Increased Plasma Noncortisol Glucocorticoid Activity in Open-Angle Glaucoma

George W. McCarry* and Bernard Schwartz

Total biologic plasma glucocorticoid activity of normal, ocular hypertensive, and open-angle glaucoma patients was compared using a glucocorticoid receptor-based competitive binding assay. Multiple linear-regression analysis was used to adjust for the effects of significant ocular and nonocular variables, including therapy for glaucoma. The glaucoma patients had significantly greater plasma glucocorticoid activities than did normal subjects. A comparison of receptor-based assay values to values obtained with a cortisol radioimmunoassay showed that significant amounts of biologic glucocorticoid activity in the plasma of the glaucoma patients could not be explained by cortisol alone. In the normal and ocular hypertensive groups, however, virtually all of the plasma glucocorticoid activity could be accounted for by cortisol. These results suggest that in open-angle glaucoma patients, noncortisol glucocorticoids are responsible for elevating biologic plasma glucocorticoid activity. Thus, open-angle glaucoma may be associated with a disturbance of the hypothalamic-pituitary-adrenal axis that pro- duces increased plasma levels of both cortisol and other noncortisol glucocorticoids. Invest Ophthalmol Vis Sci 32:1600-1608,1991

The glucocorticoid cortisol has been implicated in may elevate the biologic plasma glucocorticoid activ- the pathogenesis of ocular hypertension (OH) and pri- ity beyond that expected on the basis of the total mary open-angle glaucoma (OAG).1'2 Several clinical plasma cortisol level. This elevated glucocorticoid ac- studies show that pharmacologic levels of glucocorti- tivity could be present at normal or only slightly ele- coids, given either topically to the eye or systemically, vated total plasma cortisol levels and would go unde- can raise intraocular pressure.3"9 This elevation in in- tected with a standard cortisol radioimmunoassay traocular pressure is often accompanied by visual (RIA). We wanted to determine whether, in compari- field loss and optic disc changes characteristic of son to NOR, patients with OH and OAG have plasma OAG.10 Similarly, patients with Cushing's syndrome, glucocorticoid activity in excess of that attributable to also associated with an elevation of glucocorticoids, cortisol. especially plasma cortisol, often have OAG.11"13 Compared with normal individuals (NOR), patients Materials and Methods with OH or OAG have significantly elevated levels of total plasma cortisol1'214"17 and greater amounts and Subjects higher percentages of free plasma cortisol.18 Both OH and OAG may be associated with gluco- We studied NOR, OH, and OAG populations. All corticoids other than cortisol. These glucocorticoids subjects underwent a complete ocular examination, including measurement of intraocular pressures with the Goldmann applanation tonometer on at least two From the Department of Ophthalmology, New England Medical Centre Hospitals and Tufts University School of Medicine, Boston, separate occasions, slit-lamp gonioscopy, tonography, Massachusetts. determination of visual fields with the Goldman pe- * Currently working for Alcon Laboratories, Inc., Fort Worth, rimeter by kinetic and static means, and stereopho- Texas. tography of the optic disc. Supported in part by grants from the National Eye Institute of the The NOR consisted primarily of volunteers from National Institutes of Health, Bethesda, Maryland for research (EY00024) and for training (EY07045) and from Research to Pre- senior-citizen groups who had ocular pressures less vent Blindness, Inc., New York. than 21 mm Hg with normal visual fields and optic Presented in part at the meeting of the Association of Research in discs on at least two occasions. The OH patients had Vision and Ophthalmology, May 6, 1980, at Sarasota, Florida. normal visual fields but had ocular pressures deter- Submitted for publication: November 1, 1989; accepted De- mined on two or more independent occasions to be cembers, 1989. 21 mm Hg or more. The optic disc may or may not Reprint requests: Bernard Schwartz, MD, PhD, Department of Ophthalmology, New England Medical Centre, 750 Washington have had changes characteristic of glaucoma. The Street, Box 450, Boston, MA 02111. OAG patients had ocular pressures of 21 mm Hg or

1600

Downloaded from iovs.arvojournals.org on 09/25/2021 No. 5 INCREASED PLASMA NONCORTISOL IN OPEN-ANGLE GLAUCOMA / McCorry and Schwartz 1601

