Response criteria for intraocular retinoblastoma: RB-RECIST

Abstract

Standardized guidelines for assessing tumor response to therapy are essential for designing and conducting clinical trials. The Response Evaluation Criteria In Solid Tumors (RECIST) provide radiological standards for assessment of solid tumors. However, no such guidelines exist for the evaluation of intraocular cancer, and ocular oncology clinical trials have largely relied on indirect measures of therapeutic response—such as progression-free survival—to evaluate the efficacy of treatment agents. Herein, we propose specific criteria for evaluating treatment response of retinoblastoma, the most common pediatric intraocular cancer, and emphasize a multimodal imaging approach for comprehensive assessment of retinoblastoma tumors in clinical trials.

1 |. INTRODUCTION

Commonly accepted standards for assessing tumor response to new therapeutics are essential for designing and conducting meaningful clinical trials. Beginning in 1979 and continuing through the 1990s, such guidelines for solid tumors were developed based on radiological imaging—in particular, computed tomography (CT) scanning—in the field of general oncology. These guidelines are now widely referred to as RECIST, for Response Evaluation Criteria In Solid Tumors.^1,2 Ocular oncology, however, has never established such standards for assessing the response to treatment of intraocular tumors such as retinoblastoma, the most common intraocular malignancy in pediatric populations.^3,4 This is in part because the tumor is directly visible to the treating ocular oncologist via fundoscopic examination, and submillimeter changes that are unlikely to be detected on routine radiologic imaging are clinically relevant in this cancer.

Many tumor-specific oncology groups have modified the RECIST criteria to meet the needs of their specialty. For example, the Prostate Cancer Working Group outlined modifications for trial eligibility and design specific to prostate cancer,⁵ and the International Working Group incorporated bone marrow immunohistochemistry and flow cytometry into their lymphoma response assessment.⁶ If there is hope for ocular oncology to develop significant new treatment modalities that are comparable across centers and studies, it is important to initiate a dialogue in our subspecialty with the ultimate goal of creating commonly accepted criteria for accurately evaluating treatment responses.

2 |. RESPONSE EVALUATION CRITERIA IN SOLID TUMORS

2.1 |. The history of RECIST

Approximately four decades ago in general oncology, radiological standards were starting to be developed for the comparison of pre- and posttreatment tumor burdens in clinical trials. In order to assess tumor response to therapeutic agents, endpoints such as tumor shrinkage and time to tumor progression were frequently used.² However, it rapidly became clear that using tumor response as an objective endpoint was only possible when widely accepted standard response criteria were available and when image-based endpoints were appropriate surrogates for outcomes of survival. In 1979, the World Health Organization (WHO) was the first to publish response criteria to assess solid tumor size changes in response to therapeutic intervention.⁷ In the years that followed, the WHO guidelines were modified to accommodate new technologies and clarify concepts from the original publication.² These modifications ultimately resulted in confusion over the interpretation and comparability of clinical trial results.⁸ An International Working Group was formed in the 1990s to reconcile the divergent standards. The results of this group, known as RECIST, supported the use of CT images for lesion measurement and specifically discouraged the use of ultrasound imaging given the potential for interoperator variability and subjectivity.¹ Multiple prospective analyses subsequently validated the RECIST recommendations, which largely replaced the prior WHO criteria.²

Since RECIST was first published in 2000, the guidelines have been widely accepted by both industry and academia for clinical trials in which objective tumor response is an endpoint.² Amendments to the original RECIST recommendations (later designated as RECIST 1.0) were developed and published as RECIST 1.1 in 2009, including specifications for determining lymph node size, reducing the maximum number of target lesions per patient, and utilizing newer imaging modalities such as positron emission tomography and magnetic resonance imaging (MRI).² A summary of the WHO and RECIST guidelines are provided in Table 1.

TABLE 1.

Response criteria according to WHO and RECIST guidelines

Best response	WHO–change in sum of products of target tumor diametersa	RECIST–change in sum of longest target tumor diameters
CR	Disappearance of target lesions	Disappearance of target lesions
PR	50% Decrease	30% Decrease
SD	Neither PR nor PD criteria met	Neither PR nor PD criteria met
PD	25% Increase; no CR, PR, or SD documented before increased disease	20% Increase; no CR, PR, or SD documented before increased disease

Abbreviations: CR, complete response; PD, progressive disease; PR, partial response; RECIST, Response Evaluation Criteria In Solid Tumors; SD, stable disease; WHO, World Health Organization.

