Advances in medicine, including new treatments, frequently inspire or require new methods for categorizing disease and/or assessing the efficacy of treatment. In the case of retinoblastoma, not long after the adoption of systemic chemotherapy in the 1990s — which was adopted in the hope of minimizing radiation-related second cancers — clinicians began to recognize the need for a classification scheme that would enable them to more accurately describe responses to chemotherapy. Less than a decade later, the International Intraocular Retinoblastoma Classification (IIRC) essentially replaced the Reese-Ellsworth system that had been used since the 1960s to describe responses to radiation.

Today, clinicians continue to grade diseased eyes A through E according to the IIRC, with group A eyes having the best prognosis, and group E eyes having the poorest prognosis, even though intra-arterial chemotherapy (IAC) has now replaced systemic chemotherapy as the standard of care in developed countries. However, the existence of at least three different versions of the IIRC has posed significant challenges. Recently published survey results¹ have helped to increase awareness of the problems inherent in using multiple classification schemes — particularly when treatment outcomes are commonly reported without reference to and, in some cases, without knowledge of, the particular classification scheme used.

A BRIEF HISTORY

For most of the twentieth century — from 1903 to the mid-1990s — external beam radiation therapy (EBRT) was the only treatment that could be used to save the eyes of children with advanced intraocular retinoblastoma. Soon after the first successful case study was published in 1903, clinicians rapidly adopted EBRT, with a significant degree of overall success. Yet variations in treatment response were difficult to explain.

“No one could figure out why some responses weren’t as successful as others,” says David H. Abramson, MD, chief of ophthalmic oncology service at Memorial Sloan Kettering Cancer Center in New York. “That is, until Reese and Ellsworth developed their classification in the 1960s.”

Dr. Abramson notes that the scheme developed by Reese Ellsworth was not a staging system, nor a system used to predict whether a patient would survive or have vision, but simply a classification scheme that predicted the likelihood of globe salvage when an eye was treated with lateral photons.

“As soon as the Reese Ellsworth classification was in regular use, it became obvious that the two determinants of success were the volume of tumor and the presence of vitreous seeds,” says Dr. Abramson. “It turned out that eyes with significant seeding were not responding well to radiation.”

Later, it was discovered that the same RB1 mutation responsible for retinoblastoma also increased the risk of cancers of the bone and soft tissues later in life, particularly when children were exposed to radiation in the first year of life, leading to a search for an alternative to radiation in the late 1980s. Then, after the widespread adoption of systemic chemotherapy as the standard of care for retinoblastoma in the 1990s, it became apparent that the Reese-Ellsworth system was an inadequate tool for predicting responses to chemotherapy.

Modifications to the original IIRC developed by A. Linn Murphree, MD, of the Children’s Hospital of Los Angeles (CHLA)² came first, followed later by the development of both the Philadelphia³ and Children’s Oncology Group (COG)⁴ versions of the IIRC, which also use A to E classification schemes.

“All three of these classification schemes are good,” Dr. Abramson says, “however, despite using the same letters and similar wording, they are not describing the same phenomena. This is particularly problematic with regard to descriptions of Group D eyes, which are the most challenging and the most commonly seen eyes with retinoblastoma in the US and around the world.”

Generally speaking, group D eyes are those with nondiscrete growth or nonlocalized seeding (Figure 1). However, while the CHLA version defines D eyes as having a tumor with “exophytic or endophytic qualities, diffuse/extensive vitreous or subretinal seeding, and/or retinal detachment in >1 quadrant,”² the Philadelphia system describes the same tumors as having “subretinal and/or vitreous seeds >3 mm from the tumor margin, occupying ≤50% of the globe,”³ and the COG criteria includes the presence of vitreous and/or subretinal seeding (in addition to subretinal fluid) and increases the radius from the tumor margin to >6 mm.⁴ The discrepancies in these criteria, among others, have undermined the value of the IIRC in assessing, comparing, and communicating treatment outcomes. Moreover, a review of the literature by Dr. Abramson and colleagues revealed that only 3 out of 20 papers communicating outcomes of eyes classified according to the IIRC specified which IIRC version was used.¹

SURVEY FINDINGS

To determine the IIRC schemes used by retinoblastoma centers around the world and examine correlations between IIRC version used and a range of critical treatment decisions, Dr. Abramson and colleagues designed a 39-question survey that was sent by email to 115 ophthalmologists at 39 retinoblastoma centers worldwide.

