Psychoneuroendocrinology,Vol. 17, No. 5, pp. 485-495,1992
0306-4530/92 $5.00+0.00 01992 Persmnon Press ltd.
Printed in Great Brit~dn
EFFECTS OF ESTROGEN O N MEMORY FUNCTION IN SURGICALLY MENOPAUSAL W O M E N SUSANA M. PHILLIPS1 and BARBARA B. SHERWIN1,2 1Department of Psychology, Mt~Jill University, and 2Department of Obstetrics and Gynecology, Sir Mortimer B. Davis-Jewish General Hospital Montr6al, Qu6bec, Canada (Received 18 April 1991; infinalform 5 December 1991)
SUMMARY The effects of estrogen (E) on memory were assessed in 19 women who required a hysterectomy and bilateral oophorectomy for benign disease. Blood samples were drawn and memory tests were administered before surgery and again after 2 mo of postoperative treatment consisting of either monthly E or placebo (PL) injections. Scores on the immediate and delayed recall of paired-associates stayed at the same level in E-I~'eatedwomen, whereas they decreased significantly pre- to postoperatively in the PL-treated subjects. In the immediate recall of paragraphs, the scores of those given E improved postoperatively compared to baseline; scores remained unchanged in the PL group. No hormonal effects were apparent on the immediate or delayed recall of visual material, delayed recall of paragraphs, or digit span scores. These findings suggest that variations in specific aspects of memory function may occur in surgically menopausal women coincident with changes in plasma estrone and estradiol levels.
INTRODUCTION COMPLAINTS of memory problems and an inability to concentrate are among the symptoms that have been associated with the menopause, which is marked by dramatic decreases in plasma estradiol (E2) levels. Few studies, however, have adequately examined the association between plasma F_a levels and cognitive functioning during this reproductive phase. In two double-blind crossover studies, Campbell and Whitehead (1977) reported that estrogen (E) replacement therapy (conjugated equine E for 2 or 6 mo) was more effective than placebo (PL) in alleviating many menopause-related symptoms, including hot flushes, anxiety, and insomnia, and was associated with a significant improvement in m e m o r y in peri- and postmenopausal women. Similar results were reported in an uncontrolled study on postmenopausal women who had received estriol succinate for approximately 2 yr (Schneider, 1982). However, in both studies, the memory improvements were based only on subjective self-report data, without any formal objective psychometric tests of memory function. Evidence of improvements in cognitive functioning were provided by other researchers who used various psychometric tests. The performance of 11 postmenopausal women treated with Address correspondence and reprint requests to: Dr. Barbara B. Sherwin, Department of Psychology, McGill University, 1205 Dr. Penfield Avenue, Montrtal, Qutbec, CANADA, H3A IB1. 485
486
S.M. PHILLIPS& B. B. SHF.RWlN
E2-17[3-valerianate for 3 mo was found to be superior to the performance of 10 PL-treated subjects on various cognitive tasks, including a visual search task, a modification of the Stroop test, a choice reaction time task, and tests of concentration and attention (Fedor-Freybergh, 1977). Hackman and Galbraith 0977) also reported an improvement in scores on the Guild Memory Test in nine surgically or naturally menopausal women treated with piperazine estrone (El) sulphate for 6 mo, compared to the performance of nine subjects who received PL. However, when these data were reanalyzed by Sherwin (1988), the between-group differences were not statistically significant. In a double-blind crossover study, Sherwin (1988) reported that surgically menopausal women who received PL for 3 mo following bilateral oophorectomy showed significant decreases in test performance relative to preoperative scores in digit span, immediate recall of paragraphs, and abstract reasoning and response speed and accuracy, whereas those given hormone replacement therapy postoperatively maintained their preoperative levels of performance. Other studies have failed to detect an effect of E on cognitive function assessed with different psychometric tests. Rauramo et al. (1975) tested surgically menopausal women who received 6 mo of either E 2 valerate or PL and found no significant differences between or within the groups on the Integration Memory Test, or on tests of reaction time or logical thinking. It was noted, however, that women who did not receive E complained most often of problems with memory and concentration. Vanhulle and Demol (1976), using tests of concentration, visual memory, digit span, and learning ability, also failed to detect any changes in postmenopausal women following 3 mo of treatment with estriol compared to PL, although there was some evidence of increased attention in the E-treated women. The discrepancies in these results may be due to a number of factors. Various types of psychometric tests were administered across studies, each of which may have measured very different aspects of cognitive functioning. It is possible that some tests tapped a specific area of mental functioning (e.g., visual memory) that may be unaffected by E levels, thereby yielding negative findings, whereas other instruments