- Email: [email protected]
Radiation exposure
Radiation Exposure A New Risk Factor for Idiopathic Perifoveal Telangiectasis David A. L. Maberley, MD, MSc (Epid),1 Lawrence A. Yannuzzi, MD,1 Kurt Gitter, MD,2 Lawrence Singerman, MD,3 Emily Chew, MD,4 K. Bailey Freund, MD,1 Fabio Noguiera, MD,1 Danielle Sallas, MD,1 Ron Willson, MD,2 Kim Tillocco, COA3 Objective: To examine the association between previous radiation exposure and idiopathic perifoveal telangiectasis (IPT). Design: A multicentered, individually matched, case– control study design was used. Participants/Controls: Sixty-five case subjects were matched with 175 control subjects. Individuals with unequivocal evidence of angiographically confirmed IPT were included as cases. Control subjects were matched for center, age, and gender. Main Outcome Measure: The main exposures of interest were a history of therapeutic head or neck irradiation and environmental radiation exposure. Methods: A standardized questionnaire was administered to case and control subjects. Data were collected for the main exposures of interest as well as pertinent covariates. Conditional logistic regression was used to evaluate therapeutic and environmental radiation as risks for IPT. Results: On univariate analysis, head or neck irradiation was associated with IPT (odds ratios [OR] ⫽ 4.15, 95% confidence interval [CI] ⫽ 1.30 –13.24). While controlling for diabetes and family history of diabetes, IPT was found to be associated with both head or neck irradiation (OR ⫽ 4.06, 95% CI ⫽ 1.20 –13.76) and with environmental irradiation (OR ⫽ 6.73, 95% CI ⫽ 1.06 – 42.74). Conclusions: This study presents a previously unreported association between prior radiation exposure and IPT. Ophthalmology 1999;106:2248 –2253 Idiopathic perifoveal telangiectasis (IPT) is thought to be part of the spectrum of primary retinal telangiectasia. This disorder is typified by congenital or acquired perifoveal capillary dilation and leakage. Other features include rightangle vessels, refractile intraretinal deposits, retinal pigment epithelial hyperplasia, and choroidal neovascularization. Gass and coworkers1,2 have proposed a classification scheme that divides perifoveal telangiectasia into four cat-
Originally received: November 10, 1998. Revision accepted: August 16, 1999. Manuscript no. 98578. 1 Manhattan Eye, Ear and Throat Hospital, New York, New York. 2 East Jefferson Hospital, Touro Infirmary, New Orleans, Louisiana. 3 Mt. Sinai Hospital, Lakewood Hospital, Hillcrest Hospital, Cleveland, Ohio. 4 National Institutes of Health, National Eye Institute, Division of Biometry and Epidemiology, Washington, DC. David A. Maberley, MD, MSc (Epid), is currently at the Department of Ophthalmology, University of British Columbia, Vancouver, BC, Canada. Presented in part at the Retina Society annual meeting, Washington, DC, September 1998; and at the American Academy of Ophthalmology annual meeting, New Orleans, Louisiana, November 1998. None of the authors has any financial interests related in any way to this article. Address correspondence to David A. L. Maberley, MD, MSc (Epid), University of British Columbia/Vancouver General Hospital Eye Care Center, 2550 Willow St., Vancouver, BC, Canada V5Z 3N9 E-mail: [email protected].
2248
egories based on laterality, gender, age at onset, and lesion location. The main systemic association noted for IPT is with diabetes; to date, two case series have supported this link.3,4 Radiation therapy was used frequently in the early 1900s to examine and treat a variety of medical and nonmedical conditions. One of the first associations between radiation and delayed-onset disease was noted by Duffy and Fitzgerald, who, in 1950, linked thymus irradiation to thyroid cancer.5 Other associations between radiation exposure and cancer have been described, along with the treatment of acne vulgaris and tinea capitis, as well as other benign disorders of the head and neck.6,7 The rationale for this study arose out of the clinical suspicion of an association between previous radiation exposure and IPT. As far as the authors are aware, there is no published link between radiation exposure and IPT. This study considers the possible association between IPT and previous radiation exposure, including radiation to the head and neck and environmental radiation.
Materials and Methods All cases were identified from the hospitals and private clinics in Cleveland, New Orleans, and New York that are attended by the authors. A diagnosis of idiopathic perifoveal telangiectasis was made if there was unequivocal clinical and fluorescein angio-
Maberley et al 䡠 Radiation Exposure and Idiopathic Perifoveal Telangiectasis graphic evidence of IPT in one or both eyes. Cases had no clinical or historical evidence of branch vein occlusion, posterior inflammation, Coats disease, or radiation retinopathy. Individuals younger than 18 years of age were excluded. Patients were identified during 1997 and were all diagnosed with IPT during the 15-year interval between 1982 and 1997. The identification of potential control subjects was conducted using two sampling techniques. The method chosen depended on which was most feasible for each clinic. The first technique was used for the control subjects identified in New York and assigned a random two-digit number to each case. This number was used to select control subjects from numbered clinic file drawers. The first of the two random digits was used to identify the numbered file drawer and the second to select the distance from the front of the file from which to start looking for a control. The first two to three patients encountered whose age and gender matched those of the case and who had been seen in the same clinic in the past 3 years were contacted. Efforts were made to include three control subjects for each case; however, if by the end of each filing cabinet, only one control had been identified, the next cabinet was searched. In this circumstance, the recruitment of control subjects was terminated once two suitable control subjects were identified. In general, it was not difficult to collect two control subjects for each case. The second control selection process involved an examination of the age and gender of consecutive outpatients seen in the retina clinics in Cleveland and New Orleans. This selection protocol was performed by office assistants who recruited patients matching a prepared list of male and female ages that corresponded to cases of IPT from the same clinic. All non-IPT clinic patients whose age and gender corresponded to a case were asked to participate as control subjects once their ocular diagnosis was established. Three control subjects were sought for each case and efforts were made to mask interviewers from the ocular diagnoses of the subjects under evaluation. Institutional ethics review board approval was obtained, and informed consent was sought from all participants. In addition to individual gender and clinic matching, all control subjects were age-matched to within 5 years of the case subjects (most were within 2 years). Efforts were made to limit the number of control subjects with a single diagnosis to less than 20% of the total control population. A standardized questionnaire was administered to both case and control subjects by a trained research assistant. The specific associations of interest (radiation exposure and IPT) were not made known to the subjects during questioning. A number of potential confounders for IPT were also included in the questionnaire. The same interviewers administered all questionnaires to all patients in each center. Data were collected for the following risk factors: age, gender, medical history, medication history, presence/absence of diabetes mellitus, family history of diabetes mellitus in the absence of overt diabetes, history of therapeutic head or neck radiation, and history of environmental radiation exposure. For individuals with IPT, supplemental clinical information was recorded, including bestcorrected visual acuity at time of diagnosis, age at diagnosis, history of thyroid disease, and presence or absence of medically treated hypertension. “Head/neck radiation” exposure was documented as present if the patient had a history of therapeutic radiation for scalp infections, sinusitis, thyroid disease, thymus disease, acne, facial hair, scars, keloids, or cancer. “Environmental radiation” exposure included a history of work on a nuclear submarine, at a watch factory, at a uranium refinery, in a nuclear power plant, or at an industry using radioactive chemicals. “Any radiation” exposure was recorded dichotomously as the presence or absence of either of the preceding two radiation exposures. Diabetes was defined by
Table 1. Control Subject Diagnoses Diagnosis
No.
