Incidence of second nonocular cancers following the diagnosis of retinoblastoma in patients who received radiotherapy.
Abramson DH, Du TT, Beaverson KL. (Neonatal) Retinoblastoma in the First Month of Life. Arch Ophthalmol. 2002;120(6):738-742. doi:10.1001/archopht.120.6.738
Copyright 2002 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2002
To identify patients with retinoblastoma whose conditions were diagnosed at the age of 1 month or younger and to describe their clinical features (including ocular and patient survival) and the development of second nonocular tumors.
Materials and Methods
A retrospective study of 1831 patients. The cumulative incidence of second cancer development was analyzed using the Kaplan-Meier method.
Forty-six patients were identified as having a diagnosis of retinoblastoma at the age of 1 month or younger (mean age, 18.5 days). Family history (31 patients [67%]) exceeded leukocoria (6 patients [13%]) as the most common reason for detection. Twenty-six (56%) of the 46 patients were seen with unilateral retinoblastoma, with 22 ultimately developing cancer in the fellow eye. At the initial diagnosis, 81 (85%) of the 95 tumors were detected in zones 1 and 2. Eighty-two (93%) of the 88 subsequent tumors were located in zones 2 and 3. In the 26 patients who had unilateral retinoblastoma, 16 of the initially affected eyes and 21 of the fellow eyes were salvaged. In the 19 (44%) of 20 patients who were seen initially with bilateral retinoblastomas, 31 (82%) of the 38 eyes were salvaged. The mean follow-up was 10.9 years. The incidence of second nonocular cancers reached 54% by 23.7 years for the patients who received radiation therapy, while the incidence was 0% for the patients who did not. Four (8.7%) of the 46 patients developed metastatic disease and died; 3 of these patients had documented metastases in the first month of life (one at birth).
The most common manifesting sign of children diagnosed as having retinoblastoma in the first month of life is family history. Eyes with Reese-Ellsworth group I retinoblastomas were the most common. In patients with bilateral and unilateral retinoblastoma, new (subsequent) ocular tumors developed in a centrifugal pattern. Despite an early diagnosis, patients' eyes came to enucleation, and metastatic disease and death occurred from ocular metastases. In patients who received radiation therapy, the probability of developing second nonocular cancer is 54% by 23.7 years; no second cancers developed in patients who did not receive radiation therapy.
CHILDREN WHO have retinoblastoma usually receive its diagnosis at a young age. In the United States the mean age at diagnosis for unilaterally affected children is 25 months; while for bilaterally affected patients, it is 15 months.1 When there is a known family history and children are screened for the disease, the mean age at diagnosis is younger than 1 year.2
Prior studies have demonstrated some interesting differences exhibited by children whose condition was diagnosed in the first year of life2 and those whose condition was diagnosed in the first 6 months of life.3 Some of these are expected, others are unexpected. As expected, not only are these children's condition diagnosed at a younger age, but also their laterality is different (more commonly bilateral when diagnosed in the first year and 6 months), and their proclivity to develop subsequent intraocular tumors after treatment is greater. Surprisingly, despite the diagnosis within 1 year, or even 6 months of life, the most common intraocular Reese-Ellsworth group at diagnosis was group V (ie, massive tumors involving more than half of the retina and vitreous seeding) and the most common manifesting sign or symptom was leukocoria.1,2
As education to families with the heritable form of the retinoblastoma gene has expanded and molecular and cytogenetic techniques become more available, we have been examining children whose conditions are diagnosed at even earlier ages and we realized that the subset of children whose conditions are diagnosed in the first month of life had some very unusual and instructive features. Because there are no studies on this subset of children, we reviewed our experience at the Robert M. Ellsworth Ophthalmic Oncology Center, New York Presbyterian Hospital, New York.
