Cell Loss in the Aging Retina

Rebrionship to Lipofuscin Accumubrion ond Mocubr Degenerorion

C. Kathleen Dorey,*t Gloria Wu,*t4: David Ebensrein,* Armando Garsd,* and John J.

We examined the impact of aging on the numbers of photoreceptors and retinal pigment epithelium (RPE) cells, and the number of photoreceptors per RPE cell profile, in selected regions of 30 human eyes. The mean ratio of photoreceptors to RPE cell was higher in the macula than in the paramacula (P < 0.01) or the equatorial area {P < 0.001). We found evidence for an age-related loss of RPE in both whites (P < 0.02) and blacks (P < 0.0006), although the rate of loss in whites was significantly slower than in blacks. Photoreceptor loss in blacks was inversely correlated with age (P < 0.04). In whites, however, photoreceptor loss was very significantly and directly correlated with lipofuscin concentration in the opposing RPE (P < 0.0001) and unrelated to age. The disparity in the rates of photoreceptor and RPE cell loss produced, in older eyes, a higher ratio of photoreceptors per RPE cell profile. In the macula, the ratio for whites over 50 years of age was significantly higher (P < 0.05) than that in blacks over 50. Our data suggest that the increased phagocytic and metabolic load on the RPE cell in the macula causes a preferential age-related accumulation of lipofuscin in the RPE, which ultimately leads to photoreceptor death. This may prove a useful model of age-related macular degeneration and Stargardt's disease. Invest Ophthalmol Vis Sci 30:1691-1699,1989

The predilection of age-related retinal degeneration ties within the RPE cell1 are greater in whites who are for the macula has never been explained. One possi- at greater risk for macular degeneration.5'6 To date, ble explanation is that macular degeneration occurs no direct evidence exists to show that retinal degen- when lipofuscin accumulation in the RPE cell eration is due to lipofuscin. reaches a threshold. We and others have reported that Both the site specificity of degeneration and the the RPE in this area exhibits the greatest accumula- increased lipofuscin in the macula have been attrib- tion of lipofuscin1'2 and that the annular pattern in uted to increased exposure to environmental light.78 macular degeneration is remarkably well correlated However, a study by Kooijman,9 recently confirmed with the area of greatest lipofuscin concentration.3 by Pflibsen and colleagues,10 revealed no evidence of The preferential accumulation of lipofuscin in the increased retinal illumination in the posterior pole, macula may be partially explained by the topo- but rather a homogeneous retinal light distribution graphic distribution of RPE melanin and the negative regardless of pupil size. Thus, if the amount of light linear relationship between lipofuscin and melanin exposure is uniform, the higher levels of lipofuscin concentrations in individual human RPE cells.1 accumulation must depend on factors other than However, since the coefficient of determination, r2, light. was only 0.28 (P < 0.001) for melanin, other factors Most of the lipofuscin within the RPE may be may also influence lipofuscin concentration (eg, age, safely assumed to derive from phagocytosis of oxida- genetics or light damage). Moreover, although both tively damaged (and nondegradable) lipids in the blacks and whites exhibit age-related accumulation of photoreceptor outer segments.11"14 Therefore, RPE lipofuscin,4 the rate of accumulation4 and the quanti- cells phagocytosing greater quantities of outer segment lipoids may accumulate more lipofuscin. Evi- dence for this concept is found in a study by Katz et From the *Macular Disease Research Center, Eye Research In- 15 stitute, and fDepartment of Ophthalmology, Harvard University, al that demonstrated that lipofuscin deposition de- and ^Retina Associates, Boston, Massachusetts. creased significantly in retinal degenerate (RD) rats Presented in part at the 59th Annual ARVO Spring Meeting, only after photoreceptors were lost. In a study of the Sarasota, Florida, May 1987. aging human macula, Gartner and Henkind16 re- Supported by grant EY-06544 from the National Eye Institute ported a reduced thickness in the outer nuclear layer (CKD), and the Louis Morganstern Memorial Scientist Fund. Submitted for publication: May 18, 1988; February 27, 1989. and an increase in the number of distally displaced Reprint requests: C. Kathleen Dorey, PhD, Eye Research Insti- photoreceptor nuclei; they interpreted the latter find- tute, 20 Staniford Street, Boston, MA 02114. ing as evidence of a photoreceptor loss. Since the loss