^aWHO guidelines recommended measuring perpendicular diameters in each of at least five target lesions (measurable lesions), multiplying the two measurements for each lesion to obtain a lesion “product,” then adding each of the five products of the target tumors in each organ to give an estimate of tumor burden in that organ.

2.2 |. RECIST and ocular oncology

Although RECIST guidelines have improved the comparability amongst clinical trials of cancer therapeutics, aspects of RECIST are poorly compatible with the standard and widely accepted techniques used in ocular oncology clinics and trials. The most glaring incompatibility is the RECIST 1.1 recommendation to avoid ultrasound for measuring tumor response due to operator dependence and difficulty with standardization. Ultrasonic modalities (in conjunction with fundus photography) are the primary imaging techniques used to safely diagnose and monitor treatment response for intraocular tumors.^9,10 Ultrasound has also been accepted as the imaging utility of choice in the modified response criteria guidelines for some solid tumors, such as subcutaneous melanoma lesions (itRECIST).¹¹ Rather than ultrasound, RECIST guidelines posit CT or MRI as the acceptable imaging modalities for tumor measurements given the high reproducibility and reliability between repeated examinations.^1,2 In the setting of retinoblastoma, CT is generally contraindicated given the risk of radiation-associated secondary cancer development in pediatric patients with a heritable tumor predisposition syndrome.^3,12 MRI is the primary neuroimaging modality for the diagnosis and staging of retinoblastoma given its superior ability to detect optic nerve invasion and extraocular disease without exposing patients to harmful radiation.^13–16 MRI cannot however evaluate extent of intraocular seeding, small increases in subretinal fluid or more subtle intraocular tumor changes, therefore clinical examination with directed ocular imaging to assess tumor response is required. The need for general anesthesia to perform MRI evaluations in young children is another limiting factor in the routine application of MRI.¹⁷ When considering tumor response evaluation for retinoblastoma, it is important to choose accessible imaging modalities that do not increase our patients’ risk for secondary malignancies or other adverse effects during clinical trials.

Given the limitations set forth by RECIST guidelines, clinical trials in ocular oncology have rarely assessed objective tumor response as a primary clinical endpoint—instead, using progression-free patient and/or eye survival to evaluate the efficacy of therapeutic agents.^18,19 With the increasing diversity of novel first- and second-line therapeutics and the emphasis on eye-salvaging modalities in recent years, it is clear that the field of ocular oncology is in need of better-defined, tumor-specific response criteria and imaging recommendations for evaluating and comparing tumor response, and thus globe salvage, in clinical trials. Ideally, modifications to RECIST would be based on series of prospective, randomized multicenter trials,²⁰ although these are often neither timely nor feasible in the setting of rare diseases such as retinoblastoma. Herein, with the consensus of clinical and research experts in ocular oncology, we discuss the primary imaging techniques used to assess retinoblastoma and propose modified criteria for determining the intraocular response to treatment of retinoblastoma in a clinical trial setting.

3 |. MONITORING INTRAOCULAR TUMOR RESPONSE

3.1 |. Imaging posterior segment tumors

Ultrasound is the mainstay for imaging retinoblastoma and has been routinely used to evaluate posterior segment lesions since the early 1970s.²¹ Because of the fluid-filled, superficial positioning of the human eye, it is well-suited for ultrasonic examination. Ultrasound utilizes high-frequency sound waves that can penetrate through tissue interfaces and reflect back to the transducer, allowing for evaluation of intraocular tumors even in the presence of visually obstructive defects such as cataracts and vitreous hemorrhage.²² It is also a significantly less expensive modality than CT or MRI²³ and involves no ionizing radiation exposure, making it ideal for performing repeated tumor imaging over time and for global translation where there is a need for easily accessible, cost-effective imaging modalities. Even for lesions located anterior to the ora serrata involving the corpus ciliaris or the iris, ultra high-frequency sound waves (35–50 MHz) represent the most reliable way to document and measure tumor burden when the anterior segment is involved.²⁴ Although nonstandardized ultrasound is largely operator dependent and may therefore have limited reproducibility compared to CT or MRI,^1,2 it has nevertheless been routinely used to obtain parameters for retinoblastoma lesions.

