Asteroid Hyalosis in an Autopsy Population the University of California at Los Angeles (UCLA) Experience

CLINICAL SCIENCES Asteroid Hyalosis in an Autopsy Population The University of California at Los Angeles (UCLA) Experience

Amani A. Fawzi, MD; Baotran Vo, BS; Ryan Kriwanek; Hema L. Ramkumar, BS; Chris Cha, BS; Amy Carts; John R. Heckenlively, MD; Robert Y. Foos, MD, DVM†; Ben J. Glasgow, MD

Objectives: To study the prevalence and associations ous attachment (PϽ.001). After adjusting for age in a mul- of asteroid hyalosis (AH) in a series of autopsy eyes. tivariate logistic regression analysis, statistical significance was found only for posterior vitreous attachment Methods: Retrospective review of the University of (P=.002) and male sex (P=.046). No statistically signifi- California at Los Angeles (UCLA) autopsy eye database cant association was found with diabetes mellitus or al- from 1965 to 2000 yielded 10801 patients. The pa- cohol abuse by univariate or multivariate analysis. Analy- tients’ medical histories were reviewed for evidence of sis of the odds ratio showed a strong age effect that diabetes mellitus, hypertension, hyperlipidemia, alco- increased from 5.0 (95% confidence interval, 2.2-11.3) hol abuse, hypercalcemia, hypothyroidism, and chronic in age group 41 to 50 years, compared with 25.4 (95% renal failure. Autopsy records were searched for evi- Wald confidence interval, 8.2-77.9) in the age group of dence of optic atrophy, macular degeneration, posterior patients older than 90 years. vitreous detachment, atherosclerosis, and chronic renal failure. Asteroid hyalosis was diagnosed by examina- Conclusions: A unique epidemiological autopsy co- tion of the autopsy eyes. Univariate and multivariate hort study of AH and its systemic associations yielded a statistical methods were used to analyze our data. higher prevalence of AH than previous studies. Asteroid hyalosis was strongly correlated with age and inversely Results: The prevalence of AH was 1.96% in this au- correlated with posterior vitreous detachment. Unlike topsy population. By ␹2 analysis, AH was significantly cor- some previous reports, we found no statistically signifi- related with age (PϽ.001), male sex (P=.006), age- cant correlation between AH and diabetes mellitus. related macular degeneration (P=.02), hypertension (P=.03), atherosclerosis (PϽ.001), and posterior vitre- Arch Ophthalmol. 2005;123:486-490

STEROID HYALOSIS (AH), were reported in 11 case series and popu- also referred to as aster- lation-based reports in the literature oid hyalitis, asteroid bod- (Table 1)2,3,7,9-16 as well as 1 series of AH ies, or scintillatio nivea, in enucleated eyes.17 One study was popu- was first described by Ben- lation based3; all other reports originated son in 1894.1 The prevalence of AH has from eye clinic populations, which may re- A 2 been reported to range from 0.83% to flect selection bias for diseases that are 1.2%.3 While AH is generally considered more prevalent among outpatients in an Author Affiliations: to have only a minor impact on visual acu- eye clinic, and cannot be extrapolated to Department of Ophthalmology, ity, some patients are disturbed enough by the general population. Statistical meth- Jules Stein Eye Institute their visual symptoms to undergo surgi- ods and patient and control selection cri- (Drs Fawzi, Heckenlively, and cal treatment.4,5 The condition is often uni- teria used in these studies were difficult Glasgow; Messrs Kriwanek and lateral, yet the etiology of AH is not known. to assess and highly variable. Healthy, age- Cha; and Mss Vo, Ramkumar, Knowledge of the conditions predispos- matched controls were used in 1 study,12 and Carts), Department of ing to AH may provide clues for deter- whereas age-matched controls from the Pathology and Laboratory mining the mechanism of the formation same clinic population were used in other Medicine (Drs Foos and of asteroid bodies. Numerous studies studies.2,11,13,16 Several works did not at- Glasgow), The David Geffen have claimed an association between AH tempt to compare patients with AH with School of Medicine at 2,6-12 7,9,10,14,15,17 University of California at and diabetes mellitus (DM). Other controls. Because of small sample Los Angeles. reported associations with AH include sizes, only univariate statistical analyses 7,8 2,13 Financial Disclosure: None. hypercholesterolemia, hypertension, were used in the case-control stud- †Dr Foos died in November increased serum calcium levels,9 and ies.2,11-13,16 Only 1 report collected a large 2004. hyperopia.13 Overall, 547 cases of AH enough sample size to use multivariate sta-