more, and visual field and optic disc changes charac- second visit. They were centrifuged immediately at teristic of glaucoma. room temperature for 15 min. An aliquot of the All OAG subjects were examined with slit-lamp plasma was collected to determine the total cortisol. gonioscopy and categorized into: (1) those with in- Free cortisol was separated as described later, and creased pigmentation graded as 2+ or more in all an- then the sample and the plasma sample for total corti- gles, (2) those with the exfoliation syndrome,19 and (3) sol were frozen at -20°C. those with neither pigmentary changes in the angle nor exfoliation syndrome. Subjects in the first two Chemical Analysis categories were excluded from this study. Fifty-five of For the receptor-based competitive assay, 8-week- the 85 subjects had been evaluated for plasma free- old male New Zealand white rabbits weighing about 18 cortisol levels in a previous study. 1.6 kg were used as a source of ocular tissue. This On the day of the study, ocular pressure was mea- source was chosen both because the rabbit eye is sensi- sured after a blood sample was drawn. The values for tive to the ocular hypertensive effects of glucocorti- maximum applanation pressure were obtained from coids22"25 and because specific glucocorticoid recep- the patient's record. Areas of cupping and pallor of the tors are present in this tissue.26-27 Preparations of rab- optic discs, expressed as the percent area relative to the bit iris-ciliary body glucocorticoid receptors were area of the optic disc, were measured using a semi- made using the method previously described by quantitative technique from disc photographs made McCarty and Schwartz.27 with a Zeiss fundus camera (Carl Zeiss, Inc., Oberko- The Glucocorticoid Receptor Competitive Binding chen, Germany) and from stereophotographs taken Assay (RRA) was used for the determination of total with a Donaldson stereocamera.2021 Low-speed color plasma glucocorticoid activity. The method used was slide film (Kodachrome 25; Kodak, Rochester, NY) an adaptation of the method originally published by was used for all photographs. For each subject, blood Ballard and co-workers,28 who used rat hepatoma glu- pressure was measured in one arm on the morning of cocorticoid receptors to measure plasma glucocorti- blood sampling, with the subject in a seated position, coid activity in both rabbits and humans. with the standard mercury sphygmomanometer tech- Steroid-free plasma prepared from pooled plasma nique. from NOR was stirred at room temperature with 50 Subjects receiving phenytoin sodium, barbiturates, mg/ml of Norit A charcoal (Amend Drug and Chemi- cyproheptadine, or any form of steroid medication, cal Co., New York, NY) for 1 hr. The charcoal was including estrogens, progesterones, and antimineralo- removed by centrifugation at 25,000 X g for 15 min. corticoids (such as spironolactone) were excluded The supernatant was treated with the same procedure from the study. Seventeen subjects were receiving twice more. This procedure has been used successfully medication for systemic vascular hypertension, of by previous investigators to deplete plasma of ste- whom two were taking systemic beta-adrenergic roids.28 A cortisol RIA on the steroid-free plasma did blockers (propranolol hydrochloride). Only two sub- not detect the presence of cortisol. jects had a diagnosis of diabetes mellitus. All but 1 of Corticosteroid extracts were prepared both from 25 OAG patients and only 12 of 31 OH patients were standards and from plasma obtained from subjects. receiving medication for elevated ocular pressure. For standards, concentrations of cortisol of 2.5-40 ng Twenty subjects were receiving the beta-adrenergic in ethanol were dispensed in triplicate into clean blocker, timolol, as drops, 16 were receiving epineph- tubes. The ethanol vehicle was evaporated with N2. rine drops, 11 were receiving a carbonic anhydrase One hundred microliters of steroid-free plasma was inhibitor, 9 were receiving demecarium bromide placed into each tube, mixed, and incubated at room drops, and 27 were receiving pilocarpine drops, either temperature for 15 min. After a second mixing, 1.0 ml alone or in combination. of methylene chloride was added to all tubes, and All subjects gave informed consent before partici- each sample was extracted by vortexing for 20 sec at pating in the study. Approval for the investigation was room temperature. The aqueous phase was aspirated, obtained from the Institutional Human Investigation and 0.5 ml of the methylene chloride extract was Review Committee of the New England Medical transferred to a clean, labeled tube. This produced Center, Boston. standards of 1.25-20 ng. Using 3H-cortisol as a tracer showed that this method had a consistent recovery of Blood Samples >95%. All samples of blood for analysis for both total The analyses of samples from subjects were done plasma cortisol and free cortisol were taken between without prior knowledge of patient diagnosis. Three 9:00 AM and 11:00 AM on an outpatient visit. The 100-JUI aliquots of the samples were extracted at the samples were drawn only during or after a subject's same time and in the same way as the standards. The

Downloaded from iovs.arvojournals.org on 09/25/2021 1602 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / April 1991 Vol. 32