1691

Downloaded from iovs.arvojournals.org on 09/30/2021 1692 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / Augusr 1989 Vol. 30

regular intervals between the fovea and periphery in several eyes. Based on these results, six sites per eye were selected for counting of the photoreceptor and RPE nuclei: the fovea, the nasal posterior pole, two parafovea, and the two equatorial sites (Fig. 1). The parafoveal sites were located at half the distance from the fovea to the disc at either side of the fovea. The distance between the fovea and the disc was measured on the nasal aspect of the posterior pole, that point being designated the nasal posterior pole. The lipofuscin fluorescence measurements previously reported1 were also obtained from the same sites in these eyes. Donor eyes, obtained less than 12 hr after death, were fixed in formalin and embedded in paraffin. Pupil-optic nerve sections 8 jim thick were cut through the fovea, mounted on slides and deparaffin- ized, and coverslipped in ultraviolet-inert immersion oil. Unstained slides were used for lipofuscin measurement; adjacent sections were stained with hema- toxylin to facilitate counts of nuclei. A duplicate ad- RETINAL REGIONS EXAMINED jacent section was not always available for staining, so-final counts of RPE nuclei were made on 19 TE - TEMPORAL EQUATOR - PARAFOVEA dtfiiors: ten blacks (ages 8, 20, 27, 39, 40, 46, 53, 65,

Pu_ F FOVEA 69 and 80; mean = 43.7 years) and nine whites (ages NP - NASAL POSTERIOR POLE NE - NASAL EQUATOR 6 weeks, 10, 28, 36, 61, 70, 73, 82 and 88; mean = 51 years). Sites were examined using a Zeiss photomi- croscope III at a magnification of X320. Photorecep- Fig. 1. Retinal sites studied. tor nuclei were counted within a 90 ^m long X 8 /xm deep area from each site. Estimates of photoreceptor of photoreceptors in the macula should theoretically density assumed triangular spacing. Photoreceptor decrease the lipofuscin content of the RPE, these ob- and RPE data from identical regions were used to servations suggested to us that excessive lipofuscin in obtain a ratio for that region. the macular RPE might cause photoreceptor degen- RPE cell sizes were estimated from the number of eration or that both age-related changes might share a nuclei within an area 90 /tm long X 8 /xm deep (the common etiology. In fact we previously described a thickness of the section). To calculate the diameter of significant correlation between the accumulation of 4 the cell, we assumed that each nucleus represented a lipofuscin and the loss of photoreceptors in whites. whole cell, and that the monolayer was composed of The current study examined cell loss in the aging equal sized hexagons with an area in /xm2 of 3.5 R2, retina and its impact on the photoreceptor/RPE where R is the perpendicular distance in nm from the complex. We report the firstevidenc e that: (1) there is middle of a side to the center of the hexagon. an age-related increase in the phagocytic and metabolic load on the macular RPE; and (2) that lipofuscin accumulation is correlated with photoreceptor Statistical Methods loss in the human macula. Before further analysis, extreme values in each group were examined to identify aberrant observa- Materials and Methods tions; none were found. The effect of age on any Thirty donor eyes with no known ocular pathology given variable was first assessed by simple linear re- were studied: 15 from whites (age range 2 weeks to 88 gression; however, to compare the relative effects of years, mean age 50 years) and 15 from blacks (age several variables on the formation of lipofuscin, and range 8 to 84 years, mean age 49 years). Through the loss of photoreceptors or RPE cells, multiple lin- randomization when necessary, only one eye was in- ear regressions were performed. cluded for each donor. To evaluate a model of photoreceptor loss, the To obtain general topographic information on the model submitted for statistical analysis expressed the photoreceptor and RPE nuclei, counts were made at number of photoreceptors (PRN)j in the macula of a

Downloaded from iovs.arvojournals.org on 09/30/2021 No. 8 ELEVATED PHAGOCYTIC LOAD ON RPE CELLS IN THE AGING MACULA / Dorey er ol 1693

single individual (i) as a possible function of that per- little, and the standard deviations were usually less son's age, race, and the level of RPE lipofuscin (LF) than 10% of the mean. In the parafovea, the photore- in relative fluorescence units. Since race was known ceptor counts ranged from 115 to a low of 50 in the to influence ratio and lipofuscin concentration, to same 80-year-old eye; standard deviations were rarely permit statistical recognition of a racial influence the more than 20% of the mean. There was greater heter- model was expanded by interaction terms (ie, the ogeneity in the density of photoreceptors in the equa- product of the variable and a race indicator (RI) set torial region; 11 out of 18 eyes exhibited greater than equal to 1 for whites and 0 for blacks). Specifically, 20% variation. Although the regression line fitted to our model for statistical analysis was the following. the data indicated an annual decrease of 1-2% in the For each individual i (i = 1,..., 30): photoreceptor number in the macular and paramacular regions, none of the regions was found to PRNi = Bo + B,(Age)i + B2(LF)i + B3(RI)i exhibit a significant age-related loss of photorecep-