(REPRINTED) ARCH OPHTHALMOL / VOL 123, APR 2005 WWW.ARCHOPHTHALMOL.COM 486

Asteroid Hyalosis Population Prevalence, %, Bilateral Male-Female Patients With Source Source Selection (Sample Size) Cases, % Ratio Diabetes Mellitus, % Bergren et al2 Eye clinic Cross-sectional study with 0.83 (12 205) 19 0.94:1 29 case-control subgroup Moss et al3 Population Cross-sectional; population 1.2 (4926) 9 2.3:1 8.6 Beaver Dam, based Wis Smith7 Eye clinic Case series NA 11 1.4:1 32 Jervey and Anderson9 Eye clinic Case-control study NA 22 NS NS Hatfield et al10 Eye clinic Case-control study 0.5 NS 3:1 0.5 Luxenberg and Sime11 Eye clinic Case-control study NA 10 1.1:1 40 Yazar et al14 Eye clinic Case-control study NS 0 0.67:1 31 Lamba and Shukla15 Eye clinic Case series NA 17 1.7:1 23 Wasano et al12 Eye clinic Case-control study NS 18 1.3:1 28 Potter and Newcomb13 Eye clinic Observational case series 0.9 (1559) 7 13:1 19 Cockburn16 Eye clinic Case-control study NS 22 1.5:1 19 Present study Autopsy population Cross-sectional; population 1.96 (10 801) 19.8 2.5:1 7.2 based

Abbreviations: NA, not available; NS, not stated.

tistical analysis.3 However, that report surveyed a population from Beaver Dam, Wis, that was predominantly 2500

restricted to a white northern European background 2000 (99.4%). In addition, AH was diagnosed retrospectively by inspection of fundus photographs. The photographs 1500 were not interpretable in 3.6% of patients, most of whom were older (mean age, 71 years) and more likely to have 1000 a history of DM and cardiovascular disease. The authors No. of Patients indicated that AH was likely to be underestimated be- 500

cause photographs did not include the periphery and the 0 camera focused on the posterior retina rather than on the ≤10 11-20 21-30 31-40 41-50 51-60 61-70 71-80 81-90 ≥91 vitreous.3 Age, y The limitations of the previous studies prompted us Figure 1. Age distribution of autopsy population. to study the prevalence of several systemic diseases in association with AH. Our study population was large enough to permit multivariant analyses, included a wide ously.19 For analysis, 20227 eyes were studied as a subgroup cross-sectional population reflective of the United States in which PVD was scrutinized. in general, and, to our knowledge, is the only cohort study of the prevalence of AH in human autopsy eyes. STATISTICAL METHODS

␹2 METHODS Univariate analysis and multivariate logistic regression statistical methods were used to analyze our data. In the multivariate analysis, several models were tested using factors that Retrospective review of the University of California at Los An- were found to be statistically significant on ␹2 analysis (hyper- geles (UCLA) autopsy eye database from 1965 to 2000 in- tension, atherosclerosis, DM, hyperlipidemia, age-related macu- cluded 10801 patients (Figure 1). The patients’ medical rec- lar degeneration). We adjusted for age and sex in all of these ords were reviewed for evidence of DM, hypertension, models. In addition to studying AH cases overall, bilateral AH hyperlipidemia, alcohol abuse, hypercalcemia, hypothyroid- cases were specifically studied in several models in an attempt ism, and chronic renal failure, in addition to autopsy evidence to explore whether there were inherent differences between uni- of optic atrophy, glaucoma, age-related macular degeneration, lateral and bilateral AH. Age-specific prevalence of AH was cal- posterior vitreous detachment (PVD), atherosclerosis, and re- culated using 10-year interval age groups (Figure 2). Statis- nal diseases. Racial backgrounds were categorized according to tical analysis was done using SAS statistical package (SAS the classification provided in the report of the US census for 2000. Institute, Inc, Cary, NC) EXAMINATION OF RESULTS AUTOPSY EYES

Autopsy eyes were processed and examined using previously Review of our autopsy database yielded 212 cases (1.96%) described methods.18 Posterior vitreous detachment was diag- of AH, of whom 42 cases (0.4%; 42 of 212 patients nosed by suspension of the eyes in air as reported previ- [19.8%]) were bilateral. In addition to the bilateral cases,