%
Macular degeneration Retinal detachment/tear/lattice Vitreous disorders Macular disorders Retinal vascular disease Diabetic retinopathy Anterior segment disorders Miscellaneous
45 28 8 17 15 36 10 16
25 16 5 10 9 20 6 9
175
100
Total
a self-reported history of Type-1 or Type-2 diabetes. There were no individuals with Type-1 diabetes enrolled in this study as either case or control subjects. A family history of diabetes was recorded as positive if a first-degree relative was identified by a nondiabetic respondent as having Type-1 or Type-2 diabetes. After computation of simple descriptive statistics, each risk factor was examined individually in a conditional logistic regression model (adjusted analysis) with IPT as the dependent variable. Odds ratios (OR) and 95% confidence intervals (CIs) were determined for each risk factor from the coefficients generated in the univariate process (adjusted analysis). All variables of interest were represented dichotomously and were modeled individually using a single term in the regression equation. Each form of radiation exposure was subsequently evaluated individually with diabetes and family history of diabetes in a multivariate conditional logistic regression model. The EGRET statistical software package (Statistics and Epidemiology Research Corp., Seattle, WA) was used for these calculations.
Results There were 65 cases of IPT identified at the 3 centers between 1982 and 1997. One hundred seventy-five control subjects were recruited. The distribution of control diagnoses is presented in Table 1. There was a slightly higher representation of control subjects with macular degeneration and diabetes than was intended (20% and 25%, respectively). Individuals with IPT were compared to control subjects for basic demographic features and for the main variables of interest. These results are presented in Table 2. A closer examination of the case subjects who did and did not have radiation exposure is presented in Table 3. Men with IPT were found to have been exposed to radiation more frequently than women (on the basis of occupational exposure). There were no differences between the case subjects with radiation exposure and those without; however, presenting visual acuity was actually slightly better than for case subjects who did not have a history of radiation exposure. Specific reasons for head and neck irradiation in the case group were as follows: three patients had irradiation for acne, two patients had irradiation for squamous cell carcinoma of the head/ neck, one patient had irradiation for a facial boil, and one patient had irradiation for a keloid scar. Environmental exposures in the cases included two nuclear reactor maintenance workers and one fluoroscopist. The average latency period between radiation exposure and the diagnosis of IPT was 42 years. For the control subjects, head/neck irradiation was performed on one individual with geographic tongue, one individual with a keloid scar, one individual with acne, one individual with squamous cell carcinoma, one individual with a choroidal malignant melanoma, and one individual with childhood irradiation for thy-
2249
Ophthalmology Volume 106, Number 12, December 1999 Table 2. Characteristics of Cases and Controls*
Center (Cleveland, NO, NY) Age (average) (yrs) Sex† Head/neck radiation Environmental radiation Environmental or head/neck radiation
IPT Cases (n ⴝ 65) [%]
Controls (n ⴝ 175) [%]
21,20,24 68 40 female [61.5%] 7 [10.8%] 3 [4.6%]
46,66,63 69 (matched) 106 female [60.5%] 6 [3.4%] 2 [1.1%]
8 [12.3%]
7 [4%]
IPT ⫽ idiopathic perifoveal telangiectasis. * Some individuals had more than one source of radiation exposure. † Matched on age and sex.
mus enlargement. Environmental exposure was noted in two persons, one who worked with radio-labeled chemicals and another who was involved in nuclear testing in Nevada in the 1940s. Unadjusted analysis of risk factors for IPT was examined using conditional logistic regression. These results are presented in Table 4. Associations with IPT were found for a history of head or neck irradiation and a family history of diabetes. When both sources of radiation were considered together, an association was also found with IPT. Adjusted analysis involved re-examining OR for radiation exposure sources after adjusting for diabetes and family history of diabetes. Associations with IPT were found for a history of head/ neck irradiation (OR ⫽ 4.06; 95% CI ⫽ 1.20 –13.76) and environmental radiation exposure (OR ⫽ 6.73; 95% CI ⫽ 1.06 – 42.74). As with the unadjusted analysis, when both sources of radiation exposure were combined, an association with IPT was noted (OR ⫽ 4.54; 95% CI ⫽ 1.4 –14.70). Interestingly, an association between diabetes and IPT was not found in the course of this study.