+ B4(RI)i X (LF); + E; (1) tors. When multiple stepwise regression was used to where all the Bs are unknown parameters to be esti- identify the variables that influenced the number of mated from the data and where Ej is a random term. photoreceptors remaining [Eq. (1) in Methods], the To further identify the variables that might be in- only significant factor was race. Further analysis on fluencing lipofuscin accumulation, we modified separate racial groups revealed that the determinant equation (1) as follows. We included age in years, the with the most significant association with the number number of photoreceptors (PRN)i opposing the RPE of photoreceptors remaining was not the same in in the area of lipofuscin (LF) measurement, and a ; black and white eyes. The number of photoreceptors race indicator (RI)i equal to either 1 or 0 as before. remaining in the macula of blacks was inversely re- For each individual i (i = 1,.. ., 30): lated to age. The final model for blacks was: PRN4 2 LFj = Bo + B,(Age)i + B2(PRN)i + B3(RI)i = 22 - 0.06 (Age);; (P < 0.006, r = 0.43). However, in whites a strong inverse relationship was deter- + B4(RI)i X (Age)i + B5(RI)i X (PRN); + E; (2) mined between photoreceptor number and the We applied the Median Test17 to determine amount of lipofuscin in the opposing RPE. The rela- whether there were significant differences in the pho- tionship illustrated in Figure 3 was: PRNj = 23.7 2 toreceptor load on the RPE cell (ie, the ratio (R) of - 0.01 (LF)J; (r = 0.59; P < 0.0001). photoreceptors to RPE cell) in eyes of different ages Age was not a significant regressor of photorecep- or races, or in different areas within an eye. Since the tor number in whites (P < 0.9), even though the ac- plane of section would only rarely coincide with the cumulation of lipofuscin was age-related, and the in- diameter of the cell, we have expressed this ratio as dividuals with highest lipofuscin values ranged in age the number of photoreceptors per RPE cell profile. from 73 to 88 years. The final models predicted very The Median Test, a variation of the Chi-square test, is similar starting numbers of photoreceptors for both completely nonparametric and makes no assump- blacks (22/90 ixm) and whites (23.7/90 ^m). Because tions about the distributions of ratios. In those cases displaced photoreceptor nuclei have been considered where significance was not found with the Median evidence of photoreceptor loss, we also considered Test, we confirmed the result with the more powerful their number. Displacement of photoreceptor nuclei Mann-Whitney test17 (after first testing for equality of in the macular region was constant in blacks and variances with the Ansari-Bradley procedure). whites of all ages (Fig. 4), and the number of displaced nuclei did not correspond to the rate of pho- Results toreceptor loss. Photoreceptor Number Increase in Lipofuscin Figure 2 presents the mean and standard deviation of paired photoreceptor and RPE cell counts in six Equation (2) in Methods proposed several possible adjacent 15 nm long fields in the macula of 19 indi- factors related to the amount of lipofuscin. Among vidual eyes for which RPE cell counts were also avail- these, age, race and, in whites only, the number of able. The means in the paramacular and equatorial photoreceptors were found to be significant. The final regions :weresdetermined from six fields in each of the equation found in Table 1 indicated that the amount nasal and temporal sites. The number of photorecep- of lipofuscin in the RPE of an individual can be satis- 2 tors per 720 /urn area in the macula varied from over factorily predicted by adding: (1) a basal value (Bo) 200 in ah 8-year-old eye to a low of 75 in an 80-year- and (2) the amount accumulated during that individ- old eye. Within one individual, counts varied very ual's life. In whites, additional components adjust for

Downloaded from iovs.arvojournals.org on 09/30/2021 1694 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / August 1989 Vol. 30

PR - MACULA 220 • \

170 • i

120 • 1 i M I — • —i——— MX ] 70 - I * ! 0

20 - O 20 100 o> E CO o cs> LU O LU LU o O LU Q. 5 20 40 60 80 100 CC 140 PR - EQUATORIAL 120 -

100

80 -

60 - 40 • FT 20 20 40 60 80 100 40 60 100 AGE (yrs) AGE (yrs) Fig. 2. Counts of photoreceptors (left) and the overlying RPE cells (right) observed in three retinal regions of 18 human eyes. Mean and standard deviation of six adjacent microscope fields of 15 ^m each are presented in the macula. In the other retinal regions, two areas of six adjacent fields were counted on opposite sides of the fovea, and combined for calculation of the mean. Superimposed regression lines were not significant for the photoreceptors, but were for the RPE cells,.

their greater lipofuscin accumulation and for the = -0.07; P < 0.0006; r2 = 0.46); the difference in the greater quantity of lipofuscin observed in sites with slopes was highly significant (P < 0.0001). Among the fewer photoreceptors. Race interacts with PRN sig- eyes over 20 years old the mean RPE cell count was nificantly. The overall coefficient of determination 8.1 ±3, representing an estimated mean area of 88.9 (r2) for the final equation in Table 1 was highly signif- nm2, and a minimum diameter of 10.1 nm. icant, and an excellent fit of the data (r2 = 0.79; P < 0.0001). Phagocytic Load The rates of cell loss in photoreceptors and RPE Loss of RPE cells were not equal in this sample. In order to assess The number of RPE nuclei in the posterior pole the effects of photoreceptor and RPE cell loss on the decreased significantly with age in whites (Bi relationship between them, we determined the = -0.05, P < 0.02, r2 = 0.64) and in blacks (B, phagocytic load on the RPE, that is, the ratio of the