(REPRINTED) ARCH OPHTHALMOL / VOL 123, APR 2005 WWW.ARCHOPHTHALMOL.COM 487

0.05 Logistic Regression 0.04 P Value After Adjustment 0.03 Factor ␹2 P Value for Age and Sex Ͻ Ͻ Prevalence of AH 0.02 Age .001 .001

0.01 White .71 .33 African American .65 .44 0 Asian .91 .84 ≤10 11-20 21-30 31-40 41-50 51-60 61-70 71-80 81-90 ≥91 Male .006 .046 Age, y AMD .02 .78 DM .20 .93 Figure 2. Age-specific prevalence of asteroid hyalosis (AH). Chronic renal failure .65 NA Atherosclerosis Ͻ.001 .25 Hypertension .03 .86 there were 97 cases with only the right eye affected and Alcohol abuse .44 .37 73 with only the left eye affected. Hypothyroid .16 NA Hyperlipidemia .95 .76 AGE Hypercalcemia .72 NA Absence of PVD .15 .002 Partial PVD .009 .89 In the autopsy population, the ages ranged from 0 (still- Optic atrophy .67 NA birth) to 109 years, with a mean and median of 48 and 55 years, respectively. There were a total of 4274 pa- Abbreviations: AMD, age-related macular degeneration; DM, diabetes tients (39.59%) older than 60 years, of whom 619 (5.73%) mellitus; NA, not applicable; PVD, posterior vitreous detachment. were 81 years or older, and 74 patients (0.69%) were 90 years or older. The age of patients with AH ranged from 9 to 97 years, with a mean and median of 67 years. By ␹2 analysis, age was highly associated with AH whether uni- age groups, we found specific age subgroups (51-60 years, lateral (PϽ.001) or bilateral (P=.001), an effect that per- P=.006; 41-60 years, P=.004) that showed a statisti- sisted after adjusting for sex in a multivariate model cally significant association between AH and DM. The (PϽ.001). Age-specific prevalence increased from 0.27% age groups with the lowest association were age groups in patients younger than 40 years to 6.76% in patients younger than 40 years, 81 to 90 years, and 91 years and older than 90 years (Figure 2). older; P=.50, .53, and .73, respectively.

SEX POSTERIOR VITREOUS DETACHMENT

There were 6725 males (62%) in the autopsy popula- Of the 20227 eyes studied for PVD, 243 were found to tion. Of the patients with AH, 151 (71.2%) were male, have AH. The prevalence of AH in eyes with PVD was which was statistically significant on ␹2 analysis (P=.006) only 1.39%. No association was found for AH and PVD and was less significant after adjusting for age in a mul- in our population. However, AH was associated with the tivariate model (P=.046). absence of PVD, after adjusting for age and sex (Table 2). There was no association of partial PVD with AH after RACE adjustment for age (P=.89).

The racial backgrounds of patients in the autopsy re- OTHER CONDITIONS ports revealed that 8081 patients were white (75%); 1368, African American (13%); 216, Asian (2%); and 1119 were Age-related macular degeneration, hypertension, and atherosclerosis showed apparent statistically significant as- categorized as other racial backgrounds (10%). Of the ␹2 patients with AH, 156 were white (73.6%); 29, African sociations with AH by analysis but not after adjusting American (13.7%); 4, Asian (1.9%); and 22 other (10.4%). for age and sex in the multivariate model (Table 2). No There was no statistical significance for an association other conditions, including hypercalcemia, hyperlipid- between AH and race whether in univariate or multivar- emia, alcohol abuse, hypothyroidism, chronic renal fail- iate models. ure, and optic atrophy, showed statistically significant association with AH, whether by ␹2 or multivariate analy- DIABETES MELLITUS sis (Table 2).

Diabetes mellitus was reported in 780 patients (7.22%). COMMENT Of those patients, only 20 had AH; this was not a statistically significant association by either ␹2 or multivari- The population surveyed in this study of AH represents ate analysis. Diabetes mellitus was statistically associ- a large cross-sectional population of Los Angeles. The ra- ated with bilateral AH on ␹2 analysis but not after adjusting cial backgrounds surveyed match the demographic char- for age and sex. On subgroup analysis for the different acteristics of the US population based on the 2000 cen-