To optimize the accuracy and reproducibility of ultrasound in the setting of intraocular tumors such as choroidal melanoma, numerous studies have utilized more standardized methods of ultrasonic measurement to characterize tumor size. One of the largest trials in ocular oncology, the Collaborative Ocular Melanoma Study (COMS), combined contact B (brightness)-scan with standardized A (amplitude)-scan ultrasonic techniques to measure apical tumor height over time.^21,25 The contact B-scan employs a focused sound beam with a frequency of 10 MHz in order to identify tumor properties such as location, topography, lesion boundaries, and gross apical height (i.e., tumor thickness).^21,26 When a tumor is visualized in transverse orientation on B-scan, the image can be captured and a built-in caliper tool used to approximate the apical height of the tumor. Whereas tumor boundaries on B-scan are determined based on differences in signal brightness, A-scan uses a frequency of 8 MHz and a single nonfocused sound beam to generate a series of spikes whose amplitudes are proportional to the strength of the echo at tissue interfaces (with taller spikes corresponding to more distinct interfaces).^21,26 By measuring the distance between specific spikes that are consistent with the anterior and posterior margins of the tumor, the A-scan provides a potentially less subjective method of determining apical tumor height.²⁶ Unfortunately, the A-scan is less useful in the setting of retinoblastoma, as the highly calcified nature of these tumors substantially attenuates the ultrasonic beam and alters the spike pattern of the scan; therefore, B-scan is typically the sole ultrasonic tool used to monitor retinoblastoma apical height over time.²⁷

In addition to ultrasound, fundus photography is an essential imaging component that allows for direct visualization of intraocular tumors. The RetCam (Clarity Medical Systems, Inc., Pleasanton, CA, USA) is one of the most widely used photographic devices, employing a hand-held digital camera with a lens that gently contacts the cornea during imaging. The measurement capabilities of fundus photography are limited depending on the location of the tumor and the clarity of intervening structures, such as when the patient presents initially with a retinal detachment (Figure 1A). Generally, however, such imaging is useful for evaluating tumors in relation to important structures such as the optic nerve and macula as well as monitoring for changes over time. Additionally, fundus photos can help identify subtle abnormalities such as small vitreous seeding or minimally elevated lesions that might not otherwise be easily identified on ultrasound.³ By providing direct visualization, these images are also helpful in direct comparison of disease response and in reviewing progress with families. Therefore, fundus photography is an important adjunct to standard ultrasonic imaging of any posterior segment lesion.

FIGURE 1.

(A) Large retinal detachment with subretinal seeding associated with Group D retinoblastoma that is preventing fundoscopic evaluation of the posterior segment. (B) Fundus photograph of the same eye after systemic chemotherapy demonstrating a classic cream-colored, dome-shaped mass with associated vascularity and calcifications with scarring from consolidative therapy around the tumor margin. This highlights the significant changes the eye undergoes as it responds to therapy

In recent years, spectral domain optical coherence tomography (SD-OCT) has become an increasingly important component of retinoblastoma imaging. Hand-held SD-OCT is performed at bedside during examinations under anesthesia of retinoblastoma patients, allowing for precise evaluations of both tumor and seeding microstructure.²⁸ Whereas modalities such as ultrasound and fundoscopy are limited in the setting of submillimeter disease, high-resolution OCT can identify “invisible” tumors that cannot be appreciated clinically as well as infra-clinical involvement of the fovea and optic nerve.^29–32 While not every retinoblastoma treatment center has access to OCT, there may be a role for OCT-validated retinoblastoma measurements in future response guidelines. Given the unique strengths and limitations of individual posterior segment imaging techniques, a comprehensive multimodal imaging approach to retinoblastoma evaluation is ideal.

3.2 |. Response criteria for retinoblastoma (RB-RECIST)

The most common primary intraocular malignancy in pediatric populations is retinoblastoma, a tumor that develops from retinal progenitor cells in early childhood and often presents with leukocoria and/or strabismus.³ Active cancer cells proliferate from a primary retinal tumor but can extend into the vitreous and subretinal space. Each space may respond separately to therapy, and thus response criteria must consider all three anatomical areas.

Classically, ultrasound examination demonstrates a dome-shaped, retinal-based lesion (Figure 2). There may be associated subretinal fluid (exophytic growth pattern) suggestive of seeding, or vitreous seeding, which, if bulky, can be visualized on B-scan. Intralesional calcification of the primary retinal mass is pathognomonic for retinoblastoma, which usually presents as highly reflective deposits within the mass. The calcification increases with treatment; depending on the degree of calcification within the tumor, there may be significant concurrent shadowing posterior to the calcium deposits so that the posterior scleral wall is not well visualized.³³ In this setting, ultrasonic measurement of tumors is limited, which is why B-scan should not be the only modality used for objective assessment of tumor response.

FIGURE 2.

