Pretreatment fundus drawing andcorresponding visual field (merged central 30 and peripheral 30-60-2) (A)and fundus photograph (B) of a patient with retinoblastoma demonstrating precisecorrelation between tumor location and visual field defect.
Pretreatment fundus drawing andcorresponding visual field (merged central 30 and peripheral 30-60-2) of apatient with retinoblastoma demonstrating excellent correlation between multipletumors and discrete scotomas.
Pretreatment fundus drawing andcorresponding visual field (central 30-2 only) (A) and fundus photograph (B)of a patient with retinoblastoma with a small macular tumor that underwenta type 0 regression after external beam radiotherapy.
Pretreatment fundus drawing andcorresponding visual field (merged central 30 and peripheral 30-60-2) (A)and fundus photographs (B and C) of a patient with retinoblastoma with a visualfield defect larger than that suggested by tumor size alone. C is an enlargementof the area superotemporal to that shown in B. In A, dd indicates disc diameter.
Pretreatment fundus drawing andcorresponding visual field (merged central 30 and peripheral 30-60-2) of apatient with retinoblastoma demonstrating a brow-nose defect after externalbeam radiotherapy. This defect is caused by the anatomy of the patient's skull,not retinal disease. dd indicates disc diameter.
Pretreatment fundus drawing andcorresponding visual field (merged central 30 and peripheral 30-60-2) of apatient with retinoblastoma demonstrating a posterior tumor location and correspondingarcuate visual field defect. dd indicates disc diameter.
David H. Abramson, Mark R. Melson, Camille Servodidio. Visual Fields in Retinoblastoma Survivors. Arch Ophthalmol. 2004;122(9):1324–1330. doi:10.1001/archopht.122.9.1324
To describe the visual field defects in retinoblastoma survivors andrelate those defects to characteristics such as tumor size, tumor location,and treatment modality.
Thirty-one patients treated for retinoblastoma were included in thisstudy. Humphrey visual fields were determined in 33 eyes.
Twenty-seven patients (29 eyes, 68 tumors) had sufficient diagnosisand treatment data available for further analysis. Twenty-six of the 27 patientshad both absolute and relative visual field defects. Four types of visualfield defects were observed and correlated with location of the tumor andtherapy to the individual tumors: (1) no residual defect, (2) absolute scotoma,(3) arcuate and sector scotoma, and (4) "pseudo"–visual field defectscaused by relative enophthalmos resulting from radiation.
Patients with retinoblastoma demonstrate a variety of long-term visualfield defects after treatment for their intraocular disease. Characteristicsthat determine the size and type of defects are tumor size, tumor location,and treatment method.
A number of studies have examined various long-term outcomes in survivorsof retinoblastoma. Among the measures that have been thoroughly reviewed arepatient survival, tumor recurrence, ocular survival, visual acuity, motorand psychological development, and nonocular tumor incidence. As our abilityto treat these patients has significantly improved during the past few decades,the overwhelming majority of children in this country not only survive butalso retain near-normal to normal vision in at least 1 eye.1 Manytreatment centers are now further refining techniques to help preserve sightas well as the affected eye(s), while not compromising long-term patient survival.For patients with germline retinoblastoma mutations, much of this effort isconcentrated on therapies that minimize the risk of developing radiation-relatednonocular tumors later in life.2- 4
We are unaware of any studies specifically examining long-term visualfields in successfully treated patients with retinoblastoma. There are onlylimited case reports5 and other studies6- 8 that even mention thistopic in the current literature. Several authors have explored this questionin patients treated for ocular melanoma,9- 13 nonmalignantmelanocytoma,14 other benign choroidal neoplasia,15,16 and metastatic ocular disease,17,18 but here, too, there is a scarcityof literature.
Since retinoblastoma is a tumor of infancy and early childhood, it isimpossible for the physician to obtain pretreatment visual fields in patientswith this illness. However, many of these patients continue to be followedup closely throughout life by ophthalmologists, making it possible to performvisual field testing on them when they become young adults. Survival willremain the primary outcome measure in this patient population; however, wecannot ignore the morbidity caused by treatments that affect either centralacuity or peripheral vision. An examination of the visual fields of retinoblastomasurvivors may provide insight as to which tumor characteristics and therapiesdestroy (and which preserve) peripheral vision for these patients.