portion of the extract analyzed contained corticoste- whether the diagnostic groups differed in composition roids equivalent to 50 /xl of plasma. with respect to sex, race, season of the year of patient For the analysis of plasma from subjects, the ex- visits, and frequency of occurrence of various sys- tracts from standards and from the subjects were temic diseases, such as vascular hypertension and dia- dried under N2. Radioactive dexamethasone was then betes mellitus. A diagnosis of vascular hypertension added to all tubes so that, when it was redissolved in was presumed if the subject reported use of prescrip- 200 y\, the final concentration was 10 nM. The eth- tion medication for vascular hypertension. Diagnos- anol vehicle was dried with N2, and the tubes were tic groups were analyzed for differences in distribu- placed on ice. Two hundred microliters of rabbit iris- tion in age and systemic blood pressure using the ciliary body glucocorticoid receptor preparation (cy- Kruskal-Wallis test. tosol) was then added to all samples. All samples were The latter test was used also to compare the distri- vortexed at 15-min intervals three times, then incu- butions of RRA, RIA, and the RRA/RIA ratio of the bated at 0°C for 16-18 hr. After incubation, the re- diagnostic groups. When only two groups were being 3 ceptor-bound and unbound H-dexamethasone were compared, such as patients who were and were not separated with hydroxyapatite as previously de- being treated with a beta-adrenergic blocker (both ocu- 27 scribed. lar and systemic), the Mann-Whitney U test was used A standard curve was developed by making a ratio to compare RRA with RIA and the RRA/RIA ratios. of the amount of 3H-dexamethasone bound by the Spearman correlation coefficients were computed cytosol in the absence of a competing steroid to the and used to test for statistical independence between amount of 3H-dexamethasone bound in the presence RIA and RRA. Two-tailed Wilcoxon signed-rank of increasing quantities of unlabeled cortisol and plot- tests were done in each of the three diagnostic groups ting it against the cortisol concentration in the stan- to determine whether the frequency distributions of dards. The glucocorticoid activity of the subjects' sam- RRA differed significantly from RIA. For compari- ples was determined by using the quantity of 3H-dexa- son of correlation coefficients, the Fisher's Z transfor- methasone bound in the samples as the denominator mation was used.32 in this ratio. The total receptor-determined glucocor- Multivariate: Multiple linear regression3334 was ticoid activity (RRA) in the "cortisol equivalents" of used to examine the relationship between the experi- the unknown was then read by extrapolation from the mental findings and ocular and systemic therapy in standard curve. the presence of other, potentially influential nondiag- Total plasma cortisol was measured by RIA using a nostic variables. The RRA, RIA, and RRA/RIA ra- method outlined by Niswender et al29 and described tios were the dependent variables of interest. We previously.18 Briefly, tenfold diluted plasma was incu- transformed the data to logarithmic values to normal- bated at 60°C for 30 min to inactivate all binding ize them. proteins. Triplicate aliquots of the heat-treated Indicator variables such as meds (1) and meds (2) plasma were incubated with cortisol antiserum and were created for the purpose of identifying the differ- 3H-cortisol overnight at room temperature. The anti- ent categories of medication use, ie, no ocular medica- serum-bound cortisol was separated from the un- tion or beta-blocker use [meds (1) = 0, meds (2) = 0], bound cortisol by precipitation with ammonium sul- use of ocular medication other than timolol or deme- fate. The 3H-cortisol in the supernatant was measured carium bromide [meds (1) = 0, meds (2) = 1], and by liquid scintillation spectrometry. Quantitation of beta-blocker and/or demecarium bromide use [meds unknown samples was achieved by comparing the dis- (1)= l,meds(2) = 0]. placement of 3H-cortisol from antiserum to that dis- The univariate tests established that there were no placed by known quantities of added cortisol. The re- significant differences between OH and OAG pa- sults are expressed as micrograms/100 ml and repre- tients. Consequently, for the multiple linear-regres- sent the total cortisol. sion analysis, they were combined into one group, and only one indicator variable was needed for the Statistical Methods purpose of distinguishing ocular diagnosis (ie, the diagnosis was coded as: normal, 0 and OH or OAG, Univariate: Univariate analysis of the data was 1). The other variables included in the models be- done using two-tailed nonparametric statistical pro- cause of their potentially influential relationships cedures30'31 with a 0.05 level of significance. Nonpara- with the dependent variables included the logarithm metric procedures were chosen for the univariate anal- of total plasma cortisol (which was mean centered), ysis because they require few assumptions about the age (in yr), sex (female as 1 and male as 2), systolic (or underlying distribution from which the data were ob- diastolic) blood pressure (in mm Hg), and time of tained. Chi-square tests were done to ascertain blood sampling (measured in hours from the previous

Downloaded from iovs.arvojournals.org on 09/25/2021 No. 5 INCREASED PLASMA NONCOfYTISOL IN OPEN-ANGLE GLAUCOMA / McCorry ond Schworrz 1603

Table 1. Characteristics of study population: systemic and ocular vanables Normal Ocular hypertension One-angle glaucoma / Percentile Percentile Percentile Variable* n Median (30th, 70th) n Median (30th, 70th) n Median (30th, 70th)

Age (yr) 29 65.0 (60.5, 72.5) 31 65.0 (60.0, 70.5) 25 72.0 (66.5, 75.5) Sex/male 29 12 31 12 25 12 Race/white 29 29 31 28 25 19 Intraocular pressure (mm Hg) at time of blood sample 29 15.0 (13.3, 15.5) 31 21.0 (18.0,22.3) 24 17.8 (15.0, 18.5) Maximum intraocular pressure (mm Hg) 29 16.0 (14.0, 17.0) 31 26.0 (24.0, 34.8) 25 27.5 (22.3, 28.5) Systolic blood pressure 26 130 (125, 145) 31 140 (120, 140) 23 140 (130, 140) Diastolic blood pressure (mm Hg) 26 80 (74, 80) 31 80 (70, 80) 23 80 (78,81) Optic disc percent cuppingf 29 42.5 (40.0, 50.0) 31 55.0 (47.5,61.8) 23 70.0 (56.8, 73.8) Percent pallorft 29 6.0 (3.8, 8.0) 30 19.0 (8.5, 25.0) 23 47.5 (31.8,55.0)

