Tumour size criteria for Group D and E eyes in the International Classification System for Retinoblastoma: effects on rates of globe salvage and high-risk histopathologic features

Abstract

Purpose:

To determine the significance of large tumour size as a criteria for classifying advanced intraocular retinoblastoma, analysing rates of globe survival and high-risk (HE) histopathologic features.

Methods:

Retrospective chart review of 212 eyes diagnosed with Group D (111 eyes) or Group E (101 eyes) retinoblastoma in at least one eye from January 1, 2006 to December 31, 2016 using the Los Angeles (LA) Classification System (no tumour size criteria for Group E). The 111 Group D tumours were then reclassified to Group E using 10, 12, 14, 16, 18 mm tumour size criteria, as determined by ultrasound or magnetic resonance imaging dimensions.

Results:

For eyes in the original LA classification, 66.7% of Group D and 10.5% of Group E eyes undergoing globe preservation therapy avoided enucleation or radiotherapy (p < 0.0001; median follow-up of 33.0 months). In the LA classification, 8.5% of Group D and 26.3% of Group E enucleated globes had HE histopathologic features (p = 0.0065). When Group D eyes with tumours meeting the size criteria were reclassified to Group E, 65.7–74.4% of Group D and 16.1–36.7% of Group E eyes avoided enucleation or radiotherapy. Applying the tumour size criteria, 0–10.9% of Group D and 20.7–23.8% of Group E eyes had HE histopathologic features.

Conclusion:

Our retrospective analysis suggests that a large tumour size criteria for Group E retinoblastoma have no clinical basis, given that the LA classification system provided the greatest separation in globe salvage rates between Group D and E eyes. The LA classification system was also able to show a statistically significant difference in the rates of HE histopathologic features between Group D and E eyes. To avoid discrepancies in the literature, we recommend that centres use one uniform system for classifying advanced intraocular retinoblastoma.

Introduction

Introduced by Murphree in 2005, the International Classification System for Retinoblastoma has become the standard method for categorizing patients with intraocular retinoblastoma worldwide (Shields et al. 2006; Kim et al. 2007; Abramson et al. 2015; Scelfo et al. 2017). However, there are currently discrepancies in some aspects of the system used by various centres, particularly when classifying advanced cases (i.e. Groups D and E). The original classification for intraocular retinoblastoma [i.e. Los Angeles (LA) system] does not have a size criteria for Group E tumours, but does include eyes with neovascular glaucoma, massive intraocular haemorrhage, bloodstained cornea, massive tumour necrosis associated with aseptic orbital cellulitis, phthisis or prephthisis, tumour anterior to the anterior vitreous face, tumour touching the lens, anterior segment invasion, or diffuse infiltrating retinoblastoma (Murphree 2005). The Philadelphia classification system published by Shields et al. (2006) includes many of these criteria for Group E retinoblastoma and adds a size criteria for tumours involving >50% of the volume of the globe. A previous analysis performed at Memorial Sloan-Kettering Cancer Center (MSKCC) revealed that 5% of tumours in the LA classification system may be reclassified when the Philadelphia criteria are applied (Novetsky et al. 2009). Discrepancies in the criteria between Group D and Group E eyes continue to be problematic in the literature when comparing success rates between various centres and also create confusion when clinicians diagnose eyes with retinoblastoma that meet criteria in one system but not the other.

No previous publication of retinoblastoma patients has specifically documented the size of Group D or E tumours treated in their series. We hypothesize that there is a subgroup of Group D eyes that contain very large tumours with diffuse seeding, but do not yet fulfil Group E criteria such as neovascular glaucoma, buphthalmia, vitreous haemorrhage, anterior segment invasion, and phthisis. Interestingly, the LA classification system combines this subgroup of large tumours with all eyes that present with diffuse seeding as Group D, while the Philadelphia classification designates these eyes as Group E even though they have not yet progressed to cause destructive ocular changes such as neovascular glaucoma, buphthalmia, vitreous haemorrhage, anterior segment invasion, and phthisis. Based on the evidence published to date, it is unclear how this subgroup of eyes with very large tumours should be classified and managed to optimize outcomes. The goal of this retrospective review was to analyse the impact of adding a large tumour size criteria for classifying advanced intraocular retinoblastoma, both on rates of globe salvage and high-risk (HE) histopathologic features.

