The National Cancer Institute (NCI) SEER program includes 17 population-based cancer registries covering approximately 46% of the U.S. population from 2000 and 2021 and was used for this retrospective cohort study. This study was exempted from Institutional Review Board approval by the Albert Einstein College of Medicine, because the SEER program contains de-identified, publicly available data governed by a formal data use agreement. The study was conducted in accordance with the principles outlined in the Declaration of Helsinki and was reported following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.

Retinoblastoma patients diagnosed between January 1, 2000, and December 31, 2021, were identified by International Classification of Diseases for Oncology, Third Edition (ICD-O-3) site codes (C69.2: Retina and C69.9: Eye, not otherwise specified) and morphology codes 9510/3 to 9513/3. Initial diagnoses are typically made by ophthalmologists with additional diagnostic confirmation, and these cases are then abstracted into SEER by trained tumor registrars. Patients were excluded if diagnosed at age ≥ 15 years or if data on racial/ethnic or geographic classification were missing ( Figure 1 ).

Inclusion of patients with retinoblastoma from the Surveillance, Epidemiology, and End Results (SEER) program (2000-2021) Flow diagram of patient selection for the study. Pediatric patients diagnosed with retinoblastoma were identified from the SEER program.

The primary exposures of interest were race/ethnicity, county-level median household income, and geographic classification. Race and ethnicity were categorized based on SEER’s race and origin recode, placing patients into “Non-Hispanic White,” “Non-Hispanic Asian or Pacific Islander,” “Hispanic (All Races),” “Non-Hispanic Black,” or “Non-Hispanic American Indian/Alaska Native.” For some analyses, all non-White subgroups were aggregated into a single “Non-White” category to enable binary comparisons. Median household income was recorded by SEER as an area-level measure derived from the American Community Survey, reflecting the median household income of the patient’s county of residence, and was inflation-adjusted to 2022. Income was dichotomized as <$60,000 and ≥$60,000 to differentiate lower- and higher-income populations, with $60,000 being a threshold close to U.S. median household income for the majority of the study period. Geographic location was determined using the Rural-Urban Continuum Code, which classifies counties by population size and metropolitan adjacency. Patients who resided in any metropolitan county were grouped as “Metropolitan” and patients from all other areas were grouped as “Non-Metropolitan.”

Additional covariates included patient sex and age at diagnosis. Age was divided into four SEER-defined categories: “0 Year,” “1 to 4 Years,” “5 to 9 Years,” and “10 to 14 Years.” Sex was divided into “Male” or “Female.” While SEER also reports a “15 to 19 Years” category, this group was not included because the database does not allow distinction between patients aged 15 to 18 years and those aged 19 years. This exclusion may have reduced the total number of eligible cases, but it improved the specificity of the analysis by ensuring that the study population was restricted to pediatric patients.

Three primary outcomes were measured: (1) treatment modality, (2) disease stage at diagnosis and (3) time interval between diagnosis and initiation of treatment. Surgeries were grouped based on SEER site-specific surgery codes and collapsed into meaningful categories: (1) “Local Tumor Excision or Destruction” (laser therapy, cryotherapy, and other local ablative procedures; Codes 10-27); (2) “Enucleation” (complete surgical removal of the eye; Codes 40-41); (3) “Other Surgery” (partial excision, debulking, or radical surgery with involvement of surrounding structures; Codes 30, 50, 60, and 90); (4) “No Surgery” (patients who did not receive any surgical intervention; Code 00); and (5) “Unknown” (Code 99). Surgical procedures were categorized using the SEER program variable “RX Summ–Surg Prim Site (1998+),” which records a single surgery code representing the initial surgical procedure performed on the eye during the first course of treatment following diagnosis. While patients may receive additional surgical interventions after the initial procedure, this information was not available in the SEER program; therefore, only the initial surgery was analyzed. Because patients may receive systemic or local therapies such as chemotherapy or radiotherapy in addition to surgery, these treatments were analyzed separately. Chemotherapy and radiotherapy data were based on SEER treatment codes and were modeled as binary variables (“Yes” vs “No/Unknown”). The mode of chemotherapy administration was not specified by SEER, and intra-arterial chemotherapy could therefore not be defined separately. Stage at diagnosis was evaluated among patients with available data. Although the SEER program did not have staging data before 2004, Combined Summary Stage (CSS), which refers to SEER’s standardized system that classifies disease extent as “Localized,” “Regional,” or “Distant,” based on tumor confinement, spread to adjacent structures, or disemmination, was available from 2004 to 2021. In SEER, localized disease is defined as tumors confined to the eye or adjacent ocular structures, including the eyelid, the orbicularis muscle, tarsal plate, or tarsus. Regional disease is defined as direct invasion into contiguous orbital or ocular tissues, and distant disease is defined as metastasis to sites such as the brain and central nervous system, nasal cavity, or distant lymph nodes. Finally, time to treatment was defined as the time (in days) from the recorded date of diagnosis to initiation of first cancer treatment.

