Moreover many features of retinal development are also conserved across species, making retinoblastoma an ideal tumor for cross-species comparison.

During the past decade a series of knockout mouse models of retinoblastoma have been generated by conditionally inactivating multiple Rb family members in the developing retina. Knockout mouse models of retinoblastoma have also been valuable for studying the contribution of other tumor suppressor pathways such as the p53 pathway and for testing novel therapeutic agents for the treatment of retinoblastoma. In one study six different strains of mice that developed retinoblastoma were analyzed side by side using the same retinal progenitor–specific Cre transgene ( Chx10-Cre ) to inactivate tumor suppressor genes in the developing retina. Histopathologic analysis, gene expression profiling, and morphometric, neuroanatomic, and neurochemical analyses showed that mouse retinoblastomas faithfully recapitulate the molecular and cellular features of human retinoblastomas. However whereas the timing of retinoblastoma initiation was indistinguishable across the six strains, tumor penetrance and the rate of progression varied dramatically. These data raise the possibility that the genetic and epigenetic changes that accompany human retinoblastoma progression may not be faithfully recapitulated in the mouse, despite the remarkable interspecies similarities at the molecular and cellular level.

To more directly assess the similarities and differences across species for retinoblastoma, a recent study analyzed the aneuploidy, CNVs, somatic nucleotide variations, and epigenetic landscape of murine retinoblastoma. Similar to human retinoblastoma, mouse tumors have a low rate of single-nucleotide variations in genes. However mouse retinoblastomas have a higher rate of aneuploidy and regional and focal copy number changes, and this is dependent on the genetic lesions that initiate tumorigenesis in the developing murine retina. In addition the epigenetic landscape in mouse retinoblastoma was significantly different from human tumors, and some pathways that are candidates for molecular targeted therapy for human retinoblastoma such as SYK or myeloid cell leukemia–1 are not deregulated in genetically engineered mouse models (GEMMs). Taken together these data suggest that important differences exist between mouse and human retinoblastomas with respect to the mechanism of tumor progression, and those differences can have significant implications for translational research to test the efficacy of novel therapies for this devastating childhood cancer.

In fact, preclinical testing of combination chemotherapy (etoposide, carboplatin, and vincristine) showed dramatic differences in response between GEMMs and human orthotopic xenografts; virtually all of the GEMMs were cured of their disease, whereas no improvement was seen in progression-free survival or overall survival in the orthotopic xenograft model. It is possible that such species-specific differences could be further amplified when testing molecular targeted therapeutics directed toward processes important for maintaining genomic stability or the epigenetic landscape of retinoblastoma or targeting pathways such as SYK/myeloid cell leukemia–1 that are not deregulated in murine retinoblastoma. A careful assessment of the pathways that are deregulated in murine retinoblastomas is essential to make an accurate evaluation of the efficacy of novel therapeutics in preclinical studies focused on selecting the most promising new drugs for testing in clinical trials.

Understanding the intra- and interspecies differences is paramount when choosing the right model for the study of retinoblastoma. Unlike orthotopic xenograft models, genetic mouse models present the advantage of understanding the developmental processes that support tumor formation arising from a single cell and provide an ideal tool for genetic characterization of the contribution of individual pathways in retinoblastoma progression. On the other hand, orthotopic xenografts from patients faithfully recapitulate the mechanisms by which gene deregulation occurs in the human disease and can be more accurate for testing novel therapeutics. As a result, orthotopic xenografts are an important tool for target identification and target validation during preclinical trials, and GEMMs are important for understanding tumor initiation and progression in the developing retina.

Retinoblastoma Translational Research

Worldwide retinoblastoma is diagnosed in approximately 8000 children each year, including 250 to 300 patients in the United States. In developed countries more than 90% of patients with localized retinoblastoma disease are cured using a combination of chemotherapy, focal therapies, surgery (enucleation), and radiation therapy. However many patients in developing countries present with advanced intraocular or metastatic disease. Although mortality is low with aggressive multimodal therapy, partial or full loss of vision occurs in approximately 50% of patients with advanced bilateral retinoblastoma. In addition, significant late effects of therapy occur, including facial deformities and increased incidence of secondary malignancies. A vigorous effort is now under way to develop targeted therapies to improve ocular salvage and vision preservation and reduce late effects of therapy without compromising therapeutic outcomes.