B-scan ultrasound image demonstrating a dome-shaped retinal lesion with highly reflective deposits of intralesional calcium, consistent with retinoblastoma. Yellow and green lines represent measurements of tumor height and base, respectively

Fundus photography can further demonstrate active disease in the retina, the vitreous, and the subretinal space (Figure 1B). Although there are no established quantitative criteria that define the response of both retinal lesions and vitreous seeding to therapeutic agents, previous studies have assessed retinoblastoma treatment response based on specific tumor regression patterns (including Type 0 [no tumor remnant], Type I [completely calcified remnant], Type II [noncalcified remnant with decreased vascularity], Type III [partially calcified remnant], and Type IV [flat scar])^34,35 and the presence or absence of seeding.³⁵ Vitreous seed regression patterns have also been described, including Type 0 (no seed remnant), Type I (refringent and/or calcified residues), Type II (amorphous, often nonspherical inactive residues with or without pigment), and Type III (combination of I and II).^36,37

On clinical examination, seeding of retinoblastoma tumor cells is a well-known indicator of advanced disease. In general, the presence of intraocular seeding portends a significantly increased risk of disease recurrence and the need for secondary enucleation.^38–41 Multiple studies have also demonstrated an association between the specific morphology of seeds (as seen on fundus photography) and the overall risk of treatment failure. Seeds with cloud-like or spherical morphology, for example, are particularly difficult to eradicate and often demonstrate suboptimal response to intravitreal chemotherapy injections compared to eyes with dust seeds.^{36,38,42–44} Even with the use of targeted intravitreal chemotherapy, tumor seeding remains a common cause of treatment failure and can ultimately lead to loss of the eye.^42,45

Because both the presence of seeding and lack of tumor calcification portend tumor recurrence, which impacts overall eye survival, it is important to take these unique retinoblastoma characteristics into account. Thus, we recommend the following definitions of response criteria for each of the three anatomical spaces to be used as objective endpoints in retinoblastoma clinical trials. A combination of these responses will inform the overall disease response for the eye in patients with intraocular, nonmetastatic disease.

3.2.1 |. Response criteria for tumors

• Complete response (CR): Types 0, I, or IV regression; or Types II or III regression that have demonstrated clinical stability on fundus photography and ultrasound imaging for ≥6 months after cessation of first- and/or second-line plus local consolidation therapy. No visible residual active retinal disease; calcified and/or chorioretinal scarring is stable. Further therapy not indicated.

• Partial response (PR): Decrease in apical tumor height by ≥30% from baseline and/or Types II or III regression that have demonstrated clinical stability on fundus photography for <6 months. Often local consolidation therapy is ongoing; there is no worsening of retinal disease.

• Stable disease (SD): Decrease in apical tumor height by <30% from baseline with lack of/minimal regression also seen on fundus photography or increase in apical tumor height by <30% from baseline. Consolidation therapy is ongoing or limited. Persistent disease may be present and undergoing therapy; there is no worsening of retinal disease.

• Progressive disease (PD): Increase in tumor measurements by ≥30% from tumor nadir in at least one dimension, that is, height and/or base, and/or appearance of new lesions. OCT can aid in identifying new subclinical lesions within the retina. This includes recurrent disease, defined as a new secondary growth at any location occurring after >2 event-free months following completion of first- or second-line therapies.

On initial evaluation of each case, a single target tumor should be identified and measured whenever possible. Baseline apical height and basal dimension of the tumor should be established at diagnosis prior to treatment initiation by obtaining measurements using B-scan ultra-sonography. Per standard imaging protocol, tumors should be measured from the inner sclera to the tumor apex, not including the scleral thickness.⁴⁶ Given the known presence of subretinal fluid associated with retinoblastoma tumors,⁴⁷ the tumor should be measured without the retina in these cases. At each imaging time point, the apical height and base of the target tumor should be similarly evaluated. When there is a treatment response, there will be a nadir in the measurements; subsequent increases in tumor size are compared to this nadir to assess for progression.⁴⁸ During evaluations for tumor response, the patient will receive a single categorical response to describe the overall disease status in the eye. We recommend a ≥30% reduction in tumor height from baseline to be considered PR. Additionally, we recommend a ≥30% increase in tumor size from the nadir tumor size to be considered PD. These measurements should be at least 0.5 mm, as multiple studies have demonstrated that the normal variability in ultrasound measurements is within 0.5 mm or less.^25,49,50 While these values are consistent with previously validated thresholds for solid tumors,^51–53 we encourage that these guidelines be reassessed and similarly validated for retinoblastoma in the future. Small retinal tumors below the limit of ultrasound will be better evaluated with OCT than B-scan.