may have assessed abilities that are indeed sensitive to hormonal changes. Another possible reason for the lack of consensus in findings may be due, in part, to the fact that very different E preparations were administered in various doses, and that in most cases plasma E levels were not measured. In addition, although some studies have generally found a deterioration in mental functioning in postmenopausal women given PL compared to E-treated subjects, none have addressed the possibility that decrements in test scores may be simply an artifact of menopause-related symptoms, such as hot flushes or dysphoric mood, which may have interfered with test performance. The goal of the present study was to determine the specific pattern of memory changes that may occur coincident with changes in plasma E levels in surgically menopausal women, while attempting to control for previous methodological shortcomings. In addition to a battery of objective memory tests, radioimmunoassays of plasma E 1 and E 2 levels, and assessment of symptoms such as depressed mood, anxiety, hot flushes, and sleep disturbances, were carded out. SUBJECTS AND METHODS
Subjects Between February 1988 and July 1990, 31 Caucasian women who required a total abdominal hysterectomy and bilateral salpingo-oophorectomyfor benign disease were recruited. Exclusion criteria included a diagnosis of malignant disease and the presence of known contraindicationsto hormone replacement therapy. Subjects had to be in good general health except for the reasons necessitating the surgery, and they must never have
EFFECTS OF ESTROGEN ON MEMORY F~rNCTION
487
received E replacement therapy. Six women discontinued participation because of scheduling problems, surgical complications, or misgivings about the random assignment procedure. If the subject began experiencing severe menopausal symptoms within a few weeks postoperatively, the double-blind code was broken for ethical reasons, and E2 was administered if she had been receiving PL. Six subjects had to be dropped from the study on this basis, leaving 19 women who completed the study (mean age+ SD=48.2+ 4.7 yr; mean years of scolarity+ SD ---11.4 + 2.4).
Procedure In this 2 (test periods) × 2 (treatments) repeated-measures design, written informed consent was obtained on a form approved by a hospital ethics committee. Subjects were told that they would be participating in a study examining the effects of hormones on mood and memory. They also were informed that they would be receiving postoperatively either E or a preparation with no active hormone. Subjects were randomly assigned preoperatively, in double-blind fashion, to either an E replacement-therapy group (n = 10) or a PL group (n = 9). Test sessions were scheduled approximately 1 - 2 wk prior to surgery (baseline) and approximately 2 me postoperatively. The postoperative test session took place 2 - 4 days after the third monthly injection, the time at which plasma E levels were maximal (Sberwin, 1988). Each session took place at approximately the same time of day, in order to control for possible diurnal variations in alertness. Within each hour-long test session, a blood sample was drawn, followed by an immediate-recall phase in which the subject was asked to recall each individual memory item immediately after presentation. After all the memory tests were presented, a 20- to 30rain delay period ensued, during which the subject completed a series of questionnaires, including those listed below. The delayed-recall phase then followed, during which the subject was asked to recall test items from each of the first three subtests listed below. Drugs The E group received three monthly IM injections of 10 mg E 2 valerate (Delestrogen; Squibb, Canada), whereas the PL group received monthly injections of 1 ml sesame oil. Plasma Hormone Assays Ten milliliters blood were collected at each test session by venipuncture into heparinized Vacutainer tubes. The blood was immediately centrifuged and the plasma stored at -20 ° C until analysis. Plasma F_~and E 1 levels were determined in duplicate by radioimmunoassay. F_,z was measured with a non-extraction, double antibody technique (Garnma-B Direct Oestradiol RIA Kit, IDS Limited, Boldon, U.K.), in which an iodinated F_,z derivative is used. The antiserum is highly specific for ~ . E 1 was analyzed by an ether extraction method, with tritiated E 1 as tracer and an antibody raised to this steroid. Charcoal was used as the adsorbent. Materials Wechsler Memory Scale (WMS~ (Wechsler, 1945; Stone et al., 1946; as modified by Russell, 1975): The two forms of the WMS were counterbalanced across the two test sessions, such that half the subjects received Form I first and half Form 11. The following four subtests were administered: a) Logical Memory/Paragraph Recall: The subject listened to a short paragraph of about four to five sentences and then was asked to verbally recall what she remembered from the passage. Two paragraphs were presented once each. b) Visual Reproduction: The subject was shown a design for 10 sec and then asked to draw what she remembered seeing on the card. Three cards of designs were presented once each. Subjects were given separate sheets of paper for each drawing to