Methods

Retrospective chart review of patients diagnosed with Group D or E retinoblastoma in at least one eye from January 1, 2006 to December 31, 2016 at our centre; there were 222 Group D and E eyes treated during this time period and 212 eyes had tumour dimensions available for analysis. Tumours were classified by largest basal dimension as determined either on magnetic resonance imaging (MRI) or on B- scan ultrasound. A neuroradiologist specializing in paediatric head and neck imaging reviewed MRI scans (when available) and confirmed the accuracy of the tumour dimensions. If the actual MRI scan was not available for review, then the tumour dimensions were recorded from the original interpretation report in the patient’s chart. Tumour dimensions from B scan ultrasound images were recorded from the chart when MRI images (or its report) were not available for review. The largest determined tumour dimension was then recorded as the size for all group D and E eyes, for example, a tumour measuring 8 × 10 × 12 mm was classified as a 12 mm tumour.

Using the LA classification system (no tumour size criteria for Group E), there were 111 Group D and 101 Group E eyes. At least one of the Group E criteria in the LA classification system was confirmed to be present in the 101 Group E eyes and absent in the 111 Group D eyes (Murphree 2005). The Group D tumours were then reclassified as Group E when it met a criteria of >10, 12, 14, 16, or 18 mm in tumour size. During this time period at our centre, Group D and E eyes underwent globe salvage using either systemic chemoreduction or intra-arterial chemotherapy (IAC; either as primary or secondary therapy), combined with local consolidation using laser photocoagulation and/or cryotherapy (per previously published CHLA protocol) (Berry et al. 2013). Some unilateral Group D or E eyes were primarily enucleated when parents chose to avoid chemotherapy. A total of 135 eyes in this cohort underwent attempted globe salvage, and 77 eyes were primarily enucleated (total 212 globes). For attempted globe salvage, systemic chemotherapy with carboplatin, etoposide, vincristine (CEV) was used in the management of 132 eyes and IAC using melphalan was used in six eyes; three eyes received treatment with CEV as primary therapy and IAC as secondary therapy. Intravitreal chemotherapy injections using melphalan were used to treat vitreous seeding from 2012 to 2016 in 17 eyes. High-risk (HE) histopathologic features were defined as postlaminar optic nerve invasion, massive choroidal invasion (3 mm or greater in any dimension) or scleral invasion.

Comparison of globe salvage rates across all classification types for both Group D and E eyes was conducted using two-tailed Chi-square test at the 0.05 significance level. Comparing the rates of HE histopathologic features across all classification groups for Group D and E eyes was conducted using two-tailed Chi-square test at the 0.05 significance level and Fisher’s exact test when appropriate. All statistical analyses were performed using stata/se 14.2 (StataCorp LLC, College Station, TX, USA).

Results

In the LA classification system, 66.7% (52/78 eyes) of the Group D eyes that underwent globe salvage therapy and 10.5% (6/57 eyes) of the Group E eyes that underwent globe salvage therapy were successfully treated without enucleation or radiotherapy (median follow-up of 33.0 months). The difference in globe salvage rates between Group D and E eyes for the LA classification system was statistically significant (p < 0.0001). When globes in this cohort were reclassified using the tumour size criteria, 65.7–74.4% of Group D and 16.1–36.7% (40/109 eyes) of Group E eyes undergoing globe salvage therapy were successfully treated without enucleation or radiotherapy (median follow-up of 33.0 months; see Table 1). All of the classification schemes utilizing tumour size criteria were able to demonstrate a statistically significant difference in globe salvage rates between Group D and E eyes (p values 0.0002 or less). When comparing the Group D globe salvage outcomes between the LA classification system and the various tumour size categories, there was no statistically significant difference in globe salvage rates. However, when comparing the globe salvage rates for Group E eyes between the LA classification system and the 10 mm and 12 mm classification systems, the p values were 0.0003 and 0.0051, respectively. The Group D eyes which were reclassified according to the tumour size criteria (i.e. Group D/E) had globe salvage rates ranging between 52.4% and 80%. See Table 1 for a summary of these results on globe salvage outcomes.

Table 1.

Globe salvage outcomes of eyes with advanced retinoblastoma using the Los Angeles (LA) classification system and 10–18 mm tumour size criteria.