Unlike other solid neoplasms, measurement alone without evaluation of the quality and borders of the lesion is insufficient for noting regression in retinoblastoma. In conjunction with ultrasound, fundus photography should be employed at each examination to determine the specific regression pattern of the retinal tumors, as well as qualities such as tumor vascularity that are less distinguishable on ultrasound. While retinoblastoma tumors typically decrease in size during therapeutic regression, they may also develop varying degrees of calcification in response to therapy.³⁴ It is important to note that as retinoblastoma tumors consolidate and calcify, they can occasionally appear larger on B-scan ultrasonography compared to pretreatment measurements, particularly if subretinal fluid and confluent seeding were present at diagnosis. Therefore, ultrasonic findings should always be considered in the context of clinical photographs. A tumor may also measure significantly less overall but have a prominent recurrence on the calcium that can be noted visually; thus, both modalities are needed to assess response.

3.2.2 |. Response criteria for vitreous seeds

• CR: Types 0 or I regression; or Types II or III regression that have demonstrated clinical stability on fundus photography for ≥6 months. Further intravitreal therapy not indicated.

• PR: Unequivocal improvement in seeding based on decreased number or density of seeds and/or Types II or III regression that have demonstrated clinical stability on fundus photography for <6 months.

• SD: Neither unequivocal improvement nor progression of seeding.

• PD: Unequivocal progression of seeding based on increased number or density of seeds, conversion from dust to spheres, or the presence of new preretinal tumors.

Although ultrasonic modalities may identify bulky intravitreal seeding, the presence or absence of seeding should always be confirmed with clinical examination and fundus photography. Digital photos should be saved at baseline and at each subsequent evaluation so that the degree of seeding may be compared throughout the study. In some cases, OCT may also be useful to assess the presence and microstructure of vitreous seeds.^28,54 Of note, a PR of vitreous seeding may have different prognostic implications depending on the morphology of seeds that remain. Multiple studies have associated residual spherical seeds with resistance to therapy and significantly increased rates of relapse, whereas eyes with residual dust or calcified seeds have much lower recurrence rates.^38,42,44 Therefore, a finding of PR alone—without additional clinical context—is not necessarily predictive of future ocular survival.

3.2.3 |. Response criteria for subretinal seeds

• CR: Disappearance of all subretinal fluid and visible subretinal seeds, or calcification of all subretinal seeds for ≥6 months.

• PR: Unequivocal improvement in subretinal seeding based on decreased number or density of subretinal seeds without complete calcification, and decreased subretinal fluid.

• SD: Neither unequivocal improvement nor progression of subretinal seeding.

• PD: Unequivocal progression of subretinal seeding based on increased number or density of seeds, and/or increased subretinal fluid.

These criteria are largely influenced by a 2007 clinical trial that examined the efficacy of carboplatin as a neoadjuvant agent for intraocular retinoblastoma and defined specific seeding response criteria.³⁵ Retinoblastoma involvement of subretinal compartments is characterized by seeding that accumulates inferiorly at the ora serrata and is often associated with subretinal exudate.^9,36 Similar to vitreous seeding, subretinal fluid and seeding can be identified with ultrasound, although fundus photography is ideal for evaluating the specific location and morphology of subretinal seeds.

3.3 |. Clinical application of RB-RECIST

The purpose of RB-RECIST is to provide a standardized, objective framework for retinoblastoma response assessment in clinical trials. It is equally important to consider RB-RECIST definitions in the context of ongoing clinical decision making and risk assessment for retinoblastoma patients. In previously published retinoblastoma studies, globe salvage was equated to complete therapeutic response of intraocular cancer as defined by the treating ocular oncologist.^40,55,56 These RB-RECIST guidelines are meant to standardize the language and assessment of tumor response across all studies of retinoblastoma therapy. In the future, we encourage treatment centers to prospectively evaluate and validate the RB-RECIST recommendations in the context of relevant imaging techniques and therapeutics for retinoblastoma.

3.4 |. Special notes on assessing retinoblastoma treatment response

• The specific response to treatment of the retinal tumor versus the vitreous and/or subretinal seeding may differ at any time point (e.g., the vitreous seeding may demonstrate a PR or CR, whereas the retinal tumor may concurrently show SD or PD). While the consolidative treatment of retinal and vitreous disease is different, PD in any ocular space decreases the overall prognosis for ocular survival. Thus, any PD is indicative of disease activity within the eye. Similarly, complete therapeutic response for the entire eye is achieved only when there is no residual active disease in any intraocular space. These guidelines are directed toward evaluation of the overall therapeutic response in the intraocular space.