Downloaded from iovs.arvojournals.org on 09/30/2021 No. 8 ELEVATED PHAGOCYTIC LOAD ON RPE CELLS IN THE AGING MACULA / Dorey er ol 1695

number of photoreceptors to the number of RPE cells o in the six areas of each eye (Fig. 5), and compared the z ratios of various regions via the Median Test. Our 25- results, presented in Table 2, demonstrate that the PR = -(0.01[LF]) + 23.7

ratio is significantly higher in the macula than in the 2 paramacula (P < 0.01), or in the periphery (P r =0.I < 0.00002). Moreover, the RPE cells in the macula of p < 0.0001 whites over 50 years of age oppose more photoreceptors than do those in the macula of older blacks (P < 0.05). In fact, the ratio in older whites was significantly higher than that found in all other maculas (P < 0.015). When age effects were not included we were K unable to identify significant racial differences in the o ratios found in the macula. X 0 500 1000 1500 2000 Cursory inspection of the data indicated that the regions closer to the macula tended to have higher RPE LIPORJCSIN CONCENTRATION ratios. Statistical comparisons of these regions in eyes (arbitrary units) of all ages and both races determined that the tem- Fig. 3. Relationship between lipofuscin concentration and the poral paramacula had more photoreceptors per RPE number of photoreceptors found in the overlying macula of 15 whites. Each data point represents the mean of six independent cell profile than did the nasal (P < 0.05). If only those observations in the macular region of a single eye. A similar but less over 50 years of age were considered, the differences significant (P < 0.02) relationship was found in the paramacula. in ratio were more significant (P < 0.02). Similarly, the temporal equator had much higher ratios than did the nasal equator, whether eyes of all ages were con- no correlation between their number and age. Our sidered (P < 0.001) or only those of donors over 50 study showed an average of 2.8 + 1 displaced photo- years of age (P < 0.00002). receptor nuclei (DN) per 90 /*m in the fovea, compa- rable to the average of 33 DN/1500 jim found by Discussion Gartner and Henkind in 50-70-year-olds. If each displaced nucleus died within 1 year, the rate of photo- Aging changes are associated with cell loss and de- receptor nuclear loss would be six times that actually 18 generation in different parts of biological systems. observed in our data. This suggested that either cells In the human retina an age-related loss of ganglion with displaced nuclei required many years to die or cells has been identified by examining axonal densi- that nuclear displacement was not related to cell ties in the optic nerve,19 and age-related loss of pho- death. We also found that photoreceptor loss was toreceptors has been inferred from morphologic stud- constant over age. Therefore we concluded that dis- ies,20 and from pyschophysical demonstrations of placement of nuclei is not a valid indicator of pho- age-related declines in color perception,21 visual toreceptor death. acuity,22 sensitivity23 and foveal cone pigment density.24 However, we know of no previous attempts to identify possible causes of cell loss in the retina. E m Photoreceptor Density 0 " LEGBJD 16 n FOVEAL In 1981 Gartner and Henkind reported an obvi- D UJ 5" • PARAFOVEAL ous reduction in the thickness of the outer nuclear _j O layer (ONL) of the retina in older eyes, but did not z> 4- 1 DD D • a D count the nuclei remaining. They also described a z distal displacement of photoreceptor nuclei seen in- oc 3- a D • creasingly in the macular area after age 40. They in- Q. a terpreted the displacement as a stage of cell loss and UJ 2- • a aa a » •• •• a* c concluded that photoreceptor loss was accelerated in o those over 40 years of age. Also not counting the 25 1 - • nuclei in the ONL, Lai and Rana found signifi- _ 0 20 40 60 80 100 cantly more displaced nuclei in the retina of a 20- a AGE (yrs) year-old monkey than in a 4- or 10-year-old retina. Fig. 4. Number of photoreceptor nuclei displaced into the outer We too observed distally displaced nuclei but found segment region observed within high-power fields of 15 /

Downloaded from iovs.arvojournals.org on 09/30/2021 1696 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / August 1989 Vol. 30