	LA classification	10 mm criteria	12 mm criteria	14 mm criteria	16 mm criteria	18 mm criteria
Total globes	212	212	212	212	212	212
Group D	111	32	48	71	92	97
Group E	101	180	164	141	120	115
Group D reclassified (D/E)*		79	63	40	19	14
Group D globes salvaged	52/78 (66.7%)	18/26 (69.2%)	29/39 (74.4%)	41/57 (71.9%)	46/70 (65.7%)	48/73 (65.8%)
Group D enucleated (1/2/T)†	33/26/59	6/8/14	9/10/19	14/16/30	22/24/46	24/25/49
Group E salvaged	6/57 (10.5%)	40/109 (36.7%)	29/96 (30.2%)	17/78 (21.8%)	12/65 (18.5%)	10/62 (16.1%)
Group E enucleated (1/2/T)†	44/51/95	71/69/140	68/67/135	63/61/124	55/53/108	53/52/105
Group D versus Group E globe salvage rates (p values)	66.5%/10.5% (<0.0001)	69.2%/36.7% (0.0002)	74.4%/30.2% (<0.0001)	71.9%/21.8% (<0.0001)	65.7%/18.5% (<0.0001)	65.8%/16.1% (<0.0001)
Group D globe salvage rates: LA versus size criteria (p values)		66.7%/69.2% (0.8093)	66.7%/74.4% (0.3954)	66.7%/71.9% (0.5141)	66.7%/65.7% (0.9027)	66.7%/65.8% (0.9056)
Group E globe salvage rates: LA versus size criteria (p values)		10.5%/36.7% (0.0003)	10.5%/30.2% (0.0051)	10.5%/21.8% (0.0854)	10.5%/18.5% (0.2176)	10.5%/16.1% (0.3708)
Group D/E1 globe salvage		34/52 (65.4%)	23/39 (59.0%)	11/21 (52.4%)	6/8 (75%)	4/5 (80%)
Group D/E* Enucleated (1/2/T)†		27/18/45	24/16/40	19/10/29	11/2/13	9/1/10

^*Group D/E = reclassified Group D eyes in the LA classification system meeting tumour size criteria.

^†1/2/T = primary enucleations/secondary enucleations/total enucleations.

Analysis of the enucleated globes showed that in the LA classification system, 8.5% (5/59) of Group D eyes had HE histopathologic features and 26.3% (25/75) of Group E eyes had HE histopathologic features. The LA classification system was able to demonstrate a statistically significant difference in HE histopathologic features between Group D and E eyes (p = 0.0065). For the tumour size categories, the rates of HE histopathologic features ranged between 0–10.9% for Group D eyes and 20.7–23.8% for Group E eyes. A comparison of the rates of HE histopathologic features between Group D and Group E eyes did not show a statistically significant difference for the 10–18 mm tumour size categories (Table 2). The Group D eyes which were reclassified according to the tumour size criteria (i.e. Group D/E) had rates of HE histopathologic features ranging between 0% and 11.1%.

Table 2.

High-risk (HE) histopathologic features in advanced retinoblastoma using the Los Angeles (LA) classification system and 10–18 mm tumour size criteria.

Group D high-risk features	5/59 (8.5%)	0/14 (0%)	2/19 (10.5%)	2/30 (6.7%)	5/46 (10.9%)	5/49 (10.2%)
Group D HR features (1/2/T)	3/2/5	0/0/0	1/1/2	1/1/2	3/2/5	3/2/5
Group E high-risk features	25/95 (26.3%)	30/140 (21.4%)	28/135 (20.7%)	28/124 (22.6%)	25/108 (23.1%)	25/105 (23.8%)
Group E HR features (1/2/T)†	3/22/25	5/25/30	4/24/28	4/24/28	3/22/25	3/22/25
Group D/E high-risk features*		5/45 (11.1%)	3/40 (7.5%)	3/29 (10.3%)	0/13 (0%)	0/10 (0%)
HR features Group D versus Group E (p values)	8.5%/26.3% (0.0065)	0%/21/% (0.073)	10.5%/20.7% (0.3705)	6.7%/22.6% (0.0695)	10.9%/23.1% (0.1182)	10.2%/23.8% (0.0516)

^*Group D/E = reclassified Group D eyes in the LA classification system meeting tumour size criteria.