• During evaluations for tumor response, the patient will receive a single categorical response to describe the overall disease status in the eye. Often this occurs at monthly intervals during treatment. Tumor response should be determined relative to the baseline measurements at diagnosis.⁴⁸ However, with treatment response there will be a nadir in tumor measurements. Retinoblastoma has a well-documented and significant risk of local intraocular relapse postchemotherapy.^38,57–59 Therefore, increases in tumor size should be compared to the most recent nadir in tumor measurements.⁴⁸ For example, an eye can demonstrate PR or even CR with treatment response, but then develop recurrent growth months later (which should be defined as PD).

• It should be noted that patients with heritable retinoblastoma are predisposed to the development of new, separate retinal tumors, and this is part of the rationale for continued surveillance of these children under anesthesia. While development of a new tumor does NOT indicate failed response leading to local progression, it does indicate active disease in the eye which may require a change (or reinitiation) of therapy.

• As a tumor regresses, it may become clear that multiple separate tumors, instead of one single tumor, are present in the eye. In this case, it may be clinically impractical to measure every single tumor focus reliably. Thus, the largest tumor focus can be used as a surrogate measurement of total tumor burden.

• In clinical trials assessing tumor response with ultrasonography, serial evaluations should be performed by the same examiner whenever possible in order to minimize interoperator variability in tumor thickness measurements.²⁶

• Often a cooperative in-clinic exam is not possible, thus imaging evaluations may be performed with the pediatric patient under sedation or general anesthesia in order to attain appropriate positioning and accurate measurements.⁹ The primary purpose of each examination under anesthesia (EUA) is to assess the intraocular response to treatment. Consolidative treatment may be given during the same examination for eyes that have not responded completely to chemotherapy. Findings of disease progression often require the initiation of new treatment modalities aimed at the size and anatomic location of the active disease.

• In general, MRI is used at baseline to evaluate for optic nerve and central nervous system involvement (e.g., pinealoblastoma) as well as at the conclusion of therapy.^13–16 In many patients, particularly those with heritable retinoblastoma, MRI is obtained at 6-month intervals for central nervous system screening. MRI is also useful for detecting tumor activity in the setting of opaque media or vitreous hemorrhage.⁹ Because MRI is a relatively expensive modality and is not performed at every appointment, its role in evaluating tumor response throughout treatment is limited. We recommend ultra-sonography for monitoring tumor response, as this modality can be performed at every EUA and thus facilitates shorter time intervals between measurements.

• In current clinical practice, our understanding of retinoblastoma activity during active therapy relies almost exclusively on imaging and clinical observations. With the evolution of blood and aqueous humor based liquid biopsy platforms, there may be a future role for the inclusion of molecular biomarkers in the assessment of therapeutic response.^60,61

• Although B-scan is typically the only ultrasonic modality used to measure apical height in retinoblastoma tumors (due to the frequency of intratumoral calcifications that interfere with A-scan reads²⁷), B-scan in combination with standardized A-scan is highly recommended for measuring melanoma tumors (which are typically noncalcified), especially in clinical trials.²⁶ Given the differences in morphology and imaging of retinoblastoma and melanoma tumors, a separate set of response criteria is necessary for the evaluation of uveal melanoma and other ocular tumors.

4 |. CONCLUSIONS

Although RECIST 1.0 and 1.1 recommend against the use of ultrasound in measuring solid lesions, ultrasonography is a safe and reliable imaging modality for assessing intraocular tumor parameters—particularly in pediatric retinoblastoma populations. With an increasing number of eye-salvaging treatments available for retinoblastoma, it is important that ocular oncologists consider tumor response in the retinal, subretinal, and vitreous spaces—and not only eye and/or patient survival—as a valid clinical endpoint. Herein, we are the first to recommend standardized definitions of intraocular tumor and seeding response criteria for retinoblastoma, and we encourage the use of these guidelines in ocular oncology clinical trials and studies in the future.

Abbreviations:

CR

complete response

CT

computed tomography

EUA

examination under anesthesia

MRI

magnetic resonance imaging

PD

progressive disease

PR

partial response

RECIST

Response Evaluation Criteria In Solid Tumors

SD

stable disease

SD-OCT

spectral domain optical coherence tomography

WHO

World Health Organization