(REPRINTED) ARCH OPHTHALMOL / VOL 123, APR 2005 WWW.ARCHOPHTHALMOL.COM 488

©2005 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/25/2021 sus report.20 The wide distribution obtained is the result iners. These methods are severely limited, especially in of a broad-based autopsy eye program that included sev- patients with media opacity, and may not be reliable or eral hospitals with varying demographic populations and provide an examination of the entire eye for AH. This is no selection bias for eye diseases. The largest previous reflected by the fact that in the Beaver Dam Eye Study,3 study surveyed an eye clinic population in Philadelphia, only 96.4% of the fundus photographs were gradable for Penn, of 12205 patients and provided little demo- AH. Retrospective reports suffer several limitations, in- graphic information.2 The second largest previous study cluding incomplete data acquisition, in addition to the surveyed the population of Beaver Dam, and included nonstandard methods of data collection and examina- 4926 patients, but 99.4% of patients were white.3 tion methods. This may explain why the largest clinical The broad representation of the demographic groups report to date, which surveyed 12205 subjects, yielded in our study provides a basis to study the relationship of only 101 cases of AH (0.83%).2 The authors report that race to AH. Five previous studies compared the preva- junior residents, who consulted with senior residents or lence of AH in African American patients with white pa- attending physicians, screened the study patients. In our tients but not with other racial groups. The largest of these study, one of 2 ocular pathologists thoroughly exam- reports surveyed a largely white population (99.4%) and ined all autopsy eyes with the same methods. These fac- did not report any racial difference.3 In their popula- tors may account for the nearly 2-fold higher preva- tion, Potter and Newcomb13 similarly found no racial dif- lence of AH in our study compared with previous reports. ference. In contrast, Jervey and Anderson9 reported a An association of AH with DM has been proclaimed 2-fold higher prevalence of AH among their white pa- in several previous reports, many of which were small tients; however, no details are provided regarding the over- case series.2,7,11-16 These reports have prompted the pro- all racial distribution of their surveyed population. Simi- posal that increased permeability of basal membranes larly, Rodman et al17 found a predominance of white found in the eyes of patients with DM might lead to the individuals among patients with AH, in cases selected from release of anions and macromolecules that could ini- the Armed Forces Institute of Pathology Registry, Wash- tiate development and growth and asteroid bodies.22 When ington, DC. Interestingly, Luxenberg and Sime11 criti- evaluating the association of these 2 diseases that are both cized the conclusions of Jervey and Anderson,9 noting highly prevalent in the older population, it is impera- that the 2-fold higher prevalence of AH in white pa- tive to control for age in any statistical analysis to pre- tients in their study correlated with a 2-fold higher preva- clude reporting chance association. Recent estimates of lence of white individuals in their population. In these the prevalence of diagnosed and undiagnosed DM in the reports, little is conveyed about the baseline character- elderly population approach 1 in 5 people 65 years or istics of the population surveyed. Any survey of the ra- older.23 An additional 23% of the elderly population meet cial characteristics of a population is subject to error be- the diagnostic criteria for impaired glucose tolerance, cause racial heterogeneity in the United States is often yielding an incidence that approaches 50%. The reports underreported.21 In any case, we found no association of that found a positive correlation of DM and AH were not AH with any racial category, in either multivariate or uni- controlled7,9,14,15 and/or did not adjust for age.2,7,12,13,16 Simi- variate models. lar to our report, the Beaver Dam Eye Study,3 as well as We found a statistically significant higher prevalence other retrospective case series,10,11 failed to confirm a cor- for AH among males, which became of borderline sta- relation between DM and AH. tistical significance after adjusting for age. The associa- An association of AH with DM was found in studies tion of AH with male sex has been reported previously, derived from eye clinic populations. The eye clinic popu- although the results were not adjusted for age.3,7,10,12,13,15,16 lation is biased toward patients with DM and with vi- One of the studies included a large population base3 but sion problems.24 Yazar et al14 surveyed an eye clinic popu- only 83.1% of the population participated in the survey. lation ranging in age from 50 to 89 years. In all of these The majority of nonparticipants were older men, which reports, excluding patients who are younger than 40 years may have led to underestimation of AH and its sex as- could bias results toward finding an association, espe- sociation. Of the other reports, little is known about the cially if patients are clustered in the 40- to 60-year age baseline characteristics of the population surveyed, apart group. In the age group of 40 to 60 years, we had the high- from 1 report that originated from a veterans affairs op- est prevalence of DM in our population and found a sta- tometry clinic with a 13:1 predominance of males among tistical association with asteroid hyalosis. patients with AH.13 One obvious limitation to an autopsy study is the in- The design of an autopsy eye study offers distinct ad- ability to confirm the diagnosis of associated diseases by vantages compared with previous studies. Because the data prospective clinical blood tests or studies. However, gen- for AH and the associated diseases are collected at au- eral autopsy findings were available to confirm the his- topsy, the results reflect the lifelong prevalence of AH and tory and clinical findings. It is also likely that our meth- its associations. To our knowledge, AH has never been ods overlooked cases of early or undiagnosed DM, which documented to spontaneously disappear, consistent with represent a large proportion of individuals with DM. The the observation that prevalence dramatically increases in prevalence of DM in our study was slightly lower (7.2%) each advancing decade of life (Figure 1). The second ad- than others (8.9% in the Beaver Dam Eye Study3). A re- vantage of this autopsy study is that eyes were methodi- cent report found the proportion of undiagnosed DM to cally examined for AH, PVD, and other associated patho- range from 35% to 44% of total persons with diabetes aged logic findings. Prior studies have relied on fundus 40 to 74 years between 1988 and 1994, depending on the photographs or clinical examination by multiple exam- diagnostic methods used.25 However, the reports by Ber-