*Ocular variables are expressed as average of both eyes, t Percent cupping = area of cup/area of optic disc X 100. tt Percent pallor = area of pallor/area of optic disc X 100.

midnight). A term representing the interaction of to- tients. There were no significant differences between tal plasma cortisol and diagnosis was also included in the groups with respect to sex or systemic blood pres- the models because of differences in the relationship sure. There was a significant difference with respect to between total cortisol and each of the other dependent race between the three groups (chi-square = 8.211, variables (especially percent free cortisol) for the dif- P = 0.02) with the OAG having more blacks than ferent diagnosis groups. the OH. Comparison of optic disc cupping and pallor mea- Results surements by the Kruskal-Wallis test showed a significant difference in the distribution of these variables Univariate Analyses among the NOR, OH, and OAG groups. Multiple Comparisons for systemic and ocular characteris- comparison testing indicated that the NOR were sig- tics of subjects: Plasma samples were obtained from nificantly different from the OH group for cupping 29 NOR and 31 OH and 25 OAG patients. The char- and pallor (cupping, P = 0.001; pallor, P = 0.000). acteristics of the three groups with respect to systemic Similarly the NOR were significantly different from and ocular parameters and plasma cortisol parame- the OAG group (cupping, P = 0.000; pallor, P ters are shown in Table 1. Nonparametric multiple = 0.000), and also the OH group was significantly dif- comparison testing revealed no significant differences ferent from the OAG group (cupping, P = 0.002; pal- in the frequency distributions of age between the lor, P = 0.000). NOR and OH patients, or between the OH and the Comparisons for diagnosis: Comparisons of corti- OAG patients, or between the NOR and OAG pa- sol parameters are shown in Table 2 and Figure 1.

Table 2. Characteristics of study population experimental variable

Normal Ocular hypertension Open-Angle Glaucoma Percentile Percentile Percentile Variable n Median (30th, 70th) n Median (30th, 70th) Change* n Median (30th, 70th) Change* Radioimmunoassay-determined cortisol (RIA)(/ig/100 ml) 29 9.5 (8.8,11.8) 31 11.3 (10.3,13.6) 19.0 25 11.9 (10.0,13.6) 25.3 Radioreceptor-determined glucocorticoids (RRA) (Mg/100ml) 29 11.6 (7.6,13.5) 31 12.4 (9.6,15.8) 6.9 25 15.9 (12.4,17.7) 37.1 RRA T7T~ 29 °-97 (0-71,1.27) 31 1.02 (0.77,1.33) 5.2 25 1.10 (0.95,1.55) 13.4

* Percent change from normal group.

Downloaded from iovs.arvojournals.org on 09/25/2021 1604 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / April 1991 Vol. 32