Thirty-one patients with retinoblastoma treated at our New York centerwere included in this study. Demographic and clinical data collected frompatient charts included age at retinoblastoma diagnosis, tumor lateralityand location, and family history of retinoblastoma. We also obtained previoustreatment data, including type and number of times a particular method wasused, fundus drawings, and fundus photographs, when available.
Using the Humphrey field analyzer (Model 630; Zeiss Humphrey Systems,Dublin, Calif), we performed a merged central 30-2, peripheral 30-60-2 thresholdtest on 26 of the 31 patients. A central 30-2 threshold test was performedon 2 patients and a merged central 30-1 threshold test on 1 patient. A mergedcentral 30-1, peripheral 30-60-1 threshold test was performed on 1 patientand a full-field 120-point screening test on 1 patient. Tests were performedwith appropriate near-vision correction for the central 30° and withoutcorrection for the peripheral 30° to 60°. Visual acuity was measuredin 26 eyes by means of the Snellen chart with best correction.
After undergoing visual field testing, 4 of the 31 patients were excludedfrom this study because of incomplete data regarding their retinoblastomatreatment.
All visual field test results were interpreted by the attending ophthalmologist(D.H.A.). An absolute scotoma was defined as 3 or more contiguous points ofgreater than 10 000 apostilbs (ASB) and a relative scotoma was definedas 3 or more contiguous points with a defect ranging from 251 to 10 000ASB. Using pretreatment fundus drawings and posttreatment fundus photographs(when available), we predicted the magnitude of the visual field defect andcorrelated its position on the retina on the basis of the size and locationof a particular tumor. We then classified defects as larger, smaller, or equalto the predicted finding and noted whether their locations were consistentwith our predictions.
Visual field testing was performed in 33 eyes in 31 patients. Of these,27 patients (29 eyes) had sufficient previous treatment data useful for additionalstudy. The average age at retinoblastoma diagnosis for the 27 patients was1.2 years (range, 1 week to 84 months). The average follow-up from date ofdiagnosis was 21.8 years (range, 9.8-38.8 years). There were 68 tumors inthe 29 eyes. The number of tumors per eye ranged from 1 to 7 (median, 2).Various treatment modalities, often combinations of 2 or more methods to controla single tumor, were used in these patients (Table 1)
Twenty-six of the 27 patients had both absolute (>10 000 ASB) andrelative (≥251 ASB, ≤10 000 ASB) visual field defects related totheir tumor (Table 1,Figure 1, andFigure 2). One patient had no detectable visual field defect directlyrelated to his tumor (Figure 3).Sixteen patients had detectable visual field loss in both the central 30°and peripheral 30° to 60° portion of the test. Twenty patients haddefects within the central 30° of the visual field, whereas 19 patientshad defects within the peripheral 30° to 60°. The near-vision correctivelens was inadvertently left in place for 3 patients in the study, thus resultingin a false ring scotoma in the peripheral 30° to 60° that correspondedto the border of the corrective lens. Visual acuity was tested in 26 of the29 eyes examined for visual field data. Of these, 24 were found to have 20/40or better visual acuity.
Of the 68 total tumors, 20 produced defects that were larger than wouldbe predicted on the basis of tumor size alone (Figure 4). Eight produced defects that were smaller than predictedand 19 produced defects that were approximately equal to what was predicted.We could not adequately assess the defect size in 1 patient, as the extentof disease-related retinal damage made it impossible to determine what portionof field loss was attributable to a particular tumor. The remaining 20 tumorswere located anterior to the equator, and the eyes involved did not demonstrateany visual field defects from these tumors by means of the Humphrey programswe used.