Discussion This study showed that therapeutic head or neck irradiation was an independent risk factor for IPT on both univariate and multivariate analyses. Environmental radiation exposure was associated on adjusted analysis alone. A history of any radiation exposure was associated with IPT on both adjusted and nonadjusted analysis. The association between IPT and radiation exposure presented in this study has not been documented previously in the literature and is an important finding. Initially, the suspicion of an association between IPT and radiation exposure arose from the clinical experience of the investigators (a number of patients with IPT gave a history of previous radiation exposure). This hypothesis was supported by the clinical and pathologic resemblance of IPT to radiation retinopathy. Histopathologically, the changes in IPT do not include retinal telangiectasia but rather evidence of damaged vascular endothelial cells with thickened vascular basement membranes.8 These changes are notably panretinal and not just perifoveal. Moreover, the lack of cystoid edema or extracellular fluid accumulation also suggests that the pathologic changes found in IPT may be more
2250
indicative of radiation retinopathy or early diabetic microangiopathy than Coats disease or Leber miliary aneurysms. As a result, there may be reason to suspect that similar etiologic mechanisms may be responsible for the retinal changes seen in IPT and radiation retinopathy. It is important to note that the OR derived in this study should approximate relative risk estimates for the associations of interest because the rare-disease assumption is met with IPT—the prevalence of IPT in the general population is less than 5%. Nonetheless, the uncovered association cannot be assumed to be definitive on the basis of this single case– control study. Further research is necessary to corroborate these results but will be limited by the aging of the cohort of individuals who were exposed to excess and unnecessary radiation in the mid-1900s. There were a few results in this study that are difficult to explain. One such contradictory finding was the discrepancy between the unadjusted and adjusted OR for the relationship between environmental radiation exposure and IPT. This difference in OR was re-examined “post-priori” by recalculating OR for environmental radiation while controlling for diabetes and a family history of diabetes in separate models. The OR (for environmental radiation) while controlling for diabetes was determined to be 5.25 (95% CI, 0.85–32.29), similar to the OR generated during the unadjusted analysis. This OR changed when family history of diabetes was controlled for (OR, 6.26; (95% CI, 1.01–38.92). These findings suggest that a family history of diabetes may confound the relationship between environmental radiation exposure and IPT in a negative manner, in which the significance of the association is masked unless a family history of diabetes is adjusted for. The mechanism underlying this finding is not clear. Of the other risk factors considered in this study, a family history of diabetes was found predictive of IPT, even after controlling for the presence of diabetes on multivariate analysis. This may support the findings of an earlier study that IPT may be more commonly associated with glucose intolerance.4 Interestingly, diabetes was not associated with IPT as noted in a previous study of this disease.3 The absence of this finding may have been related to the control group selection since individuals with diabetes make up a considerable component of the retinal practices participating in this study. The major methodologic concern with any case– control study is the selection of a control group. Two important criteria that this control group meets are that individuals in the control population would have been identified as cases had they developed the outcome of interest and that the control group comes from the same population as the cases.9 The control subjects were also individually matched to case subjects. Nonetheless, the question of whether this represents the best possible control group for the study is uncertain. First, the distribution of ocular diagnoses within the control group was that of a patient population with retinal disease. As a result, there were many patients with macular degeneration and diabetes compared to a community population. Moreover, the high percentage of control subjects with age-related macular degeneration (AMD) and diabetes could affect the results of this article if there is an associa-
Maberley et al 䡠 Radiation Exposure and Idiopathic Perifoveal Telangiectasis Table 3. Features of IPT Patients* Exposed to Radiation (n ⴝ 8)
Not Exposed to Radiation (n ⴝ 57)
Comparison (P)†
71.6 2 female (25%) 2 (25%) 3 (38%) (n ⫽ 7)
67.4 38 female (67%) 19 (35%)‡ 34 (61%)‡ (n ⫽ 35)
0.28 0.05 1.00 0.27
20/35 (log VA ⫽ 0.26) 1 (14%) 1 (14%) 4.9
20/50§ (log VA ⫽ 0.41) 16 (48%) 3 (9%) 6.6
0.14 0.21 0.53 0.41
Variable Mean age (yrs) Sex Diabetes Family history of diabetes Mean VA at presentation Hypertension Thyroid disease Average duration of IPT (yrs)
VA ⫽ visual acuity; IPT ⫽ idiopathic perifoveal telangiectasias. * Visual acuities were compared using the average LogMAR visual acuity for both eyes as the variable value. Only one patient had counting fingers vision, which was arbitrarily graded as 1.6 or 20/800. † Two-tailed Student’s t test was used for continuous variable comparisons. Fisher’s exact test was used for categorical variables. ‡ n ⫽ 56. § n ⫽ 34.
tion between macular degeneration or diabetes and radiation exposure. For example, if persons with previous head/neck irradiation are less likely to develop AMD, a strong spurious relationship between IPT and radiation could result from the protective effect of radiation on AMD. There is reason to consider the associations uncovered in this study valid. It is unlikely that differential recall bias (or selection bias) played a strong roll in the radiation/IPT relationship since the association of interest has not been described previously and, as such, patients with IPT are unlikely to recall a history of radiation exposure more frequently than patients without IPT but with another ocular condition. Nondifferential recall bias could exist for “diagnostic x-rays” because of the difficulty subjects could have remembering exposure of this nature in comparison to environmental exposure or therapeutic radiation. Such an effect would be expected to weaken the true strength of association. Moreover, the role of recall bias has been shown to be of minor concern in the design and analysis of case– control studies in which the exposure of interest is of low prevalence, as is the case with radiation in our case and control groups.10 In terms of selection bias, there is no evidence to suggest that individuals with IPT would have been more likely to have been referred for ocular examination at the three research centers if they had a history of radiation exposure. A possibility does exist that the patients with IPT who had a history of radiation exposure may have had more ad-
vanced ocular disease than those without radiation exposure and were referred to the participating clinics for this reason. An examination of patients for whom visual acuity data were available did not show a difference in visual function for cases with radiation exposure compared to those without. Despite the limitations of case– control study design, these results suggest that there is evidence to support a link between IPT and radiation exposure. Idiopathic perifoveal telangiectasis is likely a multifactorial condition with radiation exposure only one of a number of contributing conditions, many of which are yet to be determined. In this and other retinal vascular conditions, radiation may work in concert with other risk factors to create a toxic milieu that leads to the development of more advanced clinical pathology. Finally, the possibility of a causal link between radiation exposure and IPT will be considered. The specific criteria that give evidence for a causal relationship between an exposure and an outcome include strength of association, biologic credibility, consistency with other investigations, a dose-response relationship, and temporal plausibility.11 Using these criteria, causality is suggested by the strength of the relationship that was found between radiation and IPT as well as the strong statistical significance of this result on multivariate testing. Temporality issues are acceptable, since radiation was temporally removed from the development of IPT for all patients. The current study used a
Table 4. Unadjusted and Adjusted Associations with Idiopathic Perifoveal Telangiectasis (Conditional Logistic Regression) Variable
Unadjusted Odds Ratio (95% confidence interval)
Adjusted Odds Ratio (95% confidence interval)
Head/neck radiation Environmental radiation Environmental or head/neck radiation
4.15 (1.30to13.24) 5.22 (0.85to31.85) 4.53 (1.46to14.06)
4.06 (1.20to13.76) 6.73 (1.07to42.74) 4.54 (1.40to14.70)
2251
Ophthalmology Volume 106, Number 12, December 1999 questionnaire-based survey that did not allow for characterization of radiation dose for either case or control subjects, preventing an estimation of dose-response. (Interestingly, there were three individuals with IPT and radiation exposure who had multiple sources of radiation exposure.) Nonetheless, the validity of these findings will have to be confirmed by future research to strengthen the possibility of a causal relationship between radiation and IPT. In conclusion, this study supports a new association between idiopathic perifoveal telangiectasis and a history of radiation exposure. Indeed, for head and neck radiation exposure, specifically, there seems to be reason to suspect an increased risk for the development of IPT. Despite efforts in the past 40 years to protect people from unnecessary radiation exposure, there is still a large cohort of older individuals in our society who received high doses of radiation while in their youth and early adulthood. As a result, we may see a decline in the incidence of IPT as the number of people who have been exposed to excessive radiation falls over the next decades. This is, of course, assuming that new and uncontrolled sources of radiation exposure do not become manifest in our environment. A final implication of this research is its reminder to physicians about the possibility of iatrogenically induced disease that still exists whenever new technologies are introduced without a full understanding of their long-term sequelae.