Descriptive statistics were calculated to characterize the cohort by demographic and clinical characteristics. Chi-square tests and odds ratios (ORs) were calculated to identify associations between exposure variables (race/ethnicity, income, geography) and the nature of surgical treatment received. Multinomial logistic regression was performed to identify independent predictors of treatment modality, adjusting for covariates, such as age at diagnosis and sex. Due to the limited count of distant metastases, disease stage was grouped into 2 categories: “Localized” and “Regional/Distant” to ensure sufficient sample size for binary logistic regression analysis. Binary logistic regression was also employed in analyzing factors associated with the receipt of chemotherapy and radiotherapy. The treatment time was compared across race/ethnicity, income groups, and geographic clusters using analysis of variance (ANOVA). Multiple linear regression was employed to examine predictors of time to treatment. To assess whether geographic context modified the associations between race/ethnicity or household income and study outcomes, interaction terms between geography and race/ethnicity and between geography and income were incorporated into regression models. Additionally, linear regression was employed in identifying trends in treatment option utilization among thestudy cohort over time. Statistical analysis was performed using IBM SPSS Statistics 29.0.2.0 (IBM Corp) and GraphPad Prism 10.4.1 for macOS (GraphPad Software, San Diego, California). Two-sided P values of <.05 were considered statistically significant.

The study population included 1375 children (1817 eyes) who were diagnosed with retinoblastoma between the years 2000 and 2021 ( Figure 1 ). The cohort was evenly distributed by sex, with 691 males (50.25%) and 684 females (49.75%). Most cases were diagnosed early in childhood, with 709 patients (51.56%) aged 1 to 4 years at the time of diagnosis and 614 patients (44.65%) diagnosed before 1 year of age. 584 children (42.47%) were Non-Hispanic White, and 791 children (57.53%) were non-White race or ethnicity. 1226 patients (89.15%) lived in metropolitan counties, while 149 patients (10.84%) resided in non-metropolitan counties. By county-level median household income, 238 children (17.31%) were from areas earning less than $60 000 per year, whereas 1137 (82.69%) were from communities at or above this level. 882 patients had localized disease (85.05%), while 137 patients (13.21%) and 18 patients (1.31%) had regional involvement and distant spread, respectively. Bilateral retinoblastoma was present in 442 patients (32.15%) while 933 patients (67.85%) had unilateral retinoblastoma ( Supplemental Table 1 ).

Enucleation was the most common interventional procedure, performed in 740 patients (53.82%). Local tumor excision or destruction was performed in 296 patients (21.53%). No surgical procedure was performed in an additional 285 patients (20.73%). The procedure was unknown for 51 patients (3.71%), and 3 patients (0.22%) were coded as “Other Surgery.” Chemotherapy was administered to 854 patients (62.10%), and 59 (4.29%) received radiotherapy ( Supplemental Table 2 ). From 2000 to 2021, enucleation significantly declined in usage (Slope: −0.685, P =.006). In contrast, globe-sparing treatments, such as localized tumor excision or destruction (Slope: 0.923, P <.001) and chemotherapy (Slope: 1.291, P <.001), increased over this period ( Figure 2 ). Radiotherapy decreased in usage from 2000 to 2021 (Slope: −0.400, P <.001).