(REPRINTED) ARCH OPHTHALMOL / VOL 123, APR 2005 WWW.ARCHOPHTHALMOL.COM 489

©2005 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/25/2021 gren et al,2 Wasano et al,12 and Potter and Newcomb13 Pathology and Laboratory Medicine, The David Geffen used history of prior diagnosis or treatment as their cri- School of Medicine at UCLA, 100 Stein Plaza, B-279, Los teria for diagnosis of DM. The diagnostic criteria for DM Angeles, CA 90095 ([email protected]). are continually updated. For our study and several clini- Funding/Support: This study was supported by The Un- cal studies, the diagnosis of DM reflects the changing cri- derwood Vitreoretinal Research Fund, Los Angeles, Calif. teria used by health care professionals during the pe- Acknowledgment: This article is dedicated to the memory riod in which the patients lived. of Dr Foos. Unlike previous studies, we did not find an association of AH with partial PVD. We limited the scope of this REFERENCES investigation to include cases that had been carefully ex- 26 amined for PVD as part of another study. Wasano et 1. Benson AH. Diseases of the vitreous: a case of monocular asteroid hyalitis. Trans al12 compared 59 consecutive cases of AH with 46 age- Ophthalmol Soc U K. 1894;14:101-104. matched, healthy controls in an eye clinic setting. The 2. Bergren RL, Brown GC, Duker JS. Prevalence and association of asteroid hyalo- authors found a lower prevalence of complete PVD in the sis with systemic diseases. Am J Ophthalmol. 1991;111:289-293. 3. Moss SE, Klein R, Klein BEK. Asteroid hyalosis in a population: the Beaver Dam AH population compared with controls and a signifi- Eye Study. Am J Ophthalmol. 2001;132:70-75. cantly higher incidence of partial PVD in the AH popu- 4. Parnes RE, Zakov ZN, Novak MA, Rice TA. Vitrectomy in patients with de- lation. They used slitlamp biomicroscopy to diagnose creased visual acuity secondary to asteroid hyalosis. Am J Ophthalmol. 1998; PVD, which can be limited in patients with cataract, me- 125:703-704. 5. Delaney YM, Oyinloye A, Benjamin L. Nd:YAG vitreolysis and pars plana vitrec- dia opacity, poor cooperation, narrow pupils, or intra- tomy: surgical treatment for vitreous floaters. Eye. 2002;16:21-26. vitreal liquid lacunae (syneresis). Our study confirms that 6. Vogt A. Handbook and Atlas of the Slit Lamp Microscopy of the Living Eye, Vol- there is a lower prevalence of PVD in patients with AH. ume III, Iris, Vitreous, Conjunctiva. Zurich, Switzerland: Schweizer Druck; 1941: However, unlike Wasano et al, our data show a lower 1014. prevalence of partial PVD in patients with AH after cor- 7. Smith JL. Asteroid hyalitis: incidence of diabetes mellitus and hypercholesterolemia. JAMA. 1958;168:891-893. recting for age. 8. Bard LA. Asteroid hyalitis: relationship to diabetes and hypercholesterolemia. Am The most statistically significant association with AH is J Ophthalmol. 1964;58:239-242. advancing age (PϽ.001). Only 1 previous report3 studied 9. Jervey ED, Anderson WB Jr. Asteroid hyalitis: a study of serum calcium levels in age-specific prevalence of AH. The population surveyed affected patients. South Med J. 1965;58:191-194. 10. Hatfield RE, Gasineau CF, Rucker CW. Asteroid bodies in the vitreous: relation- ranged from 43 to 86 years, and they found a progres- ship to diabetes and hypercholesterolemia. Mayo Clin Proc. 1962;37:513-514. sively increasing prevalence with advancing age (PϽ.001). 11. Luxenberg M, Sime D. Relationship of asteroid hyalosis to diabetes mellitus and Our study covered a wider age range of a more broadly dis- plasma lipid levels. Am J Ophthalmol. 