45.0-1 RRA with RIA using the two-tailed Wilcoxon signed- • Median rank test showed no significant difference for NOR or OH but a significant difference for OAG patients (P = 0.0264). This indicated that cortisol could account for virtually all of the receptor-determined glucocorti- 35.0- coid activity among NOR and OH patients (Table 2) but not in the OAG group; there is a significant amount of noncortisol glucocorticoid activity in the plasma of OAG patients. Comparisons for beta-blocker and other therapy for glaucoma: The OAG and OH patients were categorized according to their use of certain medications that could be readily absorbed systemically and thus possibly affect the hypothalamic-pituitary-adrenal axis. The medications of interest were beta-blocker 15.0- medications (timolol and/or propranolol hydrochloride), epinephrine, carbonic anhydrase inhibitors, and demecarium bromide. The Mann-Whitney U test showed a significant difference in the distribution of RRA (P = 0.040) between those OAG and OH 5.0- patients who were receiving topical beta-blocker medication (mean = 18.5 mg/100 ml, n = 20) and those who were not (mean = 13.5 mg/100 ml, n = 36), but no significant differences were observed in the distri- NORMALS OH OAG n = 29 n = 31 n=25 bution of RIA or the ratio RRA/RIA. A borderline significant difference was observed using the Mann- DIAGNOSIS Whitney U test in the distribution of RRA (P = 0.054) Fig. .1. Scatter plot of RRA value radioreceptor-determined glucocorticoid activity (in cortisol equivalents ^g/100 ml) for normals, between those OAG and OH patients who were re- ocular hypertensives (OH), and open-angle glaucomas (OAG). ceiving demecarium bromide (mean = 20.3 mg/100 ml, n = 9) versus those who were not (mean =14.3 mg/100 ml, n = 47). No significant differences be- Although the median values of RIA cortisol were the tween demecarium bromide users versus nonusers least for the NOR group (19.0% greater for OH and were found for RIA and the ratio RRA/RIA. 25.3% greater for OAG), the Kruskal-Wallis test No significant differences were observed using the showed no significant difference in the distributions Mann-Whitney U test for RIA, RRA, or RRA/RIA of this variable. For both the RRA and the RRA/RIA between those OAG and OH patients who were (n ratios, the medians were smallest for the NOR group = 16) versus those who were not (n = 40) receiving and greatest for the OAG group (For RRA: OH was epinephrine drops, between those OAG and OH sub- 6.9% greater and OAG was 37.1% greater than NOR; jects who were (n = 11) versus those who were not (n for RRA/RIA: OH was 5.2% greater and OAG was = 45) receiving carbonic anhydrase inhibitors, and 13.4% greater than NOR). The Kruskal-Wallis test between those who were (n = 22) versus those who showed a significant difference (H = 8.94, P = 0.0115) were not (n = 34) receiving a beta-blocker drug (topi- in the distribution of RRA among the three groups, cal and systemic). Similarly no significant differences but this was not significant for the distribution of the were seen between those who were (n = 27) versus ratio RRA/RIA. Multiple comparison testing re- those who were not (n = 29) receiving pilocarpine vealed that the major difference was between NOR drops. and OAG patients (P = 0.0025, by Mann-Whitney U Correlations between RIA and RRA: The relation- test) and between OH and OAG groups (P = 0.0543, ship between RIA and RRA values are shown in Ta- by Mann-Whitney U test). A significant difference ble 3 and Figure 2. Highly significant positive correla- was also found with the Mann-Whitney U test for the tion coefficients for the total population and the OAG ratio RRA/RIA between NOR and OAG groups (P group were found, ie, as the amount of RIA increased, = 0.0441) and a borderline significant value between the amount of RRA also increased. The correlation OH and OAG groups (P = 0.0760). coefficient was also significant for the OH group but A comparison of the frequency distributions of was of borderline significance in the NOR group.

Downloaded from iovs.arvojournals.org on 09/25/2021 No. 5 INCREASED PLASMA NONCOPJISOL IN OPEN-ANGLE GLAUCOMA / McCarry and Schwarrz 1605

Normals Ocular Hypertensives 30 r, . 0.3458 f,. 0.3802 1 p". .07 p « .09 I n * 29 25 • E 26 • o ~ 1/10

20 • 2 20 • © _ - - qui\ © - 15 " • • "5 15 • .2 - O1 _

10 10 •

5. • 5. •

. • • • • • • • • • • •• 6 8 1 12 16 18 20 M 24 26 10 14 18 22 A - - ° B 8. 12 16 20 24 . RIA (Cortisol/*g/100ml) RIA (Cortisol fig/ 100ml)

.+ +. 45 • Open Angle + Glaucomas r,< 0.5503 1

40 • p • .004 + It • 26 I

35 •

I 30 T I z J 25 •

20 •

10 • •

15 21 27 12 18 24 RIA (Cortisol fig/ 100ml) Fig. 2. Correlation between RRA (radioreceptor-determined glucocorticoid activity in cortisol equivalents /xg/100 ml) and RIA (radioimmunoassay cortisol levels jtg/100 ml) for (A^ normals, (B) ocular hypertensives, and (C) open-angle glaucomas.

These correlations were not significantly different be- temic parameters with RIA, RRA, and RRA/RIA val- tween the NOR and OAG groups or between the OH ues such as age, sex, and diastolic blood pressure with and OAG groups (Fisher's Z-transformation test). the diagnostic groups and used the whole population. The second evaluated the effect of therapy on the sys- Multivariate Analysis temic variables and types of medications in the OH The multiple-regression analyses were divided into and OAG groups, since these groups were the only two parts. The first considered the association of sys- ones receiving medications to lower ocular pressure.

Downloaded from iovs.arvojournals.org on 09/25/2021 1606 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / April 1991 Vol. 32