^†1/2/T = primary enucleations/secondary enucleations/total enucleations.

Discussion

The original classification system introduced by Murphree in 2005 was designed to predict treatment success using a combination of systemic chemotherapy and focal consolidation. Despite the growing popularity of other primary treatment modalities such as IAC over the past decade (Yamane et al. 2004; Abramson et al. 2008), the same framework of the original classification system continues to be used to predict treatment success for intraocular retinoblastoma by clinicians worldwide. As pointed out by Abramson, there are currently three slightly different versions of the classification system used by centres to treat intraocular retinoblastoma (Abramson et al. 2015; Scelfo et al. 2017). The LA system is the original version written by Murphree in 2005, while the version with a tumour size criteria for Group E eyes was published by Shields et al. (2006). The third version of the system is used by the Children’s Oncology Group and is published on their website (Children’s Oncology Group). More recently, the 8th edition of the American Joint Committee for Cancer (AJCC) classification system for retinoblastoma was published in 2017 and shares many similarities to the LA classification system since neither includes a size criteria for categorizing the advanced categories (Mallipatna et al. 2017). For example, category cT3 in the AJCC classification system is defined by the presence of phthisis, tumour invasion of the anterior chamber, neovascular glaucoma, intraocular haemorrhage and aseptic orbital cellulitis but does not include a tumour size criteria (Mallipatna et al. 2017). Data from a survey of 1728 eyes published in the AJCC Retinoblastoma chapter showed improved separation of globe salvage rates for the 8th edition of the AJCC classification system compared with version ICRB (i.e. Philadelphia system) (Mallipatna et al. 2017).

When considering the differences in the three current versions of the International Classification System, the most clinically relevant discrepancy is the size criteria for advanced tumours, which essentially reclassifies the largest Group D tumours in the LA system as being Group E tumours in the Philadelphia system. Including a size criteria for Group E tumours in our patient cohort, a significant number of the Group D eyes in the LA classification were then reclassified as being Group E; 71% (79/111) of Group D eyes with the 10 mm criteria and 12.6% (14/111) of Group D eyes with the 18 mm criteria. Therefore, many of the advanced Group D eyes in the LA classification group in our original series had tumours that met the established size criteria but did not yet fulfil the other group E criteria (e.g. neovascular glaucoma, ocular phthisis, vitreous haemorrhage or anterior segment involvement). The percentage of Group D eyes that were reclassified eyes in our analysis was greater than the 5% of patients identified in the original study performed at MSKCC in 2009, although the latter used the criteria of >50% of the globe volume rather than a specific tumour size criteria (Novetsky et al. 2009). Given the lack of uniformity in the current classification systems, the discrepancy in the criteria between Group D and Group E eyes has the potential to create a bias in the literature when comparing treatment outcomes between different centres.

A tumour size criteria for advanced retinoblastoma were introduced by the Shields group, and they proposed classifying tumours occupying more than 50% of the volume of the globe as being Group E in the International Classification System (Shields et al. 2006). It is unclear how the authors specifically applied the criteria in any individual case, but it is assumed to be an informal clinical assessment based on a review of an ultrasound image or MRI scan. Since a defined criterion was needed to perform a retrospective analysis, we decided to use the largest tumour dimension (height or basal length) on the pretreatment imaging study. The adult sized globe is 24 mm in axial length, while the length of the posterior segment in adults is approximately 20 mm. The average axial length for a normal child’s eye at birth is 16.8 mm, 20 mm at 12 months of age and 21 mm at 4 years of age (Hussain et al. 2014). We acknowledge that the Group E eyes in the Philadelphia classification that are estimated to be >50% of the globe volume may be larger than 18 mm in diameter or height, but it is inherently difficult to obtain a 20 mm tumour dimension in a child’s eye. When Group D eyes in the LA classification system were reclassified using the tumour size criteria, there was a trend towards increasing the globe salvage rates for Group D eyes in the 10, 12, 14 mm categories while rates were essentially unchanged for the 16 and 18 mm categories (Table 1). For Group E eyes, applying the tumour size criteria increased the globe salvage rates in all established size categories (10–18 mm), with the difference in globe salvage rates between the LA classification and the 10 and 12 mm size criteria reaching statistical significance. All classification systems analysed were able to show a statistically significant difference in globe salvages between Group D and E eyes, but the greatest separation was achieved with the LA classification system (no tumour size criteria).