Nineteen patients had an enlarged brow-nose defect. For purposes ofthis study, an enlarged brow-nose defect was defined as a nasal defect extendinggreater than 10° nasally and a superior defect extending greater than20° superiorly (Figure 5). Fourpatients did not have an enlarged brow-nose defect. Of the remaining 4 patients,2 had only the central 30° of the visual field tested (thus making assessmentof the brow-nose defect impossible) and the remaining 2 had such large absolutefield defects that it was impossible to determine whether there was an enlargedbrow-nose defect.
All 27 patients had either absolute or relative defects that crossedthe vertical meridian. Twenty-three of the 27 patients had defects that crossedthe horizontal meridian. Fifteen tumors in 13 patients had at least a portionof their border in the arcuate nerve fiber pathway, but only 4 patients hadevidence of an arcuate visual field defect (Figure 6).
In summary, 4 types of defects were observed: (1) no residual defect(in a patient with a type 0 regression (Figure3); (2) an absolute scotoma corresponding to the residual tumorscar reflecting initial size and focal treatment (Figure 1 and Figure 2);(3) arcuate and sector scotomas representing disruption of nerve fiber transmissionthrough a scar (Figure 4A, Figure 4B, and Figure 6); and (4) a "pseudo"–field defect (the brow-nosedefect) resulting from the enophthalmos caused by radiation (Figure 4C and Figure 5).
Our analysis of computerized visual fields, attempted on 31 retinoblastomasurvivors, demonstrated that this type of testing can be completed in thispatient population. This encouraging finding provides evidence that more ofthese patients will be able to undergo such testing in the future to helpus further correlate defects in peripheral vision with tumor and treatmentcharacteristics.
In a previous study of visual fields of choroidal melanomas, Abramsonreported, "the visual field defect caused by a melanoma is characteristic,diagnostic, and almost unique."9 In that study,100% of the tumors were associated with both an absolute scotoma and a visualfield defect (relative scotoma) in the area of the tumor itself. Ninety-ninepercent of the tumors in that study produced defects that were larger thanthe tumors with which they were associated.9
Our visual field analysis in this study could be divided into 4 groups:no residual defect, absolute scotoma, arcuate scotoma, and pseudo–visualfield defect.
For a number of reasons, the defects seen in retinoblastoma do not appearto be as predictable as those seen in choroidal melanoma. Although the primaryfindings attributable to the tumor itself are essentially consistent withwhat we would expect on the basis of size and location, the effect of tumortreatment on the visual fields is more variable than with ocular melanoma.When comparing the findings among these tumor types, one must keep in mindthat we were able to much more precisely describe defects related to melanomathan to retinoblastoma. We were able to obtain pretreatment and posttreatmentvisual fields for patients with melanoma and were thus able to better assessthe effects of treatment on the patient's visual fields. Furthermore, patientswith retinoblastoma were often treated with multiple therapies, thus makingit much more difficult to clearly delineate the long-term effects relatedto one treatment vs another.
Of the 27 patients tested, all but one had a visual field defect thatcould be directly attributed to either a tumor or its treatment. This patienthad a solitary macular tumor (Figure 3)that was treated with external beam radiotherapy and subsequently underwenta type 0 regression pattern,19 with no evidenceof residual tumor or scar. Despite this tumor's location, this patient retained20/20 visual acuity in the eye. This suggests that external beam radiotherapymay be the only treatment modality that has the potential to not only curethe disease but also spare the patient from significant long-term visual complications.Although a type 0 regression pattern occurs in only 10% to 15% of tumors treatedwith radiotherapy,19 physicians who encounterit can likely expect no visual field defect related to retinal disease. Thisinteresting finding warrants further study of these tumors to better characterizetheir cells of origin, depth of retinal invasion, and specific mechanism(s)of cell death after treatment. It is important to note that patients who havetumors that undergo type 0 regressions may still have the brow-nose anatomicdefect that we describe in this study. In the patient described herein, wewere unable to perform peripheral 30° to 60° testing and thus wereunable to assess whether this patient had such a defect.