A retrospective medical record review was carried out of all patients examined by us at the Robert M. Ellsworth Ophthalmic Oncology Center and who were diagnosed as having retinoblastoma in the first 4 weeks of life. Forty-six patients were identified; 25 patients received their diagnosis before 1990 and 21 since 1990. The following clinical data were collected: sex, family history, age at diagnosis (in days), manifesting signs and symptoms, laterality, eyes involved at the initial visit, stage of ocular disease, mean number of tumors, location of tumors, development of new ocular tumors (number and location of tumors, if any), development of second nonocular tumor, initial treatment of the tumor, length of follow-up, survival of the individual eye, and survival of the patient. Tumor location was characterized by central vs peripheral retina, using a standard retinal drawing with a macular center with a classification system that was previously published.4 Zone 1, the posterior pole, encompassed a circle centered at the macula with a radius of 1 disc diameter beyond the optic nerve. Zone 2, the equatorial zone, spanned from the periphery of zone 1 to the equator. Zone 3 included the anterior retina from the periphery of zone 2 to the ora serrata.4 Incidence of second cancer was analyzed by life-table analysis as described previously.5 Trilateral (pinealomas) retinoblastomas were classified as second nonocular neoplasms.
Forty-six patients were identified with a mean follow-up of 10.9 years(range, <1-50.9 years; median, 4 years). Patients' characteristics are listed in Table 1 and the locations of tumors are listed in Table 2.
Of the 46 children whose conditions were diagnosed in the first month of life, 26 were seen with unilateral tumors; 20 were seen with bilateral tumors. However, 22 (85%) of the 26 patients who were seen with a unilateral tumor developed tumors in the fellow eye. Thus, 42 patients (91%) eventually had bilateral disease during our follow-up period.
Prior to 1990, 21 (84%) of the 25 patients received external beam radiotherapy as their initial treatment. After 1990, however, none of the 21 patients were treated initially or subsequently by irradiation.
None of the patients who were treated without irradiation have developed second cancers to date, with a median follow-up of 4 years. Six of the patients who received irradiation developed second nonocular tumors (Figure 1). All second tumors were in the head—there were 2 pinealomas, 2 soft tissue sarcomas, 1 osteosarcoma, and 1 fibrous histiocytoma. Life-table analysis revealed that the incidence of second cancers in the patients who were irradiated was 56.4% by 23.7 years.
Patients' ocular survival, categorized by the Reese-Ellsworth clinical classification system at the initial visit, is given in Table 3. Notable is 1 patient who was seen at birth with bilateral retinoblastoma and concurrent metastatic disease. Eight (17%) of the 46 patients died during our follow-up, 4 died of metastatic retinoblastoma, and 4 died of second nonocular cancer.
Our medical record review included 46 patients who had had the diagnosis of retinoblastoma made in the first 4 weeks of life. To our knowledge, this is the largest collection of neonatal retinoblastoma reported and reports both new information and clarifies prior, smaller reports.
Neonatal cancer is estimated to occur between 1 in 16 666 births6 and 1 in 12 500 births.7 This translates into 130 cases a year in the United States; half are diagnosed in the first 24 hours of life. Although not all series agree, the most common neonatal cancers are leukemia, sarcomas, teratomas, neuroblastomas, and central nervous system tumors. In some series no cases of retinoblastoma were found. For example, in a review from Melbourne, Australia, from 1939 to 1989, there were no retinoblastomas in the series of 46 neonatal cancers.8 Similarly, in the report from the Children's Hospital of Philadelphia of 22 neonates with cancer, none had retinoblasoma.9 Some other series do report an occasional case of neonatal retinoblastoma. Of 99 cases of neonatal cancer collected from the West Midlands Health Authority Region (United Kingdom) between 1960 and 1989, 2 cases of retinoblastoma were found.10 Of 23 neonates with cancer from Duke University, Durham, NC, 4 had retinoblastoma.11 In Toronto, Ontario, where there is a large retinoblastoma center, 17 cases of neonatal retinoblastoma were reported in a total of 102 children with neonatal cancer.12 Thus, the series of 42 patients that we report represents more than 50% of all cases ever reported and the largest collected from 1 institution. A number of features of these patients suggest that they are a special group of children with some distinct and different characteristics.
In a previous retrospective medical record review of 1265 patients of all ages with retinoblastoma from our center, 32 distinct manifesting signs were identified.13 The 4 most common signs were leukocoria (56.2%), strabismus (23.6%), poor vision (7.7%), and family history (6.8%). With a younger age of diagnosis, however, family history is the more common manifesting sign. For example, of 158 children who were diagnosed as having retinoblastoma in the first 6 months of life, it has previously been reported that 16% were seen because of family history.3 For patients in this study manifesting signs were the reverse of the general retinoblastoma population: 67% (31 patients) were seen because of a family history and 13% (6 patients) were seen because of leukocoria. Despite the very early age at diagnosis, 13% were still seen because of leukocoria.