High-risk (HE) pathologic features such as postlaminar optic nerve invasion, massive choroidal invasion and scleral invasion have been associated with an increased risk of metastatic disease after enucleation, and adjuvant therapy appears to decrease this risk to very low levels (Kaliki et al. 2011, 2013; Sullivan et al. 2014; Brennan et al. 2015). The risk of metastatic disease for patients with HE histopathologic features has been proposed as the justification for performing primary enucleation for Group E eyes (Sullivan et al. 2014; Brennan et al. 2015). In our series, the rates of HE histopathologic features in our series ranged between 0–10.9% for Group D eyes and 20.7–26.3% for Group E eyes. The LA classification system was the only one that demonstrated a statistically significant difference in the rates of HE histopathologic features between Group D and Group E eyes. Interestingly, the reclassified Group D eyes in our study (i.e. Group D/E) had a risk of having HE pathologic features between 0% and 11.1%, which was similar to the results of Group D eyes that were not reclassified (0–10.9%). Therefore, in our analysis, the larger tumours that met the size criteria had rates of HE histopathologic features similar to smaller Group D tumours. High-risk (HE) pathologic features in secondarily enucleated globes may be obscured due to treatment-related effects (Zhao et al. 2011), and this may have affected the data. For all classification schemes, there were significant numbers of secondarily enucleated eyes for both Group D and E categories. In summary, the LA classification system was able to demonstrate a significant difference in the rates of HE histopathologic features between Group D and Group E eyes, and large tumour size did not appear to identify eyes at higher risk for extraocular relapse when the other Group E criteria were absent.

The Group D eyes in the LA classification system which met the size criteria and were reclassified in our analysis represent a subgroup of eyes which have larger tumours with diffuse seeding but do not yet manifest other features such as neovascular glaucoma, intraocular haemorrhage, ocular phthisis or anterior segment invasion. Surprisingly, this subgroup of eyes with larger tumours demonstrated a >50% chance of globe salvage in all size categories and had rates of HE histopathologic features ranging between 0% and 11.1%. Classifying this group of larger tumours as Group E rather than Group D may have affected clinical decision-making (i.e. primary enucleation versus salvage therapy), since Group E eyes are considered to have a low chance of globe salvage and a higher risk of invasive histopathologic features. Our analysis suggests that this subgroup of large tumours may have clinical features more in common with smaller Group D tumours than with Group E tumours in the LA classification system. In addition, combining this subgroup with the most advanced eyes that demonstrate destructive ocular features (e.g. neovascular glaucoma, intraocular haemorrhage) essentially decreased the separation in rates for globe salvage and HE histopathologic features between Group D and E eyes. We acknowledge that the majority of our patients received systemic chemotherapy during this time period for globe salvage, and the results may not applicable to a group of patients treated predominantly with IAC. Further analysis with a larger group of patients from multiple centres will likely be required to prove the statistical validity of these trends. Until we have this validation, we would recommend avoiding a tumour size criteria for classifying Group E retinoblastoma by using either the original LA classification system or the 8th edition of the AJCC classification system (Mallipatna et al. 2017).

A classification system for intraocular retinoblastoma should accurately predict which advanced eyes can be safely salvaged versus those eyes which are at significant risk for treatment failure, secondary enucleation and/or extraocular relapse. In our analysis, the LA classification system provided the greatest separation in globe salvage rates between Group D and E eyes. Similarly, the LA classification system was able to show a statistically significant difference in the rates of HE histopathologic features between Group D and E eyes. The largest Group D tumours in the LA classification system that were reclassified as Group E based on the established size criteria were found to have a >50% chance of globe salvage in our series and had relatively low rates of HE histopathologic features. Our retrospective analysis suggests that adding a tumour size criteria for classifying Group E eyes has no clinical basis and that eyes with diffuse tumour seeding have similar clinical outcomes when other features such as neovascular glaucoma, intraocular haemorrhage and anterior segment invasion are absent. We recommend centres use one uniform classification system for advanced intraocular retinoblastoma to avoid discrepancies when comparing results between clinical series.