In the 26 patients with demonstrable visual field defects, the locationof the tumor was correlated with the center of the scotoma found in our testing(Figure 1 and Figure 2). Although this correlation has not been previously characterized,we assumed its existence on the basis of previous studies of other intraoculartumors. The nature of the damage to the retina that causes the visual fielddefect has not yet been definitively characterized in either retinoblastomaor choroidal melanoma. In Abramson's previous study of choroidal melanomas,the scotoma overlying the tumors had been presumptively attributed to photoreceptordamage in the region of the tumor.9 When thesize of the scotoma increased after therapy, he speculated that the increasewas due to radiation damage to photoreceptors or impairment of local circulationwith radiation retinopathy, or was secondary to a long-standing serous detachmentof the overlying retina. However, in several of the cases in that study, thesize of the scotoma decreased after treatment. Unfortunately, we were unableto compare pretreatment and posttreatment visual fields in this study andwere thus unaware of whether such seemingly temporary damage exists for retinoblastoma.Regardless of the mechanism of defect formation, the existence of the tumor-scotomacorrelation provides valuable data that should aid the treating ophthalmologistin predicting posttreatment visual impairment for these patients. Such informationcan be useful when discussing the long-term complications of therapy withthe patient and his or her family.
When symptomatic, patients with choroidal melanoma often have complaintsrelated to visual field defects when they are first examined. Measurementof these defects can aid the clinician in observing the progress of treatmentof a given lesion. Because of the ages at presentation for retinoblastoma,such complaints are virtually never elicited from these patients. Detailedvisual testing, including visual field analysis, is usually impossible becauseof the subjective nature of many of these tests. Thus, we cannot observe treatmentprogress in retinoblastoma by means of visual field testing. In this study,we compared the defect that would be expected on the basis of the tumor sizealone with the defect found after treatment and noted that larger defectswere most often associated with tumors treated with external beam radiotherapy.
Eighteen of the 28 patients in this study were found to have enlargedbrow-nose defects (Figure 5). Ofthis group, 14 had received external beam radiotherapy as part of their retinoblastomamanagement. This finding is not surprising when we consider the incidenceof orbital and midface hypoplasia associated with this treatment modalityin patients treated at the ages found in this study.20 Somepatients clearly manifested more pronounced physical defects than others,and the magnitude of brow-nose defects seen on visual field testing variedas well. From our observations, the extent of the defect was not necessarilycorrelated with the tumor burden of a particular eye, but with the facialand orbital structure of the patient. We can thus conclude that this visualfield defect is anatomic and not related directly to optic nerve or retinaldisease, since the changes did not correlate with areas directly impactedby a tumor. This finding is interesting and provides further justificationfor the cautious use of external beam radiotherapy in patients with retinoblastoma.In some cases, however, the benefits of external beam radiotherapy clearlyoutweigh the known risks associated with this treatment modality, especiallyfor tumors that have not responded to other therapies.
It is important to emphasize that the brow-nose defect we are describingoriginates from a change in the bony architecture of the child's skull. Inindividuals with normal orbits, the boundaries of the visual field extendapproximately 60° superiorly, 75° inferiorly, 100° temporally,and 60° nasally. The prominence of the brow accounts for the discrepancybetween the superior and inferior fields, and the nose accounts for that betweenthe nasal and temporal fields. The brow-nose defect is thus not restrictedto children who have received radiotherapy for retinoblastoma. Any change(or relative change) in the size of the patient's nose or brow can alter peripheralvision. A patient who has received radiotherapy to the skull or even healthychildren and adults who have prominent brows and/or noses may have a brow-nosedefect on visual field testing. In our series, 4 of the 18 patients who metthe criteria for this defect did not receive external beam radiotherapy aspart of their retinoblastoma treatment and thus fall into this latter category.
