Purpose
To characterize treatment patterns, overall survival (OS), and risk of second malignant tumors in patients with retinoblastoma (RB) using the SEER (Surveillance Epidemiology and End Results) dataset.
Design
Cohort study.
Methods
The SEER dataset was used to identify cases of RB using ICD-03 histology codes. Special permission was granted by the SEER administration to release chemotherapy information for this study (information that is not available in the publically available SEER dataset). Treatment of RB for patients with locoregional disease was characterized as surgical therapy, radiation therapy, chemotherapy, or any form thereof across 4 time periods from 1975 to 2010. Observed-to-expected (O/E) ratios for second malignant neoplasms were calculated with the use of age-specific incidence rates.
Results
There were 1452 cases of RB identified from 1975 to 2010, with 48% of patients being male and 30% presenting with bilateral disease. Treatment patterns over time of 1220 patients (84%) with localized disease showed an increase in chemotherapy (± any treatment) from 16.5% to 50.2% and a decrease in surgery (± any treatment) from 96.2% to 88.5% and decrease in radiation from 15.2% to 4.9% from the 1975–1979 time period to the 2000–2010 time period. The 10-year OS was 93.7%, 93.7%, 97.5%, and 97% for increasing time periods (1975–1979, 1980–1989, 1990–1999, 2000–2010, respectively; P = .029). Risk of second malignant neoplasms was highest among patients treated with radiotherapy with O/E ratio of 43 compared to 30 and 5 for chemotherapy and surgery alone, respectively.
Conclusion
Treatment trends for RB show an increase in chemotherapy utilization with a decrease in radiation therapy from 1975 to 2010. Second malignant neoplasms occurred mainly in patients treated with radiation therapy. Our series demonstrates improvement in survival in contemporary time periods, which parallels a shift in therapy toward chemotherapy with a decline in radiation therapy.
Retinoblastoma is an uncommon pediatric malignancy with an age-adjusted incidence rate of 11.8 cases per million children. Over the years treatment of retinoblastoma has shifted away from enucleation and external radiation toward chemotherapy and local therapy that spares the eye. Retinoblastoma-related mortality can be due to metastases, trilateral retinoblastoma, and second malignant neoplasms. Owing to improvement in diagnosis and management of primary retinoblastoma in the United States, death in patients is more commonly due to trilateral retinoblastoma and from second malignant neoplasms rather than metastases.
Onset of metastasis typically occurs within 2 years of diagnosis and portends a dismal prognosis, with death occurring within 6 months of diagnosis of metastatic retinoblastoma. Trilateral retinoblastoma refers to an ectopic intracranial tumor in patients with germline retinoblastoma presenting within the first 5 years of life. Development of second malignant neoplasms occurs secondary to effects of germline retinoblastoma mutation that is augmented by external radiation.
Although there has been a trend toward decreased utilization of radiation and increased use of chemotherapy for treatment of retinoblastoma, to our knowledge, survival advantage of this shift in preferred therapy has not been assessed or reported. We hypothesize that there is an improvement in the survival reflecting increased use of chemotherapy. We therefore performed an analysis of the SEER (Surveillance Epidemiology and End Results) limited dataset to characterize treatment patterns for retinoblastoma over time, risk of the development of second malignant tumors and trilateral retinoblastoma, and overall survival.
Methods
We reviewed the SEER limited database from 1975 to 2010 for patients who were diagnosed with primary retinoblastoma (ICD-03 histology code: 9510-9513) of the retina/eye (ICD-03 site codes: retina – C69; eye NOS – C699). The SEER database currently captures approximately 28% of all newly diagnosed cancer cases in the United States from 18 geographic regions of the general US population. Patients with prior diagnosis of malignancy were excluded from analysis. Institutional board review was obtained at Erlanger Hospital at the University of Tennessee–Chattanooga.
Demographic variables such as age, sex, race, year of diagnosis, and geographic location (according to the SEER registry) were collected. Pathologic information included grade and extent of disease, which was characterized as localized, regional, distant, and unstaged. Treatment for retinoblastoma was categorized by surgical therapy, radiation therapy, chemotherapy, or any combination thereof. Given that a majority of patients underwent multiple treatments, the following 3 groups were created to allow for comparison: radiation ± other treatment, surgery ± other treatment, or chemotherapy ± other treatment for 4 different time periods. These groups are not mutually exclusive and thus the total percentages do not sum to 100% for a given year. Specific approval from SEER was gained to obtain chemotherapy treatment information, as this is not available in the publicly available SEER limited dataset.
All survival focused on the patients with localized or regional disease (n = 1220). Of these, 12 patients were removed (11 with no survival time, and 1 that was diagnosed at autopsy), resulting in 1208. The median follow-up time for this group was 103 months (calculated by the reverse Kaplan-Meier method). Study cutoff was November 30, 2012. Overall survival (OS) was calculated using the Kaplan-Meier method with comparison of survival using the log rank test. Statistical analysis was performed using SEER STAT software.
Second malignant neoplasms were identified by using the multiple primaries dataset in patients who underwent treatment with surgery alone (n = 665), radiation or radiation + surgery (n = 90), and chemotherapy or surgery + chemotherapy (n = 372). Observed-to-expected (O/E) ratios were calculated with the use of age-specific incidence rates (National Cancer Institute’s [SEER] Program). For patients who developed multiple malignancies after the primary disease, all of the subsequent malignancies were counted in the numerator of specific incidence rate (SIR), in agreement with the SEER Program’s incidence calculation. SIRs were evaluated via a Poisson distribution.