Major advances have recently been made in understanding the biology of retinoblastoma, including identification of several molecular targets such as the inhibition of the MDM4-p53 interaction, SYK, histone deacetylase (HDAC), and the B-cell lymphoma–2 (BCL2) family of proteins. MDM4 is overexpressed in most human retinoblastomas and has been shown to contribute to tumor formation by suppressing p53. Consequently specific inhibition of the MDM4-p53 interaction represents a promising drug target for retinoblastoma treatment. The small-molecule MDM2/MDM4 antagonist nutlin-3a has been shown to efficiently induce p53-mediated cell death in retinoblastoma cells and mouse models of retinoblastoma. Subsequent characterization of the genetic and epigenetic landscapes of retinoblastoma revealed profound increases in expression of the protooncogene SYK . Although SYK is not expressed in the normal human retina, it is upregulated in all retinoblastomas analyzed to date. SYK is also expressed at high levels in metastatic and posttreatment retinoblastomas (Brennan and Bahrami, unpublished data). Many small-molecule SYK inhibitors with diverse physiochemical properties are in development for rheumatoid arthritis and oncology applications, and some of these novel therapeutics may eventually benefit patients with retinoblastoma. Moreover, previous studies of SYK inhibition have implicated a number of downstream signaling molecules, including the BCL2 family of proteins, as mediators of the SYK survival signal. Thus small-molecule BCL2 inhibitors currently in development for other cancers, such as Obatoclax, ABT-737, and TW-37, may also prove to be effective therapeutic agents for retinoblastoma. Retinoblastomas are also highly sensitive to HDAC inhibition, and many HDAC inhibitors are currently in clinical trials.

Retinoblastoma is unique because several well-established routes of drug delivery exist that provide flexibility when considering novel therapeutics. Systemic routes can be used for some drugs but are typically limited by poor penetration across the blood-retinal barrier (BRB) and systemic toxicity concerns. Local ocular delivery routes include topical (transcorneal), periocular (transcleral), intravitreal (direct injection) and intraarterial infusion. However all routes are not amenable to all compounds because of the physiochemical property of the compounds, the effects on normal tissues, and the schedule of drug administration. Some chemotherapy drugs such as topotecan effectively cross the BRB, and equivalent intraocular pharmacokinetic (PK) profiles result from either systemic or local topotecan. Topotecan is routinely administered daily for 5 days in children with cancer, providing repeat tumor exposure in an outpatient setting. In contrast carboplatin and nutlin-3a cannot be administered systemically because of insufficient penetration of the BRB, and in these cases, subconjunctival injections have been a significantly more effective delivery route. However subconjunctival injections are only performed during examination with use of an anesthetic once every 3 weeks, precluding the opportunity for repeat dosing of the drug. Moreover reports of significant and dose-limiting local ocular toxicities of carboplatin have prevented widespread use of subconjunctival delivery for children with retinoblastoma. These studies highlight the importance of pharmacokinetics and toxicokinetics in drug-specific route selection for intraocular targets.

Because of the young age of the patient population and the relatively rarity of the disease, preclinical testing is a critical component of developing retinoblastoma therapeutics. A standardized approach for evaluating efficacy and safety using mouse genetic models of retinoblastoma and human orthotopic xenograft models has been developed, but the large number of potential compounds, routes, and formulations precludes preclinical testing of all possible combinations. Previous work suggests that, given the unique challenges of drug delivery to the eye, one way to streamline the candidate selection process is to determine which route and formulation achieve optimal PK before conducting preclinical studies. When administered using the routes and formulations that, according to PK studies, achieved target intraocular exposure, topotecan, nutlin-3a, and carboplatin were all efficacious. In contrast, despite promising in vitro cytotoxicity, subconjunctival delivery of a SYK inhibitor failed to provide efficacy in preclinical studies; this failure was related to insufficient exposure. However alternative formulations and routes of local delivery did improve PK compared with systemic delivery; intraocular concentration never reached levels required for cytotoxicity.

In conclusion, effective translational research for retinoblastoma is built on a foundation of outstanding basic research to advance our understanding of the pathways that are deregulated in this devastating childhood cancer. From there, suitable targets for therapy are identified, such as MDM4, HDAC, or SYK, and small molecules that disrupt signaling through those pathways are identified and tested for in vitro cellular effects. Next pharmacokinetic, pharmacodynamic, and toxicokinetic studies are combined with approaches to optimize formulation of the drugs for the different routes of delivery. Finally when suitable exposure is achieved with minimal toxicity, comprehensive preclinical testing using validated preclinical models is performed to compare efficacy of the new combination with current standard of care and other possible novel therapeutic approaches. Variations in schedule and integration with upfront standard of care therapy can also be tested at this stage to achieve an optimal route, dose, and schedule for a new retinoblastoma clinical trial.