minimize visual exposure to designs presented first. c) Associate Learning: The subject listened to a list of 10 word-pairs (paired-associates) presented in random order, six of which were clearly related (easy-associate; e.g., fruit-apple) and four of which had no apparent relationship (hard-associate; e.g., necktie-cracker). After hearing the list, she was asked to recall the word that was paired (e.g., apple) with each cue word (e.g., frui0. Three such trials of presentation and immediate recall were conducted. After the delay period, one final recall trial was administered. In scoring, two points were given for a correctly recalled hard-associate and one point for an easy-associate. d) Digit Span: The subject listened to a series of numbers and was instructed to verbally repeat them in the given order immediately afterwards. In order to ensure approximately identical test conditions across subjects, a manual of exact procedures was
488
S. M. PHILLIPS& B. B. SHERWIN
prepared, verbal test items were presented to subjects via a tape recorder, and verbal responses were recorded on tape and later transcribed. Protocols were scored independently by three coders who were blind to group assignment, Weehsler's (1945) original scoring criteria for the Logical Memory and Visual Reproduction subtests have been criticized as being poorly defined, which may lead to significant scoring discrepancies among coders (Mitchell, 1987). An improved scoring system therefore was developed for this study with welldefined criteria and high inter-rater reliability (r = 0.98 for Logical Memory; r = 0.96 for Visual Reproduction). In addition, alternate-form analyses indicated the two forms were equivalent under these scoring criteria (Phillips et al., unpublished data). Menopausal Index (Blatt et aL, 1953; revised by Sherwin, 1983): This 27-item paper-and-pencil questionnaire was administered pre-and postoperatively in order to monitor the frequency of various menopausal symptoms on scales ranging from 0 (never) to 7 (very often). In addition to an overall menopausal symptom score, this instrument also yields scores for three subscales: somatic, psychosomatic, and psychological symptom clusters. Mulfiole Affect Adjective Check List (MAACL) (Zuckerman & Lubin, 1965): In this brief, self-administered inventory, the subject indicated which of 132 adjectives best reflected her present mood state at the time of each of the test sessions. This instrument yields scores for current levels of anxiety, depression, and hostility, and has been shown to be valid and reliable. It has been used previously in psychoendocrine studies to measure the relationship between affect and concurrent hormone levels (e.g., Sherwin & Gelfand, 1985).
Statistical Analyses Memory, mood, and symptom data were analyzed with the SPSS-X multivariate analysis of variance (MANOVA) program. The study had a repeated-measures design, in which treatment group (E or PL) was the independent factor and time (pre- or postsurgery) was the repeated-measures factor. If a significant multivariate F statistic (p < 0.05) was found for the treatment x time interaction, the Hummel-Sligo procedure was employed to examine each dependent variable with univariate F tests. When a univariate F was significant, post-hoc pairwise comparisons were carried out with Tukey's HSD procedure. RESULTS
Baseline Subject Characteristics Student t-tests were carried out on subject age and education, and on baseline measures of memory, mood, and menopausal symptoms o f the treatment groups. There were no significant group differences on any o f these variables, suggesting that the E and PL groups were comparable on these dimensions before they underwent surgery.
Plasma Hormone Measures There was a significant group x time interaction for both E 1 (F1,17=217, p < 0 . 0 0 1 ) and F_q (F1A7=198, p < 0 . 0 0 1 ) . Preoperatively, mean plasma E 1 and E 2 levels fell within the normal physiological range and did not differ between the E and PL groups. By the second month after surgery, however, the supraphysiological levels of E 1 and F~ in the E group exceeded both their own preoperative levels ( p < 0 . 0 1 ) and the postoperative values o f the PL group ( p < 0 . 0 1 ) . In addition, the PL group experienced a significant decrease in plasma E 1 and F_q levels postoperatively, to values within the postmenopausal range (p < 0.05) (Table I).
Memory Measures There were significant group x time interactions for the f o l l o w i n g m e m o r y measures: i m m e d i a t e p a r a g r a p h recall (Fl.17 = 4.83, p < 0.05), i m m e d i a t e recall o f p a i r e d - a s s o c i a t e s (Fl.17 = 6.55, p < 0.03), and delayed recall of paired-associates (F1,17= 4.67, p < 0.05). Post.hoc pairwise comparisons indicated that the E group showed a significant increase in immediate paragraph recall scores after h o r m o n e replacement therapy c o m p a r e d to their preoperative
489
EFFECTS OF ESTROGENON MEMORY FUNCTION
scores (p < 0.05), whereas the PL group maintained their baseline scores. On the other hand, scores of the PL group deteriorated significantly pro- to postoperatively on immediate and delayed recall of paired-associates (p
0306-4530/92 $5.00+0.00 01992 Persmnon Press ltd.