Results
Overall Cohort
A total of 1452 cases of retinoblastoma were identified from 1975 through 2010. Table 1 displays the general demographic information from this cohort. The mean age at diagnosis was 1 (range: 0–35), with 48% of patients being male. The cohort was 74.1% white, 13.6% black, and 9.2% Asian.
| Age (y) | 1 (range: 0–35) |
| Sex | |
| Male | 704 (48.5%) |
| Female | 748 (51.5%) |
| Race | |
| White | 1076 (74.1%) |
| Black | 198 (13.6%) |
| Asian or Pacific Islander | 133 (9.2%) |
| American Indian/Alaska Native | 22 (1.5%) |
| Unknown | 23 (1.6%) |
| Location | |
| Atlanta (Metropolitan) – 1975+ | 84 (5.8%) |
| California – 2000+ | 184 (12.7%) |
| Connecticut – 1975+ | 89 (6.1%) |
| Detroit (Metropolitan) – 1975+ | 130 (9.0%) |
| Greater Georgia – 2000+ | 85 (5.9%) |
| Hawaii – 1975+ | 38 (2.6%) |
| Iowa – 1975+ | 105 (7.2%) |
| Kentucky – 2000+ | 41 (2.8%) |
| Los Angeles – 1992+ | 221 (15.2%) |
| New Jersey- 2000+ | 68 (4.7%) |
| New Mexico – 1975+ | 52 (3.6%) |
| Rural Georgia – 1992+ | 4 (0.3%) |
| San Francisco–Oakland – 1975+ | 108 (7.4%) |
| San Jose–Monterey – 1992+ | 48 (3.3%) |
| Seattle (Puget Sound) – 1975+ | 112 (7.7%) |
| Utah – 1975+ | 83 (5.7%) |
| Year of diagnosis | |
| 1975–1979 | 88 (6.1%) |
| 1980–1989 | 204 (14.0%) |
| 1990–1999 | 360 (24.8%) |
| 2000–2010 | 800 (55.1%) |
| Laterality | |
| Unilateral | 1021 (70.3%) |
| Bilateral | 431 (29.7%) |
| Historic stage | |
| Localized | 1039 (71.5%) |
| Regional | 181 (12.5%) |
| Distant | 113 (7.8%) |
| Unknown | 119 (8.2%) |
| Development of secondary malignancy / trilateral RB | |
| Secondary malignancy | 37 (2.5%) |
| Trilateral RB | 8 (0.6%) |
| Total secondary neoplasms | 45 (3.1%) |
Most cases of retinoblastoma in the cohort were recorded between 2000 and 2010 (55%). Twenty-eight percent of patients had bilateral disease. At initial presentation 71.6% had localized disease, 12.5% had regional disease, and 7.8% had distant disease, with 8.2% of patients having unknown stage.
Treatment Trends
For the 1220 patients with localized or regional disease, 55.1% (n = 672 patients) underwent surgical intervention alone, while the next most common intervention was surgery + chemotherapy (n = 317 patients; 26.0%) ( Table 2 ). Treatment trends over time were grouped by patients who underwent radiation ± other treatment, surgery ± other treatment, or chemotherapy ± other treatment ( Table 3 ). There was a slight decrease in utilization of surgery, from 96.2% to 88.5%, a dramatic decrease in utilization of radiation therapy (15.2% to 4.9%), and an increase in utilization of chemotherapy from 16.5% to 50.2% ( Figure 1 ). There was an increase in radiation use of 31.6% in the 1980–1989 time period, followed by a decrease to 16% in the 1990–1999 time period.
| Initial Treatment | 1975–1979 | 1980–1989 | 1990–1999 | 2000–2010 | Total |
|---|---|---|---|---|---|
| Surgery only | 58 (73.4%) | 114 (64.4%) | 179 (64.6%) | 321 (46.7%) | 672 (55.1%) |
| Radiation only | 2 (2.5%) | 15 (8.5%) | 11 (4.0%) | 2 (0.03%) | 30 (2.5%) |
| Chemotherapy only | 0 (0%) | 0 (0%) | 3 (1.1%) | 55 (8.0%) | 58 (4.8%) |
| Surgery + radiation | 6 (7.6%) | 33 (18.6%) | 19 (6.9%) | 2 (0.3%) | 60 (4.9%) |
| Surgery + chemotherapy | 9 (11.4%) | 3 (1.7%) | 45 (16.2%) | 260 (37.8%) | 317 (26.0%) |
| Chemotherapy + radiation | 1 (1.3%) | 0 (0%) | 3 (1.1%) | 5 (0.7%) | 9 (0.7%) |
| Surgery + radiation + chemotherapy | 3 (3.8%) | 8 (4.5%) | 13 (4.7%) | 25 (3.6%) | 49 (4.0%) |
| None or unknown | 0 (0%) | 4 (2.3%) | 4 (1.4%) | 17 (2.5%) | 25 (4.0%) |
| TOTAL | 79 (100%) | 177 (100%) | 277 (100%) | 687 (100%) | 1220 (100%) |
| Therapy | 1975–1979 | 1980–1989 | 1990–1999 | 2000–2010 |
|---|---|---|---|---|
| Radiation (± other treatment) | 15.20% | 31.60% | 16.60% | 4.90% |
| Chemotherapy (± other treatment) | 16.50% | 6.20% | 23.10% | 50.20% |
| Surgery (± other treatment) | 92.20% | 89.30% | 92.40% | 88.50% |
Survival
The Kaplan-Meier survival curves by time period show an overall improvement in survival in more recent time periods ( Figure 2 ). The 5-year overall survival was 93.7%, 93.7%, 97.5%, and 97% for increasing time periods (1975–1979, 1980–1989, 1990–1999, and 2000–2010, respectively; P = .029).
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