Surgery

Enucleation is indicated for large tumors filling the vitreous for which there is little or no likelihood of restoring vision, and in cases of tumor present in the anterior chamber or in the presence of neovascular glaucoma. This corresponds to group E eyes and a significant proportion of group D eyes; upfront enucleation should be offered to those patients. Enucleation should be performed by an experienced ophthalmologist. The eye must be removed intact, without seeding the malignancy into the orbit and avoiding globe perforation. For optimal staging, a long section (10 to 15 mm) of the optic nerve needs to be removed with the globe. An orbital implant is usually fitted during the same procedure, and the extraocular muscles are attached to it. In the past, orbital implants were avoided because it was believed that they would interfere with the palpation of the socket and clinical detection of orbital recurrence. However with improved understanding of the histologic risk factors and the availability of better imaging techniques to detect orbital disease, implants should be placed at the time of the enucleation. Different orbital implants are available, including polymethylmethacrylate, polyethylene, and coralline and bovine hydroxyapatite spheres. A tissue wrap to the implant is placed, which will allow the four rectus muscles to be anatomically reattached, thus providing implant motility with little resistance in the orbit. The size and type of implant are important to stimulate orbital growth. A ceramic false eye is later fitted in the orbital socket. Orbital exenteration is very seldom indicated, although it should considered in cases of tumor recurrence after radiation. For patients presenting with orbital disease, a judicious use of chemotherapy, surgery, and radiation therapy will result in effective tumor control, avoiding the need for orbital exenteration.

Focal Therapies

Focal treatments are used for small tumors (less than 3 to 6 mm) and in combination with chemotherapy. In this setting of a patient undergoing chemotherapy for cytoreduction, focal treatments should be applied to tumors that fail to calcify; it is usually recommended to allow maximal reduction in tumor size with one or two cycles of chemotherapy prior to proceeding with aggressive focal consolidation to minimize the damage to the surrounding retina, particularly when the tumor is close to visually critical structures, such as the fovea and the optic nerve. The focal therapies available to treat retinoblastoma include argon green laser, diode laser, cryotherapy, and brachytherapy. The use of each modality depends on the tumor location and size. Lasers are more commonly used to treat posterior tumors, whereas more anterior tumors are easily accessed with cryotherapy. Brachytherapy is reserved for tumors that cannot be consolidated with either of those two approaches, usually because of their size.

Two wavelengths of light are used to treat retinoblastoma. The argon green laser, with a wavelength of 532 nm, photocoagulates tissue by inducing a temperature in excess of 65°C; it has been traditionally used to treat the retina edge surrounding a tumor in an effort to deprive the tumor of its blood supply and subsequently induce tumor regression, and for the treatment of retinal neovascularization due to radiation therapy. Using an indirect ophthalmoscope, a double row of white burns is used to encircle the tumor, and powers of 250 to 350 mW and burn durations of 0.3 to 0.5 seconds are used. This technique is limited to tumors measuring no greater than 4.5 mm in base and no greater than 2.5 mm in thickness. Debate exists as to whether the tumor surface should be directly treated for fear of releasing tumor cells into the vitreous. Newer generations of the 532-nm laser permit continuous delivery similar to that of the diode laser; such modalities have led some persons to advocate its use in the treatment of retinoblastoma. If the tumor surface is treated, the desired end point remains a subtle whitening of the treated areas. Laser energy should be kept at the lowest possible level necessary to achieve the desired end point; powers in excess of 500 mW will result in tumor disruption and should be avoided. As with all focal therapies, treatment should be continued until the residual fish-flesh tumor regresses into a flat chorioretinal scar. Typically treatment with lasers is administered every 3 to 4 weeks. The optimal indication for direct treatment with the argon green laser is a residual fish-flesh mass overlying a largely calcified tumor. In such situations, the uptake of the diode laser is unpredictable. With use of the argon green laser, a whitening of the tumor can be seen, ensuring an adequate treatment. The argon laser has both direct and indirect cytotoxic effects. Photocoagulation directly kills tumor cells, and the thermal spread from the laser may have secondary hyperthermic effects that act synergistically with ongoing chemotherapy. Argon laser photocoagulation should not be performed if there is a retinal detachment overlying the tumor.