Printed in Great Brit~dn
EFFECTS OF ESTROGEN O N MEMORY FUNCTION IN SURGICALLY MENOPAUSAL W O M E N SUSANA M. PHILLIPS1 and BARBARA B. SHERWIN1,2 1Department of Psychology, Mt~Jill University, and 2Department of Obstetrics and Gynecology, Sir Mortimer B. Davis-Jewish General Hospital Montr6al, Qu6bec, Canada (Received 18 April 1991; infinalform 5 December 1991)
SUMMARY The effects of estrogen (E) on memory were assessed in 19 women who required a hysterectomy and bilateral oophorectomy for benign disease. Blood samples were drawn and memory tests were administered before surgery and again after 2 mo of postoperative treatment consisting of either monthly E or placebo (PL) injections. Scores on the immediate and delayed recall of paired-associates stayed at the same level in E-I~'eatedwomen, whereas they decreased significantly pre- to postoperatively in the PL-treated subjects. In the immediate recall of paragraphs, the scores of those given E improved postoperatively compared to baseline; scores remained unchanged in the PL group. No hormonal effects were apparent on the immediate or delayed recall of visual material, delayed recall of paragraphs, or digit span scores. These findings suggest that variations in specific aspects of memory function may occur in surgically menopausal women coincident with changes in plasma estrone and estradiol levels.
INTRODUCTION COMPLAINTS of memory problems and an inability to concentrate are among the symptoms that have been associated with the menopause, which is marked by dramatic decreases in plasma estradiol (E2) levels. Few studies, however, have adequately examined the association between plasma F_a levels and cognitive functioning during this reproductive phase. In two double-blind crossover studies, Campbell and Whitehead (1977) reported that estrogen (E) replacement therapy (conjugated equine E for 2 or 6 mo) was more effective than placebo (PL) in alleviating many menopause-related symptoms, including hot flushes, anxiety, and insomnia, and was associated with a significant improvement in m e m o r y in peri- and postmenopausal women. Similar results were reported in an uncontrolled study on postmenopausal women who had received estriol succinate for approximately 2 yr (Schneider, 1982). However, in both studies, the memory improvements were based only on subjective self-report data, without any formal objective psychometric tests of memory function. Evidence of improvements in cognitive functioning were provided by other researchers who used various psychometric tests. The performance of 11 postmenopausal women treated with Address correspondence and reprint requests to: Dr. Barbara B. Sherwin, Department of Psychology, McGill University, 1205 Dr. Penfield Avenue, Montrtal, Qutbec, CANADA, H3A IB1. 485
486
S.M. PHILLIPS& B. B. SHF.RWlN
E2-17[3-valerianate for 3 mo was found to be superior to the performance of 10 PL-treated subjects on various cognitive tasks, including a visual search task, a modification of the Stroop test, a choice reaction time task, and tests of concentration and attention (Fedor-Freybergh, 1977). Hackman and Galbraith 0977) also reported an improvement in scores on the Guild Memory Test in nine surgically or naturally menopausal women treated with piperazine estrone (El) sulphate for 6 mo, compared to the performance of nine subjects who received PL. However, when these data were reanalyzed by Sherwin (1988), the between-group differences were not statistically significant. In a double-blind crossover study, Sherwin (1988) reported that surgically menopausal women who received PL for 3 mo following bilateral oophorectomy showed significant decreases in test performance relative to preoperative scores in digit span, immediate recall of paragraphs, and abstract reasoning and response speed and accuracy, whereas those given hormone replacement therapy postoperatively maintained their preoperative levels of performance. Other studies have failed to detect an effect of E on cognitive function assessed with different psychometric tests. Rauramo et al. (1975) tested surgically menopausal women who received 6 mo of either E 2 valerate or PL and found no significant differences between or within the groups on the Integration Memory Test, or on tests of reaction time or logical thinking. It was noted, however, that women who did not receive E complained most often of problems with memory and concentration. Vanhulle and Demol (1976), using tests of concentration, visual memory, digit span, and learning ability, also failed to detect any changes in postmenopausal women following 3 mo of treatment with estriol compared to PL, although there was some evidence of increased attention in the E-treated women. The discrepancies in these results may be due to a number of factors. Various types of psychometric tests were administered across studies, each of which may have measured very different aspects of cognitive functioning. It is possible that some tests tapped a specific area of mental functioning (e.g., visual memory) that may be unaffected by E levels, thereby yielding negative findings, whereas other instruments may have assessed abilities that are indeed sensitive to hormonal changes. Another possible reason for the lack