1969;67:406-413. tributed population, which would explain the higher age- 12. Wasano T, Hirokawa H, Tagawa H, Trempe CL, Buzney SM. Asteroid hyalosis: specific prevalence found in our study (Figure 1). posterior vitreous detachment and diabetic retinopathy. Ann Ophthalmol. 1987; 19:255-258. The definite association of AH with age invites specu- 13. Potter JW, Newcomb R. Prevalence of asteroid bodies in a VA optometry clinic. lation as to the mechanism of AH. The asteroid bodies J Am Optom Assoc. 1980;51:19-25. feature a multilamellar structure comprising a mineral 14. Yazar Z, Hanioglu S, Karakoc G, Gursel E. Asteroid hyalosis. Eur J Ophthalmol. compound containing calcium, phosphorus, and oxy- 2001;11:57-61. 27 15. Lamba PA, Shukla KN. Vascular factors in asteroid hyalopathy. Can J Ophthalmol. gen, presumably a calcium apatite–like material. There 1971;6:46-51. is chondroitin-6-sulfate at the periphery and hyaluronic 16. Cockburn DM. Are vitreous asteroid bodies associated with diabetes mellitus? acid in the inner matrix.22 The stacks of parallel crystal- Am J Optom Physiol Opt. 1985;62:40-44. line lamellae of asteroid bodies are closely associated with 17. Rodman HI, Johnson FB, Zimmerman LE. New histopathological and histochemi- irregular collagen fibrils. It has been suggested that the cal observations concerning asteroid hyalites. Arch Ophthalmol. 1961;66:552- 563. aggregation of collagen fibers into parallel fibrils with ag- 18. Roth AM, Foos RY. A system of the macroexamination of eyes in the laboratory. 28 ing in combination with an anionic field produced by Am J Clin Pathol. 1973;59:674-683. N-acetylglucosamine found in hyaluronic acid may serve 19. Foos RY. Posterior vitreous detachment. Trans Am Acad Ophthalmol Otolaryngol. as a matrix on which calcium-apatite crystals could form.22 1972;76:480-497. 20. Profiles of general demographic characteristics: 2000 census of population and The reason for the lower prevalence of AH in patients housing United States Census Bureau Web site. Available at: http://www.census with PVD is not readily apparent. Perhaps AH is less likely .gov/prod/cen2000/dp1/2kh00.pdf. Accessed July 10, 2003. to form in the presence of vitreous detachment. Alter- 21. Parker JD, Madans JH. The correspondence between interracial births and multiple- natively, the composition of the fibrils and vitreous in race reporting. Am J Public Health. 2002;92:1976-1981. patients with AH may somehow prevent PVD. This raises 22. Winkler J, Lünsdorf H. Ultrastructure and composition of asteroid bodies. In- vest Ophthalmol Vis Sci. 2001;42:902-907. fundamental questions about the composition of the vit- 23. Harris MI. Epidemiology of diabetes mellitus among the elderly. Clin Geriatr Med. reous fibrils and biochemical changes associated with ag- 1990;6:703-719. ing. The biomolecular composition of the vitreous and 24. Orr P, Barrón Y, Schein OD, Rubin GS, West SK. Eye care utilization by older asteroid bodies has only recently been explored, and many Americans: the SEE project. Ophthalmology. 1999;106:904-909. 25. Harris MI. Diabetes in America: epidemiology and scope of the problem. Diabe- questions remain unanswered. tes Care. 1998;21(suppl 3):C11-C14. 26. Foos RY, Wheeler NC. Vitreoretinal juncture, synchysis senilis and posterior vitreous detachment. Ophthalmology. 1982;89:1502-1512. Submitted for Publication: December 30, 2003; final re- 27. Streeten BW. Vitreous asteroid bodies: ultrastructural characteristics and vision received August 12, 2004; accepted August 12, 2004. composition. Arch Ophthalmol. 1982;100:969-975. Correspondence: Ben J. Glasgow, MD, Department of 28. Sebag J. Ageing of the vitreous. Eye. 1987;1:254-262.

(REPRINTED) ARCH OPHTHALMOL / VOL 123, APR 2005 WWW.ARCHOPHTHALMOL.COM 490