26 Table 1. Significant covariates of lipofuscin were found in studies by Curcio and colleagues and in human RPE by Schien.28 In order to compare our values for foveal LF> = cone densities it is necessary to process the data to B0 + B1(Age)i + B2(RDi "r- B5(RI)i(!PRN)i + E; adjust for the geometry of our samples. We counted (r2 = 0.79, P < 0.0001) photoreceptors in a 720 /tm2 area (90 /im long by 8 ^m thick tissue section). The maximum cone density Variable B P value estimated from our data was 125 cones/720 ^m2. 2 Baseline 2.46 This indicates a cone spacing of 1 cone/5.76 /xm or a Age 0.004 0.0001 linear spacing of 1 cone/2.4 /xm in the fovea of the Rit 1.37 0.0001 new born. (RI) X (PRN)J - 1.17 0.0002 * LFj = the lipofuscin in individual i. t RI = race indicator: 1 = whites, 0 = blacks. Lipofuscin % (RI)j X (PRN)j is the product of the race indicator and the number of photoreceptors, a term to recognize race-related interaction in loss of pho- These data provide possibly the first evidence for toreceptors. the concept that increased lipofuscin correlates with photoreceptor loss.3'29"31 The greatest lipofuscin ac- In these data the maximum peak foveal cone den- cumulation in whites was associated with the greatest sity was 1.9 times greater than the minimum found in photoreceptor loss, (ie, photoreceptor number was four retinas with mean age 36.7 years; Curcio et al negatively correlated with the accumulation of lipo- found that the foveal cone density in four adult fuscin but not with age). The RPE of some whites had human retinas (mean age 35) exhibited a 2.9-fold more lipofuscin than would be expected for their age; range of variation.26 The latter group reported a fo- the same eyes had lower numbers of photoreceptors. veal cone density of 200,000 cells/mm2 or a linear Although we demonstrated a significant age-depen- cone spacing of 1 cone/2.2 fim. This value corre- dent accumulation of lipofuscin in both blacks and sponds well with Osterberg's data27 and with the cone whites, lipofuscin was not a strong determinant of spacing of 120 cones/degree (120 cones/270 ^m) or 1 photoreceptor loss in blacks. This apparent contra- cone/2.2 /um determined from physiological optics. diction may indicate that below a critical threshold In the macaque monkey similar foveal cone densities lipofuscin does not have negative impact on the sur- vival of photoreceptors. Studies in other model systems have also suggested that lipofuscin accumulation may reach a threshold which compromises sur- vival.3132 We consider that lipofuscin accumulation caused photoreceptor loss and not vice versa, since Katz and Eldred demonstrate^ that loss of photoreceptors in rats resulted in lower levels of lipofuscin.15'33 In both of their models, photoreceptor loss preceded a decrease in lipofuscin. A dying photoreceptor might temporarily increase the load on an RPE cell, but since a photoreceptor is replaced ap- proximately every 10 days, the decrease in the amount of material processed will be significant after only a few months.

RPE Cells Tso and Friedman34 in 1968 showed that the number of RPE cells decreased in the posterior pole with age. Our data supported this; the number of RPE cells decreased significantly with age both in blacks (P < 0.02) and in whites (P < 0.01). Also agreeing with the work of Tso and Friedman, our study found no PF F PF NP NE significant changes in the equatorial regions. 2 RETINAL REGION In this study and in others, an age-related increase Fig. 5. Ratio of mean photoreceptor number per RPE cell profile in the height of the RPE has been observed. Unfortu- in each of six 720 nm2 regions of 19 eyes. Abbreviations as in Figure nately, these data have sometimes been erroneously 1. cited as evidence that RPE cells become taller and

Downloaded from iovs.arvojournals.org on 09/30/2021 No. 8 ELEVATED PHAGOCYTIC LOAD ON RPE CELLS IN THE AGING MACULA / Dorey er ol 1697

Table 2. Comparisons of photoreceptor to RPE ratios in regions of human eyes

Region 1 vs. Region 2

Macula vs. Macula Result Significance

Ages Race Ages Race P value*

All W All B A = B N.S. >50 W >50 B A>Bf 0.05 <50 W <50 B A = B N.S. >50 All <50 All A = B N.S. >50 W All others A> B 0.05 Macula vs. Paramacula All All All All A>B 0.01 >50 All >50 All A>B 0.01 Macula vs. Equator All All All All A> B 0.0001 >50 All >50 All A>B 0.05 T-Paramacula vs. N-paramacula All All All All A> B 0.05 >50 All >50 All A> B 0.05 T-Equator vs. N-Equator All All All All A> B 0.001 >50 All >50 All A>B 0.001

W = whites; B = blacks; T = Temporal; N =Nasal. * Via Median Test, N.S.: Not Significant. t A > B: A is significantly greater than B.