Table 3. Spearman correlation coefficients of carbonic anhydrase inhibitor versus no medication), glucocorticoids determined by radioreceptor and meds (1-2) (ie, timolol drops and/or demecar- competitive binding assay (RRA) vs. cortisol ium bromide drops versus epinephrine drops, pilo- determined by radioimmunoassay (RIA) carpine drops, and a carbonic anhydrase inhibitor). No significant models could be found for RRA, RIA, Correlation coefficient P and RRA/RIA, nor were any of the independent variables significant. Total population (n = 85) 0.4568 <0.0001 Normal subjects Analysis of Other Factors (n = 29) 0.3458 0.0661 Ocular hypertensive patients Finally, additional characteristics of the subjects (n = 31) 0.3862 0.0319 were analyzed to determine other factors which could Open-angle glaucoma patients influence RRA and RRA/RIA and contribute to the (n = 25) 0.0037 0.5593 differences found between the diagnostic groups. There was a difference of borderline significance (Pearson's chi square = 0.064) in the prevalence of Table 4 shows the significant model obtained for systemic hypertension in the OH and OAG groups the first analysis. The diagnosis variable (NOR versus compared with the NOR group, with the OH and OH plus OAG) was significantly associated with the OAG groups having a greater prevalence of systemic RRA value and diastolic pressure. For the model us- hypertension. There were no significant differences ing RIA as the dependent variable, age and sex were between the groups for distribution of subjects' visits significantly associated with the RIA level. For the categorized by season of the year. model with the ratio RRA/RIA as the dependent vari- Discussion able, only diastolic pressure was significantly associated. We used the approach employed by Ballard and Since timolol and demecarium bromide therapy colleagues28 to measure the combined activities of were found to be associated with a significant increase cortisol and other glucocorticoids because glucocorti- in RRA values, an analysis was done to determine coid receptors, unlike the specific cortisol antibodies whether other medications are associated with used in the RIA, will accept both cortisol and other changes in RRA, RIA, or RRA/RIA. A multiple-re- biologically active glucocorticoids. To compare the gression model was constructed for the OH and findings from the RRA with those from the cortisol OAG groups with the following independent vari- RIA, the former were determined in cortisol equiva- ables: diagnosis (OH versus OAG), age, sex, diastolic lents, ie, the amount of glucocorticoid activity if corti- blood pressure, meds (1) (ie, timolol drops and/or sol were the only glucocorticoid present in plasma. By demecarium bromide drops versus no medication), comparing the RRA value in cortisol equivalents with meds (2) (ie, epinephrine drops, pilocarpine drops, the amount of cortisol present in plasma (measured

Table 4. Multiple regression models for total population

Standardized Dependent variable Independent variables coefficient Coefficient

LN(RRA) 2.715 0.0361 80 0.1265 Diagnosis* 0.26 0.2854 2.37 0.02 (intercept = 1.1532) (normal vs OH plus OAG) Age 0.05 0.0024 0.43 0.67 Sexf 0.03 0.0336 0.29 .0.77 Diastolic pressure 0.25 0.0119 2.26 0.03 LN(RIA) 3.762 0.0076 80 0.1671 Diagnosis* 0.18 0.1342 1.72 0.09 (intercept = 2.5034) (normal vs OH plus OAG) Age 0.22 -0.0075 -2.07 0.04 Sexf 0.28 0.1952 2.65 0.01 Diastolic pressure 0.03 0.0009 0.25 0.80 LN (ratio RRA/RIA) 2.888 0.0279 80 0.1335 Diagnosis* 0.14 0.1526 1.32 0.19 (intercept = -1.3448) (normal vs OH plus OAG) Age 0.20 0.0098 1.83 0.07 Sex| 0.16 -0.1621 -1.49 0.14 Diastolic pressure 0.24 0.0111 2.20 0.03

LN = natural logarithm; RRA = radioimmunoassay-determined cortisol * Diagnosis: 0 = .normal; 1 = OH plus OAG. levels (ng/100 ml); RIA = radioreceptor-determined glucocorticoid activity t Sex: 1 = female; 2 = male. (in cortisol equivalent, ng/100 ml).

Downloaded from iovs.arvojournals.org on 09/25/2021 No. 5 INCREASED PLASMA NONCOIYTISOL IN OPEN-ANGLE GLAUCOMA / McCorry and Schwartz 1607