References 1. Gass JD, Oyakawa RT. Idiopathic juxtafoveolar retinal telangiectasis. Arch Ophthalmol 1982;100:769 – 80. 2. Gass JDM, Blodi BA. Idiopathic juxtafoveolar retinal telangiectasis. Update of classification and follow-up study. Ophthalmology 1993;100:1536 – 46. 3. Chew EY, Murphy RP, Newsome DA, Fine SL. Parafoveal telangiectasis and diabetic retinopathy. Arch Ophthalmol 1986;104:71–5. 4. Millay RH, Klein ML, Handelman IL, Watzke RC. Abnormal glucose metabolism and parafoveal telangiectasia. Am J Ophthalmol 1986;102:363–70. 5. Duffy BJ Jr, Fitzgerald PJ. Thyroid cancer in childhood and adolescence: a report on twenty-eight cases. Cancer 1950;3: 1018 –32. 6. Thomson DB, Grammes CF, Starkey RH, et al. Thyroid abnormalities in patients previously treated with irradiation for acne vulgaris. South Med J 1984;77:21–3. 7. Modan B, Ron E, Werner A. Thyroid cancer following scalp irradiation. Radiology 1977;123:741– 4. 8. Green WR, Quigley HA, de la Cruz Z, Cohen B. Parafoveal retinal telangiectasis: light and electron microscopy studies. Trans Ophthalmol Soc UK 1980;100:162–70. 9. Rothman KJ, Greenland S. Modern Epidemiology, 2nd ed. Philadelphia: Lippincott–Raven, 1998;96 – 8. 10. Drews CD, Greeland S. The impact of differential recall on the results of case-control studies. Int J Epidemiol 1990;19:1107–12. 11. Hennekens CH, Buring JE. Epidemiology in Medicine. Toronto: Little Brown and Company, 1987;39 – 43.
Discussion by David A. Newsome, MD The authors have presented novel data associating idiopathic perifoveal telangiectasis and a history of radiation exposure. This retrospective case– control study used chart-review data from three centers. Control selection methods differed from center to center, but case selection methods appeared uniform. The data show a strong correlation between the diagnosis of idiopathic perifoveal telangiectasis (IPT), a part of the spectrum of idiopathic retinal telangiectasis, and any history of radiation exposure, especially radiation exposure to the head and neck. When head and neck irradiation and a history of environmental radiation exposure are combined, the associative effect appeared even stronger. Because IPT is uncommon and radiation exposures of the types the authors found are also uncommon, the discovery of an association in only 65 cases is intriguing. Retrospective studies have numerous well-known pitfalls. This study has not avoided all of them, although the authors, to their credit, do address many of these in a straight-forward fashion. One of the major drawbacks to the study is the risk factor collection, which omitted important potential causes of microvascular change, such as high blood pressure, for the control subjects. It is interesting that the authors found no association between the presence of diabetes mellitus and IPT.1,2 The presumed remote nature of the episodes of radiation exposure can only be inferred since data describing elapsed time between radiation exposure and diagnosis of IPT are not provided. These data would be of some interest since the authors imply that
From Touro Infirmary, Tulane Medical Center, New Orleans, Louisiana.
2252
the initial radiation–retina interaction was insufficient to produce severe clinically symptomatic radiation retinopathy and that the appearance of IPT is a late sequela of the radiation exposure. Other concerns with the study include selective and nonselective recall bias, which, considering the probable remoteness of the radiation event and the average age of the case and control subjects (68 and 69 years, respectively), is likely to be significant. The calculated association between a history of head and neck radiation exposure and environmental radiation exposure in particular and IPT appears strong in the authors’ data. The exact mechanism for such a vascular abnormality remains unclear. It is known from studies of acute radiation retinopathy, for example patients who have had orbital irradiation of an unusually large intensity for Graves disease,3 that ischemic and hemorrhagic retinal microvascular changes are hallmarks (Table 1). These ischemic changes can activate potent growth factors and peptides such as vascular endothelial growth factor. These powerful chemicals not only produce direct damage to vascular endothelium but also stimulate angiogenic changes
Table 1. Classic Radiation Retinopathy Cotton wool spots Nerve fiber layer infarcts Microaneurysms Telangiectasias Retinal hemorrhages Retinal edema, exudation
Maberley et al 䡠 Radiation Exposure and Idiopathic Perifoveal Telangiectasis that could result in the formation of telangiectasia and other microvasculopathies.4 The authors’ work should stimulate confirmatory clinical work among other future investigations. The importance of radiation retinopathy as a clinical problem, although not great, is likely to continue since all environmental sources of radiation have not been controlled. Medical irradiation can also continue to be a causative factor. The development of a useful animal model for mild chronic radiation retinopathy and its sequela could lead to an enhanced understanding of angiogenesis and vasculogenesis, important not only in a variety of retinal diseases but also in tumors. The authors are to be commended for a novel study that reminds us of the potential deleterious long-term effect of naturally occurring and iatrogenically applied radiation.