The correlation between tumor detection time and retinal topography followed a central-to-peripheral distribution. At the initial examination, most tumors were located posterior to the equator (regardless of laterality), while subsequent new tumors were usually located anterior to the original tumors and never in the fovea. In fact, most patients and eyes developed subsequent, additional tumor foci after diagnosis and successful treatment of the manifesting tumors. This central-to-peripheral ("centrifugal") development of new tumors has been previously documented in eyes with bilateral retinoblastomas and has important practical implications for the clinicians and the patients.4 For clinicians, it means that they should expect to see subsequent new, anteriorly situated tumors in the eyes of patients with bilateral retinoblastomas independent of the method of treating the eye. For the patient (and clinician), the observation that a new, subsequent tumor never developed in the fovea is reassuring. It is also useful to inform families that these new tumors are usual and expected. The follow-up schedule for patients whose conditions were diagnosed in the first month of life must be adjusted so that these tumors are detected at an early stage.
In this study, 22 (85%) of the 26 patients who initially were seen with tumors in 1 eye eventually developed bilateral disease. Previous study showed that 20% of the patients having unilateral disease diagnosed in the first 6 months of life subsequently developed bilateral disease.3 Clinicians must be aware (and they must inform families) that children diagnosed as having unilateral retinoblastoma in the first month of life will usually go on to develop bilateral disease.
As summarized in Table 3, the most common intraocular disease stage of this age group was Reese-Ellsworth group I (ie, solitary tumor, <4 disc diameters, at or behind the equator and multiple tumors, none >4 disc diameters, all at or behind the equator). This was also true of the cohort of patients whose conditions were diagnosed in the first 3 months of life14 but not true of those whose conditions were diagnosed in the first 6 months3 or 12 months of life.2 Though this was expected, it is worth emphasizing that despite the early age at diagnosis 13% of these eyes were classified as group V eyes at the initial visit; early age of diagnosis does not guarantee an early stage of intraocular (or extraocular) disease.
We successfully salvaged 68 (79%) of the 86 diseased eyes: 16 (62%) of the 26 manifesting eyes, 21 (95%) of the 22 fellow eyes that subsequently developed tumors, and 31 (82%) of the 38 eyes in patients who were seen with bilateral disease. Although all bilateral eyes with Reese-Ellsworth group I through III classifications (ie, group II: solitary tumor, 4-10 disc diameters, at or behind the equator and multiple tumors, 4-10 disc diameters, behind the equator; group III: any lesion anterior to the equator and solitary tumor,>10 disc diameters, behind the equator) were initially managed without enucleation, progressive disease forced us into enucleation in 1 (2%) of the 40 eyes in group I, 2 (8%) of the 11 eyes in group II, and 1 (20%) of the 5 eyes in group IV (ie, multiple tumors, some >10 disc diameters and any lesion extending anteriorly to the ora serrata). More than 90% (10/11) of the group V eyes required enucleation initially or after external beam radiotherapy. Even with bilateral retinoblastoma diagnosed in the first 4 weeks of life, we still had to enucleate 18 (21%) of the 86 diseased eyes, especially when the eyes were classified as Reese-Ellsworth group V.
Retinoblastoma occurs in 2 forms—germinal and nongerminal. All patients who have bilateral disease or a positive family history are assumed to have the germinal mutation.15,16 In this study, 42 of 46 patients eventually developed bilateral disease. Of the remaining 4 patients who have unilateral disease, 3 had a positive family history and 1 had an unknown family history. Therefore, at least 45 (98%) of these 46 patients had the germinal mutation. Clinicians must keep this fact in mind when decisions are made about the treatment (especially external beam irradiation) and follow-up of unilateral retinoblastoma when the condition is diagnosed during the first month of life.