by cortisol RIA), it was possible to determine whether ences between blacks and whites should not be consid- appreciable amounts of glucocorticoid activity not at- ered of significance until a further study is done with a tributable to cortisol were present. Nevertheless, the larger sample size of blacks in comparison to whites. method measures only noncortisol glucocorticoid ac- It appears that OAG, but not OH patients, have tivity, not the actual quantities of these hormones. increased amounts of nonplasma cortisol glucocorti- Since cortisol is by far the most biologically active coid in their plasma compared with NOR (37% glucocorticoid in humans, other active glucocorti- greater in OAG than in NOR groups). Furthermore, coids, such as corticosterone or 11-deoxycortisol, the multiple-regression models indicate that the in- would have to be present in amounts abnormally high crease of the plasma noncortisol glucocorticoid is as- for a human to produce a significant discrepancy be- sociated with diagnosis of glaucoma and tends to in- tween the quantity of cortisol equivalent glucocorti- crease with increased diastolic blood pressure, but it is coid activity determined by the RRA and the quantity not associated with ocular medication for glaucoma of cortisol determined by RIA.35 Although the iden- including topical and systemic beta-adrenergic tity of possible active glucocorticoids responsible for blockers. Our observations of increased RRA values such an observation cannot be determined by the for OAG patients compared with NOR support methods in this report, the data suggest that such a previous studies in which OAG and OH was found to phenomenon may exist in primary OAG compared be associated with elevated plasma cortisol.1'214"18 with NOR and OH. The increase in RRA is also associated with a signifi- We confined our study to subjects who had no evi- cant increase in RIA in OH and OAG groups but a dence of the pigmentary dispersion syndrome and/or borderline significant increase in NOR (Table 3). exfoliation. This decision was based on our previous Therefore, there may be hypothalamic-pituitary-adre- study which showed statistically significant greater nal abnormalities in OH and OAG patients which differences with NOR when such subjects with pig- cause them to produce significant amounts of other mentary dispersion and/or exfoliation were excluded glucocorticoids in addition to cortisol. In diseases re- from the analysis.18 sulting in severe adrenal abnormalities, such as Cush- Our results indicate that when compared with ing's syndrome36 and hyperaldosteronism,37 gluco- NOR or OH,,there were significant differences in the corticoids other than cortisol and aldosterone are pro- RRA and the RRA/RIA ratios of the OAG group. duced. The observation that OH patients are not as These differences were not significant for OH versus significantly associated with increased RRA values as NOR and were borderline in significance between are OAG patients may result from the greater severity OH and OAG. Furthermore the mean RRA/RIA ra- of disease, or that OH may be a mixed group of pa- tio was found to be significantly greater than unity for tients, some of whom are essentially normal and the jOAG (with RRA significantly greater than RIA others of whom are in the early stages of glaucoma. for OAG). With multiple-regression analysis, a signifi- The identification of the other plasma noncortisol cant difference of RRA was also observed between the glucocorticoids would be important in determining NOR versus the OH plus OAG groups. the nature of any hypothalamic-pituitary-adrenal ab- The subjects in this study consisted of 29 whites and normality in glaucoma. no blacks in the NOR group, 28 whites and 3 blacks in the OH group, and 25 whites and 6 blacks in the OAG Key words: glucocorticoid, receptor, glaucoma, ocular hy- group. When the data for RIA, RRA, and RRA/RIA pertension, plasma glucocorticoids -values were categorized by race, there was no trend for the OH for the blacks being different from the whites Acknowledgments (median values: RIA whites, 12.3; RIA blacks, 10.7; The authors thank Alice Wysocki, BS, for performing the RRA whites, 13.5; RRA blacks, 15.4; RRA/RIA chemical analytic determinations, Mark Manning, BS, for whites, 1.18; RRA/RIA blacks 1.09). However, for statistical analysis, and Sarah Boardman, BA, for editorial the OAG, the blacks appeared to have lower values assistance. than the whites (median values [30th and 70th per- centiles]: RIA whites, 13.7 [9.4, 16.3]; RIA blacks, References 11.0 [10.6, 11.3]; RRA whites, 18.6 [14.0, 19.0]; RRA 1. Schwartz B and Levene RZ: Plasma cortisol difference between blacks, 13.0 [10.4, 15.6]; RRA/RIA whites, 1.45 normal and glaucomatous patients before and after dexametha- [0.95, 1.79]; RRA/RIA blacks, 1.09 [0.95, 1.55]). The sone suppression. Arch Ophthalmol 87:369, 1972. lower values for blacks tend to minimize the differ- 2. Schwartz B and Seddon J: Increased plasma cortisol in ocular hypertension. Arch Ophthalmol 99:1791, 1981. ences between whites among the NOR, OH, and 3. Bernstein NH and Schwartz B: Effects of long-term systemic OAG groups. However, the number of blacks in the steroids on ocular pressures and tonographic values. Arch Oph- OAG was only six, and any interpretation of differ- thalmol 68:742, 1961.

Downloaded from iovs.arvojournals.org on 09/25/2021 1608 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / April 1991 Vol. 32

