Screening for Retinoblastoma: Presenting Signs as Prognosticators of Patient and Ocular Survival

METHODS

We conducted a retrospective review of 1831 retinoblastoma patients who were seen at our center between 1914 and June 2000. Data were collected from patient charts, including sex, disease laterality, date of diagnosis, age at diagnosis, the presenting sign, the presenting eye (oculus dexter = right eye, oculus sinister = left eye, oculus uterque = both eyes), the individual who initially noted the presenting sign that led to the referral to our specialty center, the method of detection, the Reese-Ellsworth (RE) group of the presenting eye(s), whether there was a family history of retinoblastoma known at the time of diagnosis, the enucleation status of presenting eyes and fellow eyes (if involved), last date of follow-up for the patient, the vital status of the patient, and whether the patient was known to have developed a second neoplasm.

We grouped the 32 different presenting signs previously described by our center⁹ into 8 classifications for the purpose of comparing the major presenting signs of retinoblastoma: leukocoria, strabismus (nonspecified, esotropia, and exotropia), “poor vision,” family history, “red eye,” proptosis (proptosis and growth/metastases), unknown, and “other.” Of the 7 categories established for who detected the initial presenting sign, 5 major groups were used for our analyses: family/friends, pediatrician/primary care physician, ophthalmologist, other (school, patient, other), and unknown.

All presenting eyes were classified according to the RE Classification System for retinoblastoma¹⁴; “a” and “b” subtypes were grouped. We reported the RE frequencies for right eyes because previous studies have demonstrated that there is no difference in the distribution of tumors between right and left retinas.¹⁵

For all patients in this series (AP), patients without a family history of retinoblastoma (−FH); patients with a family history of the retinoblastoma (+FH); and patients who had a known family history and presented to our center because of their family history (screened family history [SFH]), demographic, and outcome information were analyzed using statistical methods. Patients for whom the family history status of retinoblastoma was not known or not recorded were not included in subset analysis of patient or ocular outcome. We evaluated patient survival during the follow-up period from the date of diagnosis in the following cohorts: all −FH patients, −FH patients who presented with leukocoria or strabismus, all +FH patients, +FH patients who received tumor surveillance by routine ophthalmic examinations, and +FH patients who did not receive screening. We excluded deaths as a result of other primary neoplasms from our analysis to obtain a clear understanding of the survival probability as a result of metastatic retinoblastoma.

Ocular survival was evaluated for all presenting eyes during the follow-up period from the date of diagnosis in the following patient groups: all −FH patients, −FH patients who presented with leukocoria or strabismus, all +FH patients, +FH patients who received tumor surveillance by routine ophthalmic examinations, and +FH patients who did not receive screening. The presenting eye is defined as the eye(s) in which the abnormality was initially noted. A diagnosis of retinoblastoma was defined as the presence of 1 or more retinal tumors detected on a dilated funduscopic examination using indirect ophthalmoscopy and scleral depression. For ocular survival analysis, the event variable was enucleation. We used the Kaplan-Meier method for incidence and survival analyses. We compared survival among cohorts using the log rank test with a 5% significance level. All statistics were calculated using WinSTAT add-in for Microsoft Excel version 2000.1 (R. Fitch Software; Cambridge, MA).

RESULTS

Outcome Data

Patient survival and ocular survival were compared in patients who presented with leukocoria, strabismus, or family history in Tables 3 to 6⇓⇓⇓⇓ and Figs 1 to 4⇓⇓⇓⇓. All survival data specifically exclude patient deaths as a result of second, nonocular tumors.

• Download figure
• Open in new tab
• Download powerpoint

Fig 1.

Kaplan-Meier survival curves for −FH patients who initially present at the time of diagnosis with leukocoria or strabismus.

• Download figure
• Open in new tab
• Download powerpoint

Fig 2.

Ocular survival curves for all eyes initially presenting at the time of diagnosis with leukocoria or strabismus in −FH patients. The endpoint of the Kaplan-Meier analysis is enucleation.

• Download figure
• Open in new tab
• Download powerpoint

Fig 3.

A positive family history of retinoblastoma: impact on patient survival and ocular outcome. Patient survival analysis excluded deaths from second, nonocular cancers. Ocular survival was calculated for all eyes that initially presented with retinal tumors, and enucleation was the event variable.

• Download figure
• Open in new tab
• Download powerpoint

Fig 4.

Patients with a positive family history of retinoblastoma: impact of early tumor surveillance on patient and ocular survival. Patient survival analysis excluded deaths from second, nonocular cancers. Ocular survival rates were calculated for all eyes that initially presented with retinal tumors.

View this table:

• View inline
• View popup

TABLE 3.

Patient Survival Rates at 1 and 5 Years for −FH Patients Who Presented With Leukocoria or Strabismus

View this table:

• View inline
• View popup

TABLE 4.

Ocular Survival Rates at 1 and 5 Years for All Presenting Eyes With Leukocoria or Strabismus in −FH Patients

View this table:

• View inline
• View popup

TABLE 5.

Impact of Positive Family History Status on Patient Survival and Ocular Survival Rates

View this table:

• View inline
• View popup

TABLE 6.