References 1. Chew EY, Murphy RP, Newsome DA, Fine SL. Parafoveal telangiectasias and diabetic retinopathy. Arch Ophthalmol 1986;104:71–5. 2. Millay RH, Klein ML, Handelman IL, Watzke RC. Abnormal glucose metabolism and parafoveal telangiectasia. Am J Ophthalmol 1986;102:363–70. 3. Kinyoun JL, Kalina RE, Brower SA, et al. Radiation retinopathy after orbital irradiation for Graves’ ophthalmopathy. Arch Ophthalmol 1984;102:1473– 6. 4. Cao Y, Linden P, Shima D, et al. In vivo angiogenic activity and hypoxia induction of heterodimers of placenta growth factor/vascular endothelial growth factor. J Clin Invest 1996; 98:2507–11.
Fellowships Available The Heed Ophthalmic Foundation provides fellowship support for advanced study or research in ophthalmology. Applicants must be graduates of medical schools accredited by the American Medical Association and citizens of the United States. Deadline for submission of applications: January 15th for fellowship starting in July. Please direct all inquiries and requests for application forms to: Froncie A. Gutman, M.D., Secretary The Heed Ophthalmic Foundation Cleveland Clinic Foundation 9500 Euclid Avenue, A31 Cleveland, OH 44195
2253
Originally received: November 10, 1998. Revision accepted: August 16, 1999. Manuscript no. 98578. 1 Manhattan Eye, Ear and Throat Hospital, New York, New York. 2 East Jefferson Hospital, Touro Infirmary, New Orleans, Louisiana. 3 Mt. Sinai Hospital, Lakewood Hospital, Hillcrest Hospital, Cleveland, Ohio. 4 National Institutes of Health, National Eye Institute, Division of Biometry and Epidemiology, Washington, DC. David A. Maberley, MD, MSc (Epid), is currently at the Department of Ophthalmology, University of British Columbia, Vancouver, BC, Canada. Presented in part at the Retina Society annual meeting, Washington, DC, September 1998; and at the American Academy of Ophthalmology annual meeting, New Orleans, Louisiana, November 1998. None of the authors has any financial interests related in any way to this article. Address correspondence to David A. L. Maberley, MD, MSc (Epid), University of British Columbia/Vancouver General Hospital Eye Care Center, 2550 Willow St., Vancouver, BC, Canada V5Z 3N9 E-mail: [email protected].
2248
egories based on laterality, gender, age at onset, and lesion location. The main systemic association noted for IPT is with diabetes; to date, two case series have supported this link.3,4 Radiation therapy was used frequently in the early 1900s to examine and treat a variety of medical and nonmedical conditions. One of the first associations between radiation and delayed-onset disease was noted by Duffy and Fitzgerald, who, in 1950, linked thymus irradiation to thyroid cancer.5 Other associations between radiation exposure and cancer have been described, along with the treatment of acne vulgaris and tinea capitis, as well as other benign disorders of the head and neck.6,7 The rationale for this study arose out of the clinical suspicion of an association between previous radiation exposure and IPT. As far as the authors are aware, there is no published link between radiation exposure and IPT. This study considers the possible association between IPT and previous radiation exposure, including radiation to the head and neck and environmental radiation.
Materials and Methods All cases were identified from the hospitals and private clinics in Cleveland, New Orleans, and New York that are attended by the authors. A diagnosis of idiopathic perifoveal telangiectasis was made if there was unequivocal clinical and fluorescein angio-
Maberley et al 䡠 Radiation Exposure and Idiopathic Perifoveal Telangiectasis graphic evidence of IPT in one or both eyes. Cases had no clinical or historical evidence of branch vein occlusion, posterior inflammation, Coats disease, or radiation retinopathy. Individuals younger than 18 years of age were excluded. Patients were identified during 1997 and were all diagnosed with IPT during the 15-year interval between 1982 and 1997. The identification of potential control subjects was conducted using two sampling techniques. The method chosen depended on which was most feasible for each clinic. The first technique was used for the control subjects identified in New York and assigned a random two-digit number to each case. This number was used to select control subjects from numbered clinic file drawers. The first of the two random digits was used to identify the numbered file drawer and the second to select the distance from the front of the file from which to start looking for a control. The first two to three patients encountered whose age and gender matched those of the case and who had been seen in the same clinic in the past 3 years were contacted. Efforts were made to include three control subjects for each case; however, if by the end of each filing cabinet, only one control had been identified, the next cabinet was searched. In this circumstance, the recruitment of control subjects was terminated once two suitable control subjects were identified. In general, it was not difficult to collect two control subjects for each case. The second control selection process involved an examination of the age and gender of consecutive outpatients seen in the retina clinics in Cleveland and New Orleans. This selection protocol was performed by office assistants who recruited patients matching a prepared list of male and female ages that corresponded to cases of IPT from the same clinic. All non-IPT clinic patients whose age and gender corresponded to a case were asked to participate as control subjects once their ocular diagnosis was established. Three control subjects were sought for each case and efforts were made to mask interviewers from the ocular diagnoses of the subjects under evaluation. Institutional ethics review board approval was obtained, and informed consent was sought from all participants. In addition to individual gender and clinic matching, all control subjects were age-matched to within 5 years of the case subjects (most were within 2 years). Efforts were made to limit the number of control subjects with a single diagnosis to less than 20% of the total control population. A standardized questionnaire was administered to both case and control subjects by a trained research assistant. The specific associations of interest (radiation exposure and IPT) were not made known to the subjects during questioning. A number of potential confounders for IPT were also included in the questionnaire. The same interviewers administered all questionnaires to all patients in each center. Data were collected for the following risk factors: age, gender, medical history, medication history, presence/absence of diabetes mellitus, family history of diabetes mellitus in the absence of overt diabetes, history of therapeutic head or neck radiation, and history of environmental radiation exposure. For individuals with IPT, supplemental clinical information was recorded, including bestcorrected visual acuity at time of diagnosis, age at diagnosis, history of thyroid disease, and presence or absence of medically treated hypertension. “Head/neck radiation” exposure was documented as present if the patient had a history of therapeutic radiation for scalp infections, sinusitis, thyroid disease, thymus disease, acne, facial hair, scars, keloids, or cancer. “Environmental radiation” exposure included a history of work on a nuclear submarine, at a watch factory, at a uranium refinery, in a nuclear power plant, or at an industry using radioactive chemicals. “Any radiation” exposure was recorded dichotomously as the presence or absence of either of the preceding two radiation exposures. Diabetes was defined by
Table 1. Control Subject Diagnoses Diagnosis
No.