As alluded to already, the numerous concerns regarding the potentialside effects of external beam radiotherapy have caused physicians to becomemore wary of its use as a primary therapeutic choice in this patient population.In the past 25 years, much attention has been directed toward the increasedrisk of second tumors in patients with hereditary disease. In light of this,ophthalmic oncologists are increasingly turning to local methods of tumorcontrol. Of these, the 2 most frequently used treatments are photocoagulation(transpupillary thermotherapy) and cryotherapy. Recently, there has been significantinterest in using chemoreduction before focal therapy for improved tumor control.4,21- 25 Toreduce the systemic toxic effects associated with the chemotherapeutic agentsused, a number of centers have begun to evaluate periocular drug deliveryfor these patients.26- 28 Whilewe do not expect that the chemoreduction itself would contribute to a newvisual field defect (unless the agent itself is associated with significantretinal toxic effects), we must consider the effect that the local therapythat follows it might have on the patient's peripheral vision.
Cryotherapy was used in the treatment of 14 tumors in this study. Thismethod is most often used to treat small tumors anterior to the equator. Ofthe tumors examined in this study, 10 treated with cryotherapy were anteriorto the equator. The 4 tumors located posterior to the equator that receivedcryotherapy also underwent additional treatment with different methods. Weare therefore unable to separately evaluate the visual field defect specificallyassociated with cryotherapy. Its use, however, raises an interesting question.Do tumors anterior to the equator cause any visual field defects, and, ifso, what are they? In this study, we found no demonstrable visual field defectsfrom these lesions. From this, we can conclude that the therapeutic choicefor treating tumors anterior to the equator does not have a material impacton long-term peripheral vision if the effects of the treatment do not extendposterior to the equator and there is no damage to the anterior componentsof the visual axis.
Photocoagulation was used in the treatment of 18 tumors in this study.Six of these tumors were anterior to the equator and thus presented the problemdescribed earlier in relation to cryotherapy. Six tumors posterior to theequator were treated with photocoagulation alone. In 4 of these cases, thevisual field defect detected by our testing was larger than that predictedon the basis of fundus drawings of the tumor (Figure 4). This is most likely because retinoblastoma photocoagulationinvolves destruction of the blood vessels feeding the tumor. In many cases,areas of retina bordering these vessels are destroyed as well, and we thussee a scotoma larger than the tumor itself.
We have noted that the large majority of the tumors described producedvisual field defects that crossed either or both of the vertical or horizontalmeridians. This is not surprising when we consider that the defects are correlatedwith tumor locations, and we would not expect the tumors themselves to havea particular location preference with regard to the vertical or horizontalmeridians. This is important insofar as it impacts later ophthalmologic examinationof these patients and can permit ophthalmologists to distinguish defects relatedto retinoblastoma from those seen in more common conditions such as glaucoma.
Among the most surprising findings of this study was the small numberof arcuate visual defects we encountered. Although 15 tumors in 13 patientswere totally or partly in the region of the arcuate fiber bundle, only 4 producedthis classic defect.10 We cannot explain thisphenomenon, but it certainly raises important questions about the type ofdamage caused to retinal cells by tumors and subsequent therapy. It is importantto recognize that arcuate pathway defects in these patients should not alwaysbe attributed to retinoblastoma. In fact, as the findings in this study suggest,we do not expect most tumors in the posterior pole to produce such defects.If they are encountered, the treating ophthalmologist should explore the possibilityof causes other than retinoblastoma.
In the past 30 years, the management of retinoblastoma has changed dramatically.Increasing numbers of patients are surviving and retaining sight in at least1 eye. In this study, we have examined another important outcome measure inthis disease: long-term visual field results. It is our hope that the findingspresented herein will provide useful information for physicians as they assesstumors and choose among therapeutic modalities. We also hope that this informationserves the additional purpose of helping to predict and inform patients withretinoblastoma and their families about the long-term complications of theirdisease. Retinal imaging technology now enables us to precisely document tumorsize and appearance throughout patients' lives. Those now benefiting fromthese advances will soon be able to have their visual fields accurately tested,and these improvements should significantly enhance our ability to assessthe correlation between tumor or treatment damage and peripheral vision.
Correspondence: David H. Abramson, MD, Ophthalmology Oncology Service,Memorial Sloan-Kettering Cancer Center, 70 E 66th St, New York, NY 10021 (ICANCERMD@aol.com).
Submitted for publication June 17, 2002; final revision received February12, 2004; accepted April 6, 2004.