The survey, which was completed by 29% of physicians invited to participate, provided information about a total of 1807 eyes classified as Group D. Among the centers providing this information, 27% used the CHLA version of the IIRC, 33% used the COG version, 23% used the Philadelphia version, and 17% were unsure of which system they used.

Outcomes reported by the centers surveyed varied considerably, depending on the IIRC version used. Perhaps most notably, centers using the Philadelphia version classified the greatest number of eyes as Group D and enucleated the fewest eyes upfront (prior to treatment).

“As we noted in our paper, the Group D eye in the Philadelphia version of the IIRC has the lowest cutoff with regard to distance of subretinal or vitreal seeds from the tumor, and in this version the criteria for a Group D eye does not include the presence of subretinal fluid or retinal detachment,” says Dr. Abramson. “So a D eye in the Philadelphia scheme is very close to a C eye in one of the alternate schemes, and most centers tend to have a good deal of success treating C eyes. The Philadelphia scheme doesn’t distort anything — it is simply a different way of presenting criteria. But centers using this scheme aren’t getting better results, they’re simply using a different classification.”

The survey results also showed that enucleation rates varied greatly among centers worldwide (between 0% and 100%), with approximately 27% of group D eyes enucleated upfront. Systemic chemotherapy was the most common primary treatment choice for 53% of group D eyes, although this finding was likely due in part to the treatments available during the study period, which ended in 2012. Overall, the survey confirmed that the use of different versions of the IIRC makes it difficult to compare success across centers.

“By conducting this survey, we accomplished pretty much what we had hoped to accomplish. Our goal was not to suggest a new scheme, or to put forth an argument for giving up two schemes and using only one,” notes Dr. Abramson. We wanted to point out the differences in these schemes and the importance of being aware of them when reading the literature and when reporting outcomes. Journals also must pay attention to these differences in classification schemes. Authors must indicate which classification scheme they are using, and of course, that implies that they must know which scheme they are using.”

Dr. Abramson notes that comparison of success across centers worldwide is also confounded by inconsistencies with regard to how enucleation rates are reported.

“It is essential that centers clearly state how many eyes are removed up front. Many centers around the world are not providing this information, and instead are reporting outcomes only on the eyes they treat, regardless of how many they removed prior to treatment,” explains Dr. Abramson.

MOVING FORWARD

Ultimately, Dr. Abramson believes that awareness of the different IIRC systems used and how they might play a role in the interpretation of outcomes is more important than efforts toward developing yet another system — even one that might more accurately report outcomes with IAC.

“It may be true that IIRC classifications are not especially good at predicting success with IAC, and for that reason, some have suggested that we need a new classification scheme for IAC,” says Dr. Abramson, “but I’m not sure that’s the answer, because if we had a new system for IAC, we wouldn’t be able to compare IAC with systemic chemotherapy.”

“If we design a new classification scheme to fit every new treatment that emerges,” he adds, “we’ll find that it becomes more difficult to compare outcomes across treatments. We already know, for example, that it’s impossible to compare systemic chemotherapy outcomes using the IIRC with radiation treatment outcomes using the Reese Ellsworth system.” NRP

REFERENCES

1. Scelfo C, Francis JH, Khetan V, et al. An international survey of classification and treatment choices for group D retinoblastoma. Int J Ophthalmol. 2017;10(6):961-967.
2. Murphree AL. Intraocular retinoblastoma: the case for a new group classification. Ophthalmol Clin North Am. 2005;18(1):41-53.
3. Shields CL, Mashayekhi A, Demirci H, Meadows AT, Shields JA. Practical approach to management of retinoblastoma. Arch Ophthalmol. 2004;122(5):729-735.
4. Children’s Oncology Group. Just Diagnosed. Staging. Available at https://www.childrensoncologygroup.org/index.php/newlydiagnosedwithretinoblastoma ; Accessed Aug. 16, 2017.