%
Macular degeneration Retinal detachment/tear/lattice Vitreous disorders Macular disorders Retinal vascular disease Diabetic retinopathy Anterior segment disorders Miscellaneous
45 28 8 17 15 36 10 16
25 16 5 10 9 20 6 9
175
100
Total
a self-reported history of Type-1 or Type-2 diabetes. There were no individuals with Type-1 diabetes enrolled in this study as either case or control subjects. A family history of diabetes was recorded as positive if a first-degree relative was identified by a nondiabetic respondent as having Type-1 or Type-2 diabetes. After computation of simple descriptive statistics, each risk factor was examined individually in a conditional logistic regression model (adjusted analysis) with IPT as the dependent variable. Odds ratios (OR) and 95% confidence intervals (CIs) were determined for each risk factor from the coefficients generated in the univariate process (adjusted analysis). All variables of interest were represented dichotomously and were modeled individually using a single term in the regression equation. Each form of radiation exposure was subsequently evaluated individually with diabetes and family history of diabetes in a multivariate conditional logistic regression model. The EGRET statistical software package (Statistics and Epidemiology Research Corp., Seattle, WA) was used for these calculations.
Results There were 65 cases of IPT identified at the 3 centers between 1982 and 1997. One hundred seventy-five control subjects were recruited. The distribution of control diagnoses is presented in Table 1. There was a slightly higher representation of control subjects with macular degeneration and diabetes than was intended (20% and 25%, respectively). Individuals with IPT were compared to control subjects for basic demographic features and for the main variables of interest. These results are presented in Table 2. A closer examination of the case subjects who did and did not have radiation exposure is presented in Table 3. Men with IPT were found to have been exposed to radiation more frequently than women (on the basis of occupational exposure). There were no differences between the case subjects with radiation exposure and those without; however, presenting visual acuity was actually slightly better than for case subjects who did not have a history of radiation exposure. Specific reasons for head and neck irradiation in the case group were as follows: three patients had irradiation for acne, two patients had irradiation for squamous cell carcinoma of the head/ neck, one patient had irradiation for a facial boil, and one patient had irradiation for a keloid scar. Environmental exposures in the cases included two nuclear reactor maintenance workers and one fluoroscopist. The average latency period between radiation exposure and the diagnosis of IPT was 42 years. For the control subjects, head/neck irradiation was performed on one individual with geographic tongue, one individual with a keloid scar, one individual with acne, one individual with squamous cell carcinoma, one individual with a choroidal malignant melanoma, and one individual with childhood irradiation for thy-
2249
Ophthalmology Volume 106, Number 12, December 1999 Table 2. Characteristics of Cases and Controls*
Center (Cleveland, NO, NY) Age (average) (yrs) Sex† Head/neck radiation Environmental radiation Environmental or head/neck radiation
IPT Cases (n ⴝ 65) [%]
Controls (n ⴝ 175) [%]
21,20,24 68 40 female [61.5%] 7 [10.8%] 3 [4.6%]
46,66,63 69 (matched) 106 female [60.5%] 6 [3.4%] 2 [1.1%]
8 [12.3%]
7 [4%]
IPT ⫽ idiopathic perifoveal telangiectasis. * Some individuals had more than one source of radiation exposure. † Matched on age and sex.
mus enlargement. Environmental exposure was noted in two persons, one who worked with radio-labeled chemicals and another who was involved in nuclear testing in Nevada in the 1940s. Unadjusted analysis of risk factors for IPT was examined using conditional logistic regression. These results are presented in Table 4. Associations with IPT were found for a history of head or neck irradiation and a family history of diabetes. When both sources of radiation were considered together, an association was also found with IPT. Adjusted analysis involved re-examining OR for radiation exposure sources after adjusting for diabetes and family history of diabetes. Associations with IPT were found for a history of head/ neck irradiation (OR ⫽ 4.06; 95% CI ⫽ 1.20 –13.76) and environmental radiation exposure (OR ⫽ 6.73; 95% CI ⫽ 1.06 – 42.74). As with the unadjusted analysis, when both sources of radiation exposure were combined, an association with IPT was noted (OR ⫽ 4.54; 95% CI ⫽ 1.4 –14.70). Interestingly, an association between diabetes and IPT was not found in the course of this study.
Discussion This study showed that therapeutic head or neck irradiation was an independent risk factor for IPT on both univariate and multivariate analyses. Environmental radiation exposure was associated on adjusted analysis alone. A history of any radiation exposure was associated with IPT on both adjusted and nonadjusted analysis. The association between IPT and radiation exposure presented in this study has not been documented previously in the literature and is an important finding. Initially, the suspicion of an association between IPT and radiation exposure arose from the clinical experience of the investigators (a number of patients with IPT gave a history of previous radiation exposure). This hypothesis was supported by the clinical and pathologic resemblance of IPT to radiation retinopathy. Histopathologically, the changes in IPT do not include retinal telangiectasia but rather evidence of damaged vascular endothelial cells with thickened vascular basement membranes.8 These changes are notably panretinal and not just perifoveal. Moreover, the lack of cystoid edema or extracellular fluid accumulation also suggests that the pathologic changes found in IPT may be more
2250
indicative of radiation retinopathy or early diabetic microangiopathy than Coats disease or Leber miliary aneurysms. As a result, there may be reason to suspect that similar etiologic mechanisms may be responsible for the retinal changes seen in IPT and radiation retinopathy. It is important to note that the OR derived in this study should approximate relative risk estimates for the associations of interest because the rare-disease assumption is met with IPT—the prevalence of IPT in the general population is less than 5%. Nonetheless, the uncovered association cannot be assumed to be definitive on the basis of this single case– control study. Further research is necessary to corroborate these results but will be limited by the aging of the cohort of individuals who were exposed to excess and unnecessary radiation in the mid-1900s. There were a few results in this study that are difficult to explain. One such contradictory finding was the discrepancy between the unadjusted and adjusted OR for the relationship between environmental radiation exposure and IPT. This difference in OR was re-examined “post-priori” by recalculating OR for environmental radiation while controlling for diabetes and a family history of diabetes in separate models. The OR (for environmental radiation) while controlling for diabetes was determined to be 5.25 (95% CI, 0.85–32.29), similar to the OR generated during the unadjusted analysis. This OR changed when family history of diabetes was controlled for (OR, 6.26; (95% CI, 1.01–38.92). These findings suggest that a family history of diabetes may confound the relationship between environmental radiation exposure and IPT in a negative manner, in which the significance of the association is