Temporal trends in retinoblastoma management in the SEER program by race/ethnicity, income, and geography (2000-2021)

Each panel represents the annual proportion of SEER-registered retinoblastoma patients receiving the respective therapy.

Variation in treatment modality was observed by race/ethnicity ( P =.014) and geographic residence ( P =.006), but not by median household income ( P =.566). Non-Hispanic White patients were less likely to undergo enucleation compared with non-White patients (OR: 0.679, 95% CI: 0.548-0.842; P <.001). Similarly, children living in metropolitan counties had lower odds of enucleation relative to those in non-metropolitan counties (OR: 0.652, 95% CI: 0.459-0.926; P =.016) ( Table 1 ). These associations persisted after adjusting for covariates, as both Non-Hispanic White race (aOR: 0.757, P =.010) and metropolitan residence (aOR: 0.692, P =.002) were independently associated with lower odds of enucleation ( Supplemental Table 3 ). Geographic classification significantly modified the association between race/ethnicity and enucleation (geography × race, P =.018). Non-White children residing in non-metropolitan counties had higher odds of enucleation compared with non-White children in metropolitan counties (aOR: 1.785, 95% CI: 1.123-2.829; P =.015), whereas racial differences in enucleation were attenuated in metropolitan regions (aOR: 1.143, 95% CI: 0.896-1.472; P =.290) ( Supplemental Table 4 ).

Table 1

Odd Ratios (ORs) for Receiving Treatment Modalities by Race/Ethnicity, Geographic Classification, and Median Household Income.

Retinoblastoma Treatment	Race/Ethnicity		P value	Geographic Classification		P value	Income		P value
	Non-Hispanic White	Nonwhite		Metropolitan	Non-Metropolitan		<$60 000	≥$60 000
Enucleation	0.679 (0.548-0.842)	1.000 [Reference]	<.001	0.652 (0.459-0.926)	1.000 [Reference]	.016	1.040 (0.786-1.377)	1.000 [Reference]	.784
Localized Tumor Excision/Destruction	1.261 (0.974-1.633)	1.000 [Reference]	.078	1.209 (0.784-1.864)	1.000 [Reference]	.390	0.878 (0.619-1.244)	1.000 [Reference]	.463
No Surgery	1.430 (1.101-1.858)	1.000 [Reference]	.007	2.172 (1.288-3.664)	1.000 [Reference]	.003	0.871 (0.611-1.242)	1.000 [Reference]	.446
Radiotherapy	1.163 (0.682-1.985)	1.000 [Reference]	.579	1.113 (0.496-2.497)	1.000 [Reference]	.795	0.910 (0.465-1.779)	1.000 [Reference]	.782
Chemotherapy	0.954 (0.765-1.189)	1.000 [Reference]	.676	1.523 (1.082-2.144)	1.000 [Reference]	.015	1.004 (0.753-1.339)	1.000 [Reference]	.979

Only gold members can continue reading. Log In or Register to continue

Share this:

• X
• Facebook

Related

Related posts:

1. Long-Term Follow-Up of PLACE and SPECTRA Trials: Outcomes After Successful and Unsuccessful Half-Dose Photodynamic Therapy for Chronic Central Serous Chorioretinopathy
2. Swept-Source Optical Coherence Tomography Angiography Vascular Metrics as Biomarkers for Renal Function in Patients with Diabetes Mellitus
3. Hyperreflective Foci Contiguous With the Retinal Pigment Epithelium Associated With Visual Function in Aging, Early, and Intermediate Age-Related Macular Degeneration: MACUSTAR Study Report
4. Noninfectious Posterior Uveitides — Atypicals, Variants, and Masquerades: The Jungle of Differential Diagnosis

Stay updated, free articles. Join our Telegram channel

Join