Second nonocular cancers in children with the germinal form of retinoblastoma are well known.17- 23 The cumulative incidence of second cancer is 1% per year, reaching 51% (SD, 6.2%), 50 years after the diagnosis of retinoblastoma.24 Factors that have been shown to contribute to this include the presence of the RB1 mutation, treatment with radiation, dose of radiation, presence of lipomas, and recently, patient age at irradiation.23,25,26
Prior reports of neonatal retinoblastoma suggest a high incidence of second cancers. For example, in the report on neonatal cancer from Denmark, 2 cases of retinoblastoma were found and 1 patient died of a subsequent second cancer (osteosarcoma).27 Similarly, of the 4 neonatal retinoblastomas reported from Duke University, 2 patients developed trilateral retinoblastoma.11 Of the 14 cases reported with neonatal retinoblastoma from the United Kingdom, 3 patients developed trilateral retinoblastoma and 2 others developed sarcomas in the irradiated field.
Although our data suggest similar alarming patterns, we must be careful in drawing firm conclusions because of the few patients in our study. However, we have almost equal numbers of patients with bilateral retinoblastoma whose conditions were diagnosed in the first month of life treated with radiotherapy(21 patients) or without irradiation (25 patients). Follow-up is different for these 2 groups, as all of the patients who received radiation therapy were treated before 1990 (and, therefore, have a longer follow-up period) and most of those who did not receive radiotherapy have been treated since 1990 (with a shorter follow-up period). Life tables are usually used to adjust for such problems but in our series most of the patients who received no irradiation have been diagnosed and followed up for fewer than 10 years.
Life-table analysis of the irradiated group (21 patients) revealed a second cancer incidence of more than 56.4% by 23.6 years after the diagnosis of retinoblastoma. Not only is this higher than the 1% per year incidence but it is also higher than the 2% per year that we previously reported in children irradiated in the first year of life. This strongly suggests that the children with bilateral retinoblastoma diagnosed and treated in the first month of life are at the highest risk for the development of second cancers.
Although follow-up is relatively short for the patients who did not receive irradiation based on the results here and in prior publications that have demonstrated a low but important increased incidence of second cancers in patients with nonirradiated retinoblastoma, we suspect that the differences between the patients who were irradiated and not irradiated will prove to be statistically and clinically significant.
If radiotherapy for ocular tumors is contemplated in the patients whose condition was diagnosed in the first month of life, the consequences for subsequent second cancer development must be carefully weighed and explained to the family. Not only are these children at risk for subsequent second cancers, they may actually be even more susceptible to the harmful effects of irradiation because of their early age.
In general, children with neonatal cancers have poor survival rates compared with older children who were diagnosed as having cancer. Although many do survive, the death rate (approximately 50%) and complications of treatment are significant. Prior reports have also emphasized a high death rate in children with neonatal retinoblastoma. For example, of the 17 cases reported from Toronto, 4 patients (24%) died. Of the 4 cases reported from Duke University, 2 patients died of retinoblastoma. Four (8.7%) of our patients died of metastatic retinoblastoma. In 1 case metastatic disease was evident at birth. In 2 others metastatic disease was detected within the first month of life. A fourth case had buphthalmos and rupture of the globe during surgery, leading to orbital tumor and metastatic disease. Children detected with retinoblastoma in the first month of life may manifest and/or develop metastatic disease and die. The very early diagnosis of retinoblastoma is still associated with significant mortality.
Twenty-eight (61%) of our 46 patients were girls. While this did not attain statistical significance in our study, 70% of all children with neonatal cancers are girls,11 though some series have a male preponderance.26
This cohort of patients whose condition was diagnosed in the first month of life did well, although some of these children did not develop central vision (because of tumors in the macula of one or both eyes), some did not retain their eyes, some died of metastatic disease, and many progressed to develop fatal second cancers apparently related to the therapeutic radiation that effectively cured the ocular cancer. The very early diagnosis of retinoblastoma does not guarantee vision, ocular, or patient survival and may have contributed to subsequent death from second nonocular cancers.
Submitted for publication June 7, 2001; final revision received February 12, 2002; accepted February 28, 2002.
This study was supported in part by a grant from the Sam and May Rudin Family Foundation, New York, NY (Dr Abramson).
Corresponding author and reprints: David H. Abramson, MD, 70 E 66th St, New York, NY 10023 (e-mail: ICancerMD@aol.com).