narrower with age. In fact we know of only two pre- macula and decreased in all other regions. In the eight vious studies that have examined the diameters of pairs of eyes over age 20 in her study, the RPE cell RPE cells in various regions of the eye, only one of density ranged from 29-^5 cells/0.004 mm2 or 29-35 which considered the effect of age. Young's study of cells/3969 nm2. The area of an RPE cell would there- two young monkeys35 was consistent with our data in fore range from 113-137 nm2 and, assuming hexago- the approximate number of photoreceptors per RPE nal packing, the diameters would be from 11.4- cell and in the location of the highest ratio of photo- 12.5/xm. receptors to RPE cell in the posterior pole—in the In our sample of 15 eyes over 20 years of age, the human macula and in the monkey "perifovea." mean count of RPE cells in the macula was 8.1 ± 3 Young has made the only precise measurement of cells/720 nm2. With hexagonal packing the average RPE cell areas; in the young monkey fovea and para- adult RPE cell would occupy 88.9 ^m2, and would fovea the cells occupy areas of 186.3 and 336.2 ^m2, have a diameter of 10.1 /im, considerably smaller respectively.35 Assuming hexagonal packing, the di- than the 14 ^m previously reported for macular RPE ameters would be 14.6 in the fovea and 19.6 in the cells.34'35 However, our enumeration of RPE cells parafovea. Tso and Friedman34 estimated the average considered each nucleus to be equivalent to a cell. In RPE diameter to be 14 nm in the human posterior fact many of the nuclei counted would not represent pole. cells wholly within the section, and the number of Streeten36 has extensively studied the development RPE cells would be overestimated. Consequently of the RPE from fetal through adult life. She reported both the size of the RPE cell and the number of pho- regional differences in the size of the human RPE cell toreceptors per RPE cell profile are underestimates of and an age-related increase in pleomorphism, but de- the true situation. Based on the discrepancy in size scribed an age-related increase in size only in the estimation our data may overestimate RPE cell den- areas adjacent to the ora. A further comment in the sity, and underestimate phagocytic load by as much discussion that "cell densities tended to decrease in as 40%. It is likely that Streeten's counting technique all areas but the macula, where they tended to in- also overestimated the actual number (and underes- crease" has been misapplied to the aging macula. timated RPE size), since she reported that decreasing Streeten demonstrated that between the ages of 0 and the size of the counting window caused an increase in 3 years, RPE cell density increased sharply in the the estimate of cell density. Although we could not

Downloaded from iovs.arvojournals.org on 09/30/2021 1698 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / Augusr 1989 Vol. 30

reliably distinguish all multinucleated RPE cells, situations characterized by elevated levels of lipofus- their contribution to the overestimate would be cin: in the macula more than the periphery, in whites small; only 3% of human RPE are binucleate.37 more than blacks, in older eyes more than younger ones. Moreover, we have identified: (1) a strong nega- Phagocytic Load tive correlation between the amount of RPE lipofus- The relative increase in the number of photorecep- cin and the number of photoreceptors present in the tors served by the older macular RPE cell is a conse- eye; and (2) that age seems to exacerbate the demands quence of a disproportionate loss of RPE cells vis-a- on the RPE, particularly in the macula. Not only is vis the photoreceptors. The phagocytic load on the there an age-related elevation in RPE lipofuscin, but RPE can be estimated by considering the ratio of also a discrepancy between the rates of loss of pho- photoreceptors to RPE cell. We cannot be sure that toreceptors and RPE cells, producing higher ratios of the observed increase is entirely due to age, since our photoreceptors to RPE cell in the macula of older sample of those over 50 years of age contained more individuals. Taken together, these two lines of evi- eyes from whites than from blacks. However, it