+FH Patients: Impact of Early Screening on Patient Survival and Ocular Survival

DISCUSSION

Retinoblastoma is the seventh most common pediatric malignancy in the United States,¹⁶ with an overall incidence of 1/20 000. It is the only pediatric cancer routinely screened for in the general population of the United States. The current screening protocol for retinoblastoma is a clinical examination, with disease detection dependent on the physician’s recognition of at least 1 of the many well-described presenting signs for retinoblastoma. The Committee on Practice and Ambulatory Medicine of the American Academy of Pediatrics (AAP) recommends that a physician perform an eye evaluation at all well-infant and well-child visits, beginning in the newborn period. Between 9 and 11 preventive pediatric health care visits occur from birth to 2 years of age, and the eye evaluation at each visit should include the red reflex test as a screening test for abnormalities in the posterior segment and opacities of the visual axis.^17,18 The American Academy of Ophthalmology also supports pediatric eye screening evaluations during the newborn period and at subsequent visits by a physician or a nurse.¹⁹ The red reflex test is recommended at newborn to 3 months, 6 months to 1 year, 3 years, 5 years, and >5 years. The demonstration of a white pupillary reflex (leukocoria) is an indication for a referral. An assessment for strabismus is also made at a well-child visit.¹⁶ Despite an apparent consensus that screening for retinoblastoma ought to be performed by looking for leukocoria and strabismus, there is little information regarding the prognostic value of the clinical criteria used to detect retinoblastoma or the effectiveness of these screening recommendations.

Our large retrospective study included 1831 retinoblastoma patients, the majority of whom presented with leukocoria (54%) or strabismus (19%). Despite a screening mechanism in place for retinoblastoma, most patients (80%) had their presenting sign initially detected by a family member or a friend. Of those patients who presented with leukocoria, an even larger percentage (92%) were detected by family/friends. Pediatricians detected 8% of all presenting signs; <4% of leukocoric-presenting eyes of −FH and 7% of strabismic eyes in −FH patients.

According to pediatricians, this low detection rate is not attributable to noncompliance with AAP eye examination procedures; 94% to 98% of pediatricians reported routinely performing the red reflex test in surveys performed in 1988, 1993, and 1998.^20,21 If we assume, therefore, that pediatricians are routinely screening children in the United States for retinoblastoma by checking for leukocoria, then this strategy seems to be inadequate for disease detection. A number of reasons that pediatricians may miss the detection of leukocoria have been suggested. These include the following: 1) tumors that arise in the periphery may be observed by family and friends at home who have access to a view of the child’s eye from multiple angles versus a pediatrician looking at the patient from directly forward,²² 2) underutilization of well-child care visits despite financial coverage of well-child visits,²³ 3) bright lights in a pediatric office, 4) uncooperative children,²⁴ 5) miotic pupils, 6) improper red reflex testing techniques, and 7) a low clinical suspicion. Despite possible explanations, it remains a striking observation that family members and friends detect >90% of cases of leukocoria.

Leukocoria itself seems to be a reliable endpoint for detecting local and therefore curable disease. It is associated with RE group V disease⁵: group Va indicates the presence of massive tumors involving more than half of the retina, and group Vb indicates vitreous seeding. The majority of presenting eyes in our study were classified as group V, and, despite laterality of disease, although patient survival rates were high in patients who presented with leukocoria or strabismus (Table 3), ocular survival rates were low (Table 4). A total of 91% of all eyes that presented with leukocoria in the −FH patient cohort were lost via enucleation by 5 years; of 702 unilateral patients, <4% of leukocoric eyes survived, and of 828 bilateral patients, <30% survived. Perhaps intuitive is the observed higher ocular survival rate in bilateral versus unilateral patients who present with leukocoria or strabismus, as more aggressive treatments may have been used in an attempt to save an eye(s) in the bilateral patients. Ocular survival was significantly improved in all eyes that presented with strabismus compared with leukocoria, but survival was still poor with 83% of all strabismic eyes enucleated by 5 years. No significant improvement was observed in the ocular salvage rates of eyes that presented with strabismus versus leukocoria in bilateral patients. Our study was a historical overview of our experience, evaluating patient and ocular survival from the point of disease detection, in particular, detection of leukocoria, strabismus, and a positive family history. We did not distinguish between earlier treatment modalities (external beam radiation, enucleation) and newer alternative therapies such as chemotherapy, cryotherapy, brachytherapy, and laser therapy. Undoubtedly, changes in the treatment of retinoblastoma have and will continue to have an impact on survival rates, in particular, ocular survival and the preservation of useful vision.