masked unless a family history of diabetes is adjusted for. The mechanism underlying this finding is not clear. Of the other risk factors considered in this study, a family history of diabetes was found predictive of IPT, even after controlling for the presence of diabetes on multivariate analysis. This may support the findings of an earlier study that IPT may be more commonly associated with glucose intolerance.4 Interestingly, diabetes was not associated with IPT as noted in a previous study of this disease.3 The absence of this finding may have been related to the control group selection since individuals with diabetes make up a considerable component of the retinal practices participating in this study. The major methodologic concern with any case– control study is the selection of a control group. Two important criteria that this control group meets are that individuals in the control population would have been identified as cases had they developed the outcome of interest and that the control group comes from the same population as the cases.9 The control subjects were also individually matched to case subjects. Nonetheless, the question of whether this represents the best possible control group for the study is uncertain. First, the distribution of ocular diagnoses within the control group was that of a patient population with retinal disease. As a result, there were many patients with macular degeneration and diabetes compared to a community population. Moreover, the high percentage of control subjects with age-related macular degeneration (AMD) and diabetes could affect the results of this article if there is an associa-
Maberley et al 䡠 Radiation Exposure and Idiopathic Perifoveal Telangiectasis Table 3. Features of IPT Patients* Exposed to Radiation (n ⴝ 8)
Not Exposed to Radiation (n ⴝ 57)
Comparison (P)†
71.6 2 female (25%) 2 (25%) 3 (38%) (n ⫽ 7)
67.4 38 female (67%) 19 (35%)‡ 34 (61%)‡ (n ⫽ 35)
0.28 0.05 1.00 0.27
20/35 (log VA ⫽ 0.26) 1 (14%) 1 (14%) 4.9
20/50§ (log VA ⫽ 0.41) 16 (48%) 3 (9%) 6.6
0.14 0.21 0.53 0.41
Variable Mean age (yrs) Sex Diabetes Family history of diabetes Mean VA at presentation Hypertension Thyroid disease Average duration of IPT (yrs)
VA ⫽ visual acuity; IPT ⫽ idiopathic perifoveal telangiectasias. * Visual acuities were compared using the average LogMAR visual acuity for both eyes as the variable value. Only one patient had counting fingers vision, which was arbitrarily graded as 1.6 or 20/800. † Two-tailed Student’s t test was used for continuous variable comparisons. Fisher’s exact test was used for categorical variables. ‡ n ⫽ 56. § n ⫽ 34.
tion between macular degeneration or diabetes and radiation exposure. For example, if persons with previous head/neck irradiation are less likely to develop AMD, a strong spurious relationship between IPT and radiation could result from the protective effect of radiation on AMD. There is reason to consider the associations uncovered in this study valid. It is unlikely that differential recall bias (or selection bias) played a strong roll in the radiation/IPT relationship since the association of interest has not been described previously and, as such, patients with IPT are unlikely to recall a history of radiation exposure more frequently than patients without IPT but with another ocular condition. Nondifferential recall bias could exist for “diagnostic x-rays” because of the difficulty subjects could have remembering exposure of this nature in comparison to environmental exposure or therapeutic radiation. Such an effect would be expected to weaken the true strength of association. Moreover, the role of recall bias has been shown to be of minor concern in the design and analysis of case– control studies in which the exposure of interest is of low prevalence, as is the case with radiation in our case and control groups.10 In terms of selection bias, there is no evidence to suggest that individuals with IPT would have been more likely to have been referred for ocular examination at the three research centers if they had a history of radiation exposure. A possibility does exist that the patients with IPT who had a history of radiation exposure may have had more ad-
vanced ocular disease than those without radiation exposure and were referred to the participating clinics for this reason. An examination of patients for whom visual acuity data were available did not show a difference in visual function for cases with radiation exposure compared to those without. Despite the limitations of case– control study design, these results suggest that there is evidence to support a link between IPT and radiation exposure. Idiopathic perifoveal telangiectasis is likely a multifactorial condition with radiation exposure only one of a number of contributing conditions, many of which are yet to be determined. In this and other retinal vascular conditions, radiation may work in concert with other risk factors to create a toxic milieu that leads to the development of more advanced clinical pathology. Finally, the possibility of a causal link between radiation exposure and IPT will be considered. The specific criteria that give evidence for a causal relationship between an exposure and an outcome include strength of association, biologic credibility, consistency with other investigations, a dose-response relationship, and temporal plausibility.11 Using these criteria, causality is suggested by the strength of the relationship that was found between radiation and IPT as well as the strong statistical significance of this result on multivariate testing. Temporality issues are acceptable, since radiation was temporally removed from the development of IPT for all patients. The current study used a
Table 4. Unadjusted and Adjusted Associations with Idiopathic Perifoveal Telangiectasis (Conditional Logistic Regression) Variable
Unadjusted Odds Ratio (95% confidence interval)
Adjusted Odds Ratio (95% confidence interval)
Head/neck radiation Environmental radiation Environmental or head/neck radiation
4.15 (1.30to13.24) 5.22 (0.85to31.85) 4.53 (1.46to14.06)
4.06 (1.20to13.76) 6.73 (1.07to42.74) 4.54 (1.40to14.70)
2251
Ophthalmology Volume 106, Number 12, December 1999 questionnaire-based survey that did not allow for characterization of radiation dose for either case or control subjects, preventing an estimation of dose-response. (Interestingly, there were three individuals with IPT and radiation exposure who had multiple sources of radiation exposure.) Nonetheless, the validity of these findings will have to be confirmed by future research to strengthen the possibility of a causal relationship between radiation and IPT. In conclusion, this study supports a new association between idiopathic perifoveal telangiectasis and a history of radiation exposure. Indeed, for head and neck radiation exposure, specifically, there seems to be reason to suspect an increased risk for the development of IPT. Despite efforts in the past 40 years to protect people from unnecessary radiation exposure, there is still a large cohort of older individuals in our society who received high doses of radiation while in their youth and early adulthood. As a result, we may see a decline in the incidence of IPT as the number of people who have been exposed to excessive radiation falls over the next decades. This is, of course, assuming that new and uncontrolled sources of radiation exposure do not become manifest in our environment. A final implication of this research is its reminder to physicians about the possibility of iatrogenically induced disease that still exists whenever new technologies are introduced without a full understanding of their long-term sequelae.