The diode laser, with a wavelength of 810 nm, was adapted for the treatment of retinoblastoma based on its success in the treatment of choroidal malignant melanoma. The laser energy is readily absorbed by melanin pigment and proved successful in the treatment of small melanocytic tumors. Low-power settings coupled with a large spot size (2 to 3 mm) and prolonged delivery achieved hyperthermia (45°C to 65°C) at subphotocoagulation temperatures. The end result was a sustained penetrant burn with direct cytotoxic effects. The term transpupillary thermotherapy , or TTT, was coined to describe this technique. The methods of TTT were subsequently adapted to treat retinoblastoma. The infrared laser is focused directly on the tumor surface using either an ophthalmic microscope adapter or a large spot size indirect ophthalmoscope adapter. When using the indirect ophthalmoscope for TTT, it is important to ensure that the correct adapter is being used, because adapters designed for diode photocoagulation will not provide a continuous wavelength delivery necessary for hyperthermia. For small tumors the underlying retinal pigment epithelium aids in absorption and transfer of the heat to the tumor. The desired end point is a subtle whitening of the tumor free of hemorrhage. Powers are titrated based on the tumor size and pigmentation of the underlying retinal pigment epithelium but should not exceed 600 mW when there is visible uptake by the surrounding retinal pigment epithelium. Tumors with significant elevation or residual elements that rest on adjacent calcified tumor do not readily whiten with TTT. Prolonged treatment sessions of 5 to 10 minutes using powers of 600 to 800 mW may be needed to achieve a therapeutic effect. Even with such prolonged treatment, a visible change in the tumor may not be seen until follow-up examination 3 to 4 weeks later. If available the argon green laser may provide a more effective treatment. The sequential administration of thermotherapy with carboplatin enhances the antitumor effect by increasing the platinum-DNA adducts, for which reason thermochemotherapy is becoming a very important component in the treatment of intraocular retinoblastoma.

Cryotherapy is used for the treatment of small equatorial and peripheral lesions, measuring no more than 3.5 mm in base and no more than 2 mm in thickness. Cryotherapy is directly cytotoxic, causing cell lysis by disruption of the cell membrane that occurs after the formation of cytoplasmic ice crystals. Nitrous oxide gas is used to deliver a transscleral freeze via a cryoprobe under direct visualization. The lesion to be treated is visualized with an indirect ophthalmoscope, and the scleral underlying the tumor is indented with a cryoprobe. A succession of three freezes with three intervening thaws is applied to the tumor. An “ice ball” should be seen to encompass the tumor and adjacent vitreous with each freeze. Treatment should be limited to the area of the tumor to minimize damage to adjacent normal retina. Multiple tumors may be treated at one examination, but extensive treatments should be avoided. Aggressive cryotherapy may cause serous retinal detachments and iatrogenic retinal breaks. Although each cryotherapy unit may vary, we target a freezing temperature of −70°C during treatment. Treatments should be repeated at 3- to 4-week intervals until the lesion regresses to a flat chorioretinal scar. Posterior tumors can also be treated with cryotherapy. The conjunctiva is opened in the quadrant of the tumor to be treated. The curved cryoprobe is passed along the curvature of the eye and the tumor is indented. Once the tumor has been identified, the triple freeze-thaw is applied. This “cut-down” cryotherapy is usually reserved for recurrent posterior tumors not amenable to laser treatment. Also in addition to its effect on tumor control, cryotherapy contributes to increase the intraocular penetration of chemotherapy agents, presumably through disruption of the blood-vitreous barrier. In general local control rates of 70% to 80% can be achieved. Complications of focal treatments include transient serous retinal detachment, retinal traction and tears, and localized fibrosis.

Chemotherapy

Chemotherapy is indicated in patients with extraocular disease, in the subgroup of patients with enucleated eyes with intraocular disease and high-risk histologic features, and in patients with bilateral disease in conjunction with aggressive focal therapies. Agents with documented efficacy include microtubule inhibitors (vincristine and paclitaxel), platinum compounds (cisplatin and carboplatin), topoisomerase II inhibitors (teniposide and etoposide), alkylating agents (cyclophosphamide and ifosfamide), anthracyclines (doxorubicin and idarubicin), and topoisomerase I inhibitors (topotecan). The first clinical experience with chemotherapy in the treatment of retinoblastoma was reported with the use of nitrogen mustard in 1953. Institutional experiences in the management of extraocular retinoblastoma were reported subsequently, usually applying regimens modeled after the treatment regimens for metastatic neuroblastoma. Since then the role of chemotherapy has continued to expand, and it is now a main component in the management of intraocular disease.

Radiotherapy

Retinoblastoma is a very radiosensitive tumor. Radiotherapy in combination with focal treatments can provide excellent tumor control. However because radiation therapy increases the risk of second malignancies, contemporary management of intraocular retinoblastoma is designed to avoid or delay its use. The role of irradiation is mainly as salvage management for eyes that have failed to respond to chemotherapy and focal treatments, usually because of progression of vitreous and subretinal seeding. Because most patients with intraocular retinoblastoma who undergo radiotherapy have multifocal disease, the entire retinal surface needs to be irradiated to a uniform dose. Several techniques can be used, usually through lateral or anterior fields. Recommended total doses are 40 to 45 Gy, in 180 to 200 cGy fractions, although doses of 36 Gy and even lower may be effective in conjunction with other techniques. Radioactive plaque technique is useful when treating localized tumors, both because the procedure time is short and because a high dose of irradiation is delivered to the areas of interest while minimizing irradiation effects in extraocular structures.