is dence suggest the following sequence of events: (1) clear that the ratio in the macula of older whites is the unequal loss of photoreceptors and RPE cells higher than in any other group. Our sample indicated causes (2) an increase in the phagocytic load on the that the discrepancy in the rates of cell loss was RPE cells, particularly in the macula; (3) the elevated greater in whites than in blacks. The increase in ratio phagocytic load causes the well documented greater implies that when one RPE cell dies the adjacent cells RPE lipofuscin accumulation in the macula, which expand to fill the space. As the number of photore- may in turn (4) accelerate photoreceptor death. Alter- ceptors per remaining RPE cells increases, the meta- natively, (5) the extensive compromise of RPE cyto- bolic demands on the individual RPE cell will also plasm by massive accumulations of lipofuscin may increase. Theoretically, the aging RPE cells would compromise essential RPE functions enough to in- also bear larger phagocytic loads, and consequently crease the vulnerability of the macular RPE in older accumulate lipofuscin at a faster rate. Both the whites or even (6) cause the death of the photorecep- greater accumulation of lipofuscin in whites1 and the tors and/or RPE cells. We therefore hypothesize that increase in rate of accumulation in older eyes1'2 could age-related macular degeneration may be lipofuscin- be the natural result of the increased phagocytic load related. This hypothesis predicts that factors in- on the older RPE cells. creasing RPE lipofuscin, including genetic factors (Stargardt's disease, fundus flavi,38 cone and rod de- Could the observed increase in ratio be incorrect? 39 40 As indicated above, all of our estimates of phagocytic generation, Best's disease, decreased ocular pig- mentation16), environmental factors (light-induced load are very conservative. However, since loss of 1433 RPE is considered the underlying process in the so- retinal damage ), or nutritional deficiencies in vitamins C and E7'41 would increase the risk for degen- called dry type of macular degeneration, we cannot 32 exclude the possibility that the older whites in this eration of photoreceptors, RPE cells, or both. case had a prodermal stage of age-related macular Key words: lipofuscin, cell loss, photoreceptors, retinal pig- degeneration, and that our observations simply dem- ment epithelium (RPE), phagocytic load, age-related macu- onstrate early events in this process. Moreover, al- lar degeneration, Stargardt's disease though our data were carefully obtained and the differences observed were highly significant, we Acknowledgments recognize the presence of an uncontrolled variable— The authors are grateful to Susan A. Curran for editorial swelling in the macula after death.2 Our data were assistance and to Drs. Stanley Schein of Massachusetts Eye obtained from donor eyes and we may safely assume and Ear Infirmary and Yoshihura Torige of UC Irvine for that none was fixed sooner than 2 hr after death. It is their advice concerning the relationship between nuclear counts and actual cell number in tissue sections. We thank conceivable that disproportionate lateral swelling in Dr. Francois Delori for his critical reading of the manu- the macular RPE cells could contribute to the ob- script and for many helpful discussions during the course of served increase in the ratio. However, if we accept this work. The assistance of Karlotta Fitch made it feasible this argument we must also accept its corollary, that to use the specimens on which well documented lipofuscin the macula swells more in older whites. In either case measurements had already been made. The support and encouragement of Mr. David Dresnick is gratefully ac- our data present strong evidence that the RPE cells in knowledged. the macula of older whites can not be considered equivalent to the RPE cells in other regions or to the References macula of younger eyes. 1. Weiter JJ, Delori FC, Wing GL, and Fitch KA: Retinal pig- In conclusion, we have demonstrated that the ment epithelial lipofuscin and melanin and choroidal melanin number of photoreceptors per RPE cell is higher in in human eyes. Invest Ophthalmol Vis Sci 27:145, 1986.