References 1. Gass JD, Oyakawa RT. Idiopathic juxtafoveolar retinal telangiectasis. Arch Ophthalmol 1982;100:769 – 80. 2. Gass JDM, Blodi BA. Idiopathic juxtafoveolar retinal telangiectasis. Update of classification and follow-up study. Ophthalmology 1993;100:1536 – 46. 3. Chew EY, Murphy RP, Newsome DA, Fine SL. Parafoveal telangiectasis and diabetic retinopathy. Arch Ophthalmol 1986;104:71–5. 4. Millay RH, Klein ML, Handelman IL, Watzke RC. Abnormal glucose metabolism and parafoveal telangiectasia. Am J Ophthalmol 1986;102:363–70. 5. Duffy BJ Jr, Fitzgerald PJ. Thyroid cancer in childhood and adolescence: a report on twenty-eight cases. Cancer 1950;3: 1018 –32. 6. Thomson DB, Grammes CF, Starkey RH, et al. Thyroid abnormalities in patients previously treated with irradiation for acne vulgaris. South Med J 1984;77:21–3. 7. Modan B, Ron E, Werner A. Thyroid cancer following scalp irradiation. Radiology 1977;123:741– 4. 8. Green WR, Quigley HA, de la Cruz Z, Cohen B. Parafoveal retinal telangiectasis: light and electron microscopy studies. Trans Ophthalmol Soc UK 1980;100:162–70. 9. Rothman KJ, Greenland S. Modern Epidemiology, 2nd ed. Philadelphia: Lippincott–Raven, 1998;96 – 8. 10. Drews CD, Greeland S. The impact of differential recall on the results of case-control studies. Int J Epidemiol 1990;19:1107–12. 11. Hennekens CH, Buring JE. Epidemiology in Medicine. Toronto: Little Brown and Company, 1987;39 – 43.
Discussion by David A. Newsome, MD The authors have presented novel data associating idiopathic perifoveal telangiectasis and a history of radiation exposure. This retrospective case– control study used chart-review data from three centers. Control selection methods differed from center to center, but case selection methods appeared uniform. The data show a strong correlation between the diagnosis of idiopathic perifoveal telangiectasis (IPT), a part of the spectrum of idiopathic retinal telangiectasis, and any history of radiation exposure, especially radiation exposure to the head and neck. When head and neck irradiation and a history of environmental radiation exposure are combined, the associative effect appeared even stronger. Because IPT is uncommon and radiation exposures of the types the authors found are also uncommon, the discovery of an association in only 65 cases is intriguing. Retrospective studies have numerous well-known pitfalls. This study has not avoided all of them, although the authors, to their credit, do address many of these in a straight-forward fashion. One of the major drawbacks to the study is the risk factor collection, which omitted important potential causes of microvascular change, such as high blood pressure, for the control subjects. It is interesting that the authors found no association between the presence of diabetes mellitus and IPT.1,2 The presumed remote nature of the episodes of radiation exposure can only be inferred since data describing elapsed time between radiation exposure and diagnosis of IPT are not provided. These data would be of some interest since the authors imply that
From Touro Infirmary, Tulane Medical Center, New Orleans, Louisiana.
2252
the initial radiation–retina interaction was insufficient to produce severe clinically symptomatic radiation retinopathy and that the appearance of IPT is a late sequela of the radiation exposure. Other concerns with the study include selective and nonselective recall bias, which, considering the probable remoteness of the radiation event and the average age of the case and control subjects (68 and 69 years, respectively), is likely to be significant. The calculated association between a history of head and neck radiation exposure and environmental radiation exposure in particular and IPT appears strong in the authors’ data. The exact mechanism for such a vascular abnormality remains unclear. It is known from studies of acute radiation retinopathy, for example patients who have had orbital irradiation of an unusually large intensity for Graves disease,3 that ischemic and hemorrhagic retinal microvascular changes are hallmarks (Table 1). These ischemic changes can activate potent growth factors and peptides such as vascular endothelial growth factor. These powerful chemicals not only produce direct damage to vascular endothelium but also stimulate angiogenic changes
Table 1. Classic Radiation Retinopathy Cotton wool spots Nerve fiber layer infarcts Microaneurysms Telangiectasias Retinal hemorrhages Retinal edema, exudation
Maberley et al 䡠 Radiation Exposure and Idiopathic Perifoveal Telangiectasis that could result in the formation of telangiectasia and other microvasculopathies.4 The authors’ work should stimulate confirmatory clinical work among other future investigations. The importance of radiation retinopathy as a clinical problem, although not great, is likely to continue since all environmental sources of radiation have not been controlled. Medical irradiation can also continue to be a causative factor. The development of a useful animal model for mild chronic radiation retinopathy and its sequela could lead to an enhanced understanding of angiogenesis and vasculogenesis, important not only in a variety of retinal diseases but also in tumors. The authors are to be commended for a novel study that reminds us of the potential deleterious long-term effect of naturally occurring and iatrogenically applied radiation.
References 1. Chew EY, Murphy RP, Newsome DA, Fine SL. Parafoveal telangiectasias and diabetic retinopathy. Arch Ophthalmol 1986;104:71–5. 2. Millay RH, Klein ML, Handelman IL, Watzke RC. Abnormal glucose metabolism and parafoveal telangiectasia. Am J Ophthalmol 1986;102:363–70. 3. Kinyoun JL, Kalina RE, Brower SA, et al. Radiation retinopathy after orbital irradiation for Graves’ ophthalmopathy. Arch Ophthalmol 1984;102:1473– 6. 4. Cao Y, Linden P, Shima D, et al. In vivo angiogenic activity and hypoxia induction of heterodimers of placenta growth factor/vascular endothelial growth factor. J Clin Invest 1996; 98:2507–11.
Fellowships Available The Heed Ophthalmic Foundation provides fellowship support for advanced study or research in ophthalmology. Applicants must be graduates of medical schools accredited by the American Medical Association and citizens of the United States. Deadline for submission of applications: January 15th for fellowship starting in July. Please direct all inquiries and requests for application forms to: Froncie A. Gutman, M.D., Secretary The Heed Ophthalmic Foundation Cleveland Clinic Foundation 9500 Euclid Avenue, A31 Cleveland, OH 44195
2253