4. Armaly MF: Effect of corticosteroids on intraocular pressure metric visual field loss in glaucoma. In Proceedings of 22nd and fluid dynamics: I. The effect of dexamethasone in the nor- International Congress of Ophthalmology, Vol 2. Paris, Mas- mal eye. Arch Ophthalmol 70:482, 1963. son, 1974, pp. 633-638. 5. Becker B and Mills DW: Corticosteroids and intraocular pres- 21. Donaldson DD, Prescott R, and Kennedy S: Simultaneous ste- sure. Arch Ophthalmol 70:505, 1963. reoscopic fundus camera incorporating a single axis. Invest 6. Schwartz B: The response of ocular pressure to corticosteroids. Ophthalmol Vis Sci 19:289, 1980. In International Ophthalmology Clinics: Corticosteroids and 22. Jackson RT and Waitzman MB: Effect of some steroids on the Eye, Schwartz B, editor. Boston, Little, Brown, 1966, pp. aqueous humor dynamics. Exp Eye Res 4:112, 1965. 929-989. 23. Wood DC, Contaxis I, Sweet D, Smith JC, and Van Dolah J: 7. Godel V, Feilder-Ofry V, and Stein R: Systemic steroids and Response of rabbits to corticosteroids I. Am J Ophthalmol ocular fluid dynamics I. Acta Ophthalmol (Copenh) 50:656, 63:841, 1967. 1972. 24. Lorenzetti OJ: Effects of corticosteroids on ocular dynamics in 8. Kimura R and Mackawa N: Effect of orally-administered hy- rabbits. J Pharmacol Exp Ther 175:763, 1970. drocortisone in the ocular tension in primary open angle glau- 25. Levene RA, Rothberger M, and Rosenberg S: Corticosteroid coma subjects. Acta Ophthalmol (Copenh) 54:430, 1976. glaucoma in the rabbit. Am J Ophthalmol 78:505, 1974. 9. Weinreb RN, Polansky JR, Kramer SG, and Baxter JD: Acute 26. Weinstein BI, Altman K, Gordon GG, Dunn M, and Southern effects of dexamathasone on intraocular pressure in glaucoma. AL: Specific glucocorticoid receptor in the iris-ciliary body of Invest Ophthalmol Vis Sci 26:170, 1985. the rabbit. Invest Ophthalmol 16:973, 1977. 10. Goldmann H: Cortisone glaucoma. In International Ophthal- 27. McCarty GR and Schwartz B: Increased concentration of glu- mology Clinics: Corticosteroids and the Eye, Schwartz B, edi- cocorticoid receptors in rabbit iris-ciliary body compared to tor. Boston, Little, Brown, 1966, pp. 991-1003. liver. Invest Ophthalmol Vis Sci 23:525, 1982. 11. Tartar J: Glukom bei Morbus Cushing. Graefes Arch Clin Exp 28. Ballard PL, Carter JP, Graham BS, and Baxter JD: A radiore- Ophthalmol 139:793, 1938. ceptor assay for evaluation of the plasma glucocorticoid activity of natural and synthetic steroids in man. J Clin Endocrinol 12. Bayer IM and Neuner HP: Cushing-Syndrom und erhohter Metab 41:290, 1975. Augeninnendruck. Wiener Duetsche Medizinische Wo- 29. Niswender GD, Akbar AM, and Nett TM: Radioimmunoassay chenschrift 92:1791, 1967. procedures for quantification of steroid hormones. In Hor- 13. Neuner HP and Bayer JM: Das Verhalten des Augeninnen- mones in Human Blood: Detection and Assay, Antoniades druckes Sowie des Brachialis-und des Opthalmica-Blut- HN, editor. Cambridge, MA, Harvard University Press, 1976, Druckes beim Cushing-Syndrome vor und nach operativen pp. 751-776. Eingriffen an den Nebennieren: Ein Statistischer Vergleich. 30. Siegel S: Nonparametric Statistics for the Behavioral Sciences. Doc Ophthalmol 27:221, 1969. New York, McGraw-Hill, 1956, pp. 116-127. 14. Khasanov N and Pomoshchnikov NI: Glucocorticoid function 31. Hollander M and Wolfe DA: Nonparametric Statistical Meth- of the adrenal cortex in glaucoma patients. Vestn Oftalmol 2:34,1971. ods. New York, John Wiley and Sons, 1973. 15. Weinstein P and Mori E: Plasma-Cortisol und Glaukom. Klin 32. Morrison D, editor: Multivariate Statistical Methods. New Monatsbl Augenheilkd 161:277, 1972. York, McGraw-Hill, 1976. p 104. 16. Rosenberg S and Levene R: Suppression of plasma cortisol in 33. KJeinbaum DG and Kupper LL: Apphcu regression Analysis normal and glaucomatous patients. Arch Ophthalmol 92:6, and Other Multivariable Methods. North Scituate, MA, Dux- 1974. bury Press, 1978. 17. Freedman HJ, David R, Vander Walt L, and Luntz MH: 34. Draper NR and Smith H: Applied Regression Analysis. New Plasma cortisol suppression response in the South African York, John Wiley and Sons, 1981. black population with glaucoma. Br J Ophthalmol 60:786, 35. Rousseau GC, Baxter JD, and Tompkins GM: Glucocorticoid 1976. receptors: Relations between steroid binding and biological ef- 18. Schwartz B, McCarty G, and Rosner B: Increased plasma free fects. J Mol Biol 67:99, 1972. cortisol in ocular hypertension and open-angle glaucoma. Arch 36. Cope CL: Adrenal Steroids and Disease. Philadelphia, JB Lip- Ophthalmol 105:1060, 1987. pincott, 1972, pp. 313-336. 19. Roth N and Epstein BL: Exfoliation syndrome. Am J Ophthal- 37. Guthrie GP: Multiple plasma steroid responses to graded mol 89:477, 1980. ACTH infusions in patients with primary aldosteronism. J Lab 20. Schwartz B: Correlation of pallor of the optic disc with asym- Clin Med 98:364, 1981.

Downloaded from iovs.arvojournals.org on 09/25/2021