Patients with a known family history of retinoblastoma constitute a discrete and interesting subpopulation of retinoblastoma patients: they offer the opportunity to evaluate the impact of early disease awareness on disease detection and disease outcome. Approximately 10% to 15% of patients with newly diagnosed retinoblastoma have a positive family history of the disease, thereby inheriting their RB1 gene mutation.¹⁶ Germline RB1 alterations underlie heritable forms of retinoblastoma, including inherited cases of bilateral and unilateral disease, as well as bilateral cases in which there is no family history of the disease. Alfred Knudson’s 2 mutational event model for the development of retinoblastoma provided the statistical support for the observation that patients who inherit 1 RB1 gene mutation develop retinal tumors earlier (15 months) and, in most cases, multifocal tumors than patients who develop the nonhereditary form of the disease (32 months).²⁵ Individuals with a germline RB1 mutation are predisposed to develop retinal tumors in childhood and nonocular cancers throughout their lifetime.²⁶

The recognition of a positive family history of retinoblastoma has practical implications for patients and their family and clinicians with regard to detection and outcome. For the clinician, a positive family history is an indication for an ophthalmologic examination in the newborn nursery, as well as comprehensive, serial eye evaluations by an ophthalmologist.^17,19 Aggressive surveillance is recommended for all patients with a family history of retinoblastoma until at least 28 months. If tumors are discovered, then follow-up is recommended until 6.75 years of age.²⁷ Although there is no consensus on recommended screening protocols for nonocular cancers in survivors of heritable retinoblastoma, clinicians who follow such patients ought to be aware of the association for education and for clinical management purposes.

In families with a history of retinoblastoma, some parents of an at-risk individual may be cognizant of the cause and autosomal dominant pattern of inheritance of the disease, as well as the recommended earlier timing of eye examinations and screening. Their infants may present for routine clinical examinations from birth, not because the parents noted any ocular manifestations of the disease but because parents were aware of the potential benefits of an earlier diagnosis on outcome and the complications associated with a delay in diagnosis. An increased risk for local tumor invasion, particularly in younger patients who present with squint rather than leukocoria, and higher rates of death and blindness in a cohort of patients whose bilateral retinoblastoma was diagnosed between 1945 and 1970, have been described as complications associated with a delay in the diagnosis of retinoblastoma.^28,29 In our study, 264 patients were eligible for early tumor surveillance on the basis of the presence of a family history of retinoblastoma. Approximately one third of these retinoblastoma patients initially presented to our center in the newborn period because of the family history and subsequently received routine clinical examinations as screening for retinal tumors. Patient survival was high (>93% at 5 years), and the ocular salvage rate in the screened-patient group was improved by 44%, as demonstrated by a 68% rate survival of all presenting eyes at 5 years versus 38% ocular survival in the non-SFH cohort. Furthermore, they were diagnosed at an earlier stage of disease, with >50% of presenting eyes classified as RE group 1 (ie, solitary tumor, <4 disk diameters, at or behind the equator and multiple tumors, none >4 disk diameters, all at or behind the equator). These results are not surprising, given the possibility that more aggressive surveillance and (presumably) more aggressive attempts to control tumors were undertaken in the SFH cohort. It is surprising that 32% of the presenting eyes were still lost to enucleation. This latter observation may suggest that despite earlier disease detection and despite treatment modalities, eyes and vision are still lost in this disease.

The present recommended strategy for the detection of retinoblastoma relies on a pediatrician’s recognition of leukocoria, as demonstrated by an abnormal red reflex test. Our data suggest that the red reflex test may be an inadequate tool for the detection of retinoblastoma and/or is not being performed as recommended. For improving the accuracy of the red reflex examination, perhaps a stronger emphasis needs to be placed within pediatric training programs and within continuing education programs for physicians on how to perform this test correctly. A recent policy statement released by the AAP’s Section on Ophthalmology describing the indications for and the technique to perform the red reflex examination³⁰ reflects a growing effort in addressing the issue of the screening of young eyes. In fact, a limitation of our current study may be that we assessed outcome of screening practices over an 86-year period, when an analysis of outcome by years might yield additional information about screening current practices.

In addition to improving methods for detection, the point at which we seek to detect retinoblastoma needs reevaluation: detection at the point of leukocoria, according to our study, is effective for saving lives but ineffective for saving eyes and vision. The data generated from those patients who had a positive family history of retinoblastoma and were screened for retinal tumors from birth suggests that detection at an earlier stage will allow more children to be treated successfully so that fewer eyes need to be removed. Although one third of +FH patients presented for early tumor surveillance, one third still presented with late-stage disease and leukocoria. This observation emphasizes the need to develop and implement strategies that 1) educate families with heritable retinoblastoma regarding the benefits of early tumor surveillance and 2) raise awareness of retinoblastoma to geneticists, obstetricians, and pediatricians so that children with a family history of retinoblastoma can be referred to and seen by an ophthalmologist immediately. Similarly, if dilated funduscopic examinations of all children are proposed as a newborn screening-type approach to pediatric screening for retinoblastoma, then strategies need to be developed and implemented that effectively 1) detect retinal tumors over the at-risk period and 2) educate both primary care physicians and prospective parents about the benefits and limitations of newborn screening for retinoblastoma. However, until these new screening strategies are developed, implemented, and evaluated for their effectiveness, the need remains for a pediatrician to screen for retinoblastoma by accurately performing the red reflex examination, with an immediate referral on a positive or abnormal result. Similarly, an immediate referral for an ophthalmic evaluation is prudent when a pediatrician notes strabismus or a family history of retinoblastoma, as these are risk factors for the presence of retinal tumors.