Downloaded from iovs.arvojournals.org on 09/30/2021 No. 8 ELEVATED PHAGOCYTIC LOAD ON RPE CELLS IN THE AGING MACULA / Dorey er ol 1699

2. Feeney-Burns L, Hilderbrand ES, and Eldridge S: Aging 21. Ohta Y and Kato H: Colour perception changes with age. Mod human RPE: Morphometric analysis of macular, equatorial Probl Ophthalmol 17:345, 1975. and peripheral cells. Invest Ophthalmol Vis Sci 25:195, 1984. 22. Weale RA: Senile changes in visual acuity. Trans Ophthalmol 3. Weiter JJ, Delori FC, and Dorey CK: Central sparing in annu- Soc UK 95:36, 1975. lar macular degeneration. Am J Ophthalmol 106:286, 1988. 23. McFarland RA, Domey RG, Warren AB, and Ward DC: 4. Wu G, Ebenstein D, Weiter JJ, and Dorey CK: Photoreceptor Dark-adaptation as a function of age: I. A statistical analysis. J nuclei: A relationship to aging and to lipofuscin. Invest Gerontol 15:149, 1960. Ophthalmol Vis Sci 28:263, 1987. 24. Kilbride PE, Hutman LP, Fishman M, and Read JS: Foveal 5. Gregor Z and Joffe L: Senile macular changes in the black cone pigment density difference in the aging eye. Vision Res African. Br J Ophthalmol 62:547, 1978. 26:321, 1986. 6. Weiter JJ, Delori FC, Wing GL, and Fitch KA: Relationship of 25. Lai YL and Rana MW: A study of photoreceptor-retinal pig- senile macular degeneration to ocular pigmentation. Am J ment epithelium complex: Age-related changes in monkeys Ophthalmol 99:185, 1985. (42267). Proc Soc Exp Biol Med 181:371, 1986. 7. Katz ML, Parker KR, Handelman GJ, Bramel TL, and Dratz 26. Curcio CA, Sloan KR Jr, Packer O, Hendrickson AE, and EA: Effects of antioxidant nutrient deficiency on the retina and Kalina RE: Distribution of cones in human and monkey ret- retinal pigment epithelium of albino rats: A light and electron ina: Individual variability and radial asymmetry. Science microscopic study. Exp Eye Res 34:339, 1982. 236:579, 1987. 8. Weiter JJ, Delori FC, Wing GL, and Fitch K: Relationship of 27. Osterberg GA: Topography of the layer of rods and cones in senile macular degeneration to ocular pigmentation. Am J the human retina. Acta Ophthalmol 13(Suppl 6):1, 1935. Ophthalmol 99:185, 1985. 28. Schein SJ: Anatomy of macaque fovea and spatial densities of 9. Kooijman AC: Light distribution on the retina of a wide-angle foveal representation. J Comp Neurol 269:479, 1988. theoretical eye. J Opt Soc Am 73:1544, 1983. 29. Young RW: Pathophysiology of age-related macular degener- 10. Pflibsen KP, Pomerantzeff O, and Ross RN: Retinal illumi- ation. Surv Ophthalmol 31:291, 1987. nance using a wide-angle model of the eye. J Opt Soc Am A 30. Young RW: A theory of retinal disease. In New Directions in 5:146, 1988. Ophthalmic Research, Sears ML, editor. New Haven, Yale 11. Essner E and Novikott'AB: Human hepatocellular pigments University Press, 1981, pp. 237-270. and lysosomes. J Ultrastruct Res 3:374, 1960. 31. Boulton M and Marshall J: Effects of increasing numbers of 12. Brunk U and Ericsson JL: Electron microscopical studies on phagocytic inclusions on human retinal pigment epithelial rat brain neurons: Localization of acid phosphatase and mode cells in culture: A model for aging. Br J Ophthalmol 70:806, of formation of lipofuscin bodies. J Ultrastruct Res 38:1, 1972. 1986. 13. Dabholkar AS and Ogawa K: Ultracytochemical changes in 32. Sohal RW and Donato H: Effects of experimentally altered the distribution of acid phosphatase in the liver cells of vitamin lifespan on the accumulations of fluorescent age pigments in E deficient and vitamin E supplemented rats. Acta Histochem the housefly Musca domeslica. Exp Gerontol 13:335, 1978. Cytochem 11:195, 1978. 33. Katz ML and Eldred GE: Retinal light damage reduces auto- 14. Feeney-Burns L and Eldred GE: The fate of the phagosome: fluorescent pigment deposition in the retinal pigment epithe- Conversion to "age pigment" and impact on human retinal lium. Invest Ophthal Vis Sci 30:37, 1989. pigment epithelium. Trans Ophthalmol Soc UK 103:416, 34. Tso MOM and Friedman E: The retinal pigment epithelium: 1984. III. Growth and development. Arch Ophthalmol 80:214, 1968. 35. Young RW: The renewal of rod and cone outer segments in the 15. Katz ML, Drea CM, Eldred GE, Hess HH, and Robison WG rhesus monkey. J Cell Biol 49:303, 1971. Jr: Influence of early photoreceptor degeneration on lipofuscin 36. Streeten B: Development of the human retinal pigment epithe- in the retinal pigment epithelium. Exp Eye Res 43:561, 1986. lium and the posterior segment. Arch Ophthalmol 81:383, 16. Gartner S and Henkind P: Aging and degeneration of the 1969. human macula. Br J Ophthalmol 65:23. 1981. 37. Tso MOM and Friedman E: The retinal pigment epithelium: I. 17. Conover WJ: Practical Nonparametric Statistics, 2nd Edition. Comparative histology. Arch Ophthalmol 78:641, 1967. New York, John Wiley and Sons, 1980, pp. 171,216. 38. Eagle RC Jr, Lucier AC, Bernadino VB, and Janoff M: Retinal 18. Ordy JM and Brizzee KR: Functional and structural age dif- pigment epithelial abnormalities in fundus flavimaculatus: A ferences in the visual system of man and nonhuman primate light and electron microscopical study. Ophthalmology models. In Sensory Systems and Communication in the El- 87:1189, 1980. derly, Aging, Vol. 10, Ordy JM and Brizzee KR, editors. New 39. Rabb MF, Tso MO, and Fishman GA: Cone-rod dystrophy: A York, Raven Press, 1979, pp. 13-50. clinical and histopathologic report. Ophthalmology 93:1443, 19. Balazsi AG, Rootman J, Drance SM, Schulzer M, and Douglas 1986. GR: The effect of age on the nerve fiber population of the 40. Weingeist TA, Kobrin JL, and Watzke RC: Histopathology of human optic nerve. Am J Ophthalmol 97:760, 1984. Best's macular dystrophy. Arch Ophthalmol 100:1108, 1982. 20. Marshall J, Grindle J, Ansell PL, and Borwein B: Convolution 41. Robison WG Jr, Kuwubara T, and Bieri JG: Deficiencies of in human rods: An ageing process. Br J Ophthalmol 63:181, Vitamins E and A in the rat: Retinal damage and lipofuscin 1979. accumulation. Invest Ophthalmol Vis Sci 19:1030, 1980.

Downloaded from iovs.arvojournals.org on 09/30/2021