of consensus in findings may be due, in part, to the fact that very different E preparations were administered in various doses, and that in most cases plasma E levels were not measured. In addition, although some studies have generally found a deterioration in mental functioning in postmenopausal women given PL compared to E-treated subjects, none have addressed the possibility that decrements in test scores may be simply an artifact of menopause-related symptoms, such as hot flushes or dysphoric mood, which may have interfered with test performance. The goal of the present study was to determine the specific pattern of memory changes that may occur coincident with changes in plasma E levels in surgically menopausal women, while attempting to control for previous methodological shortcomings. In addition to a battery of objective memory tests, radioimmunoassays of plasma E 1 and E 2 levels, and assessment of symptoms such as depressed mood, anxiety, hot flushes, and sleep disturbances, were carded out. SUBJECTS AND METHODS
Subjects Between February 1988 and July 1990, 31 Caucasian women who required a total abdominal hysterectomy and bilateral salpingo-oophorectomyfor benign disease were recruited. Exclusion criteria included a diagnosis of malignant disease and the presence of known contraindicationsto hormone replacement therapy. Subjects had to be in good general health except for the reasons necessitating the surgery, and they must never have
EFFECTS OF ESTROGEN ON MEMORY F~rNCTION
487
received E replacement therapy. Six women discontinued participation because of scheduling problems, surgical complications, or misgivings about the random assignment procedure. If the subject began experiencing severe menopausal symptoms within a few weeks postoperatively, the double-blind code was broken for ethical reasons, and E2 was administered if she had been receiving PL. Six subjects had to be dropped from the study on this basis, leaving 19 women who completed the study (mean age+ SD=48.2+ 4.7 yr; mean years of scolarity+ SD ---11.4 + 2.4).
Procedure In this 2 (test periods) × 2 (treatments) repeated-measures design, written informed consent was obtained on a form approved by a hospital ethics committee. Subjects were told that they would be participating in a study examining the effects of hormones on mood and memory. They also were informed that they would be receiving postoperatively either E or a preparation with no active hormone. Subjects were randomly assigned preoperatively, in double-blind fashion, to either an E replacement-therapy group (n = 10) or a PL group (n = 9). Test sessions were scheduled approximately 1 - 2 wk prior to surgery (baseline) and approximately 2 me postoperatively. The postoperative test session took place 2 - 4 days after the third monthly injection, the time at which plasma E levels were maximal (Sberwin, 1988). Each session took place at approximately the same time of day, in order to control for possible diurnal variations in alertness. Within each hour-long test session, a blood sample was drawn, followed by an immediate-recall phase in which the subject was asked to recall each individual memory item immediately after presentation. After all the memory tests were presented, a 20- to 30rain delay period ensued, during which the subject completed a series of questionnaires, including those listed below. The delayed-recall phase then followed, during which the subject was asked to recall test items from each of the first three subtests listed below. Drugs The E group received three monthly IM injections of 10 mg E 2 valerate (Delestrogen; Squibb, Canada), whereas the PL group received monthly injections of 1 ml sesame oil. Plasma Hormone Assays Ten milliliters blood were collected at each test session by venipuncture into heparinized Vacutainer tubes. The blood was immediately centrifuged and the plasma stored at -20 ° C until analysis. Plasma F_~and E 1 levels were determined in duplicate by radioimmunoassay. F_,z was measured with a non-extraction, double antibody technique (Garnma-B Direct Oestradiol RIA Kit, IDS Limited, Boldon, U.K.), in which an iodinated F_,z derivative is used. The antiserum is highly specific for ~ . E 1 was analyzed by an ether extraction method, with tritiated E 1 as tracer and an antibody raised to this steroid. Charcoal was used as the adsorbent. Materials Wechsler Memory Scale (WMS~ (Wechsler, 1945; Stone et al., 1946; as modified by Russell, 1975): The two forms of the WMS were counterbalanced across the two test sessions, such that half the subjects received Form I first and half Form 11. The following four subtests were administered: a) Logical Memory/Paragraph Recall: The subject listened to a short paragraph of about four to five sentences and then was asked to verbally recall what she remembered from the passage. Two paragraphs were presented once each. b) Visual Reproduction: The subject was shown a design for 10 sec and then asked to draw what she remembered seeing on the card. Three cards of designs were presented once each. Subjects were given separate sheets of paper for each drawing to minimize visual exposure to designs presented first. c) Associate Learning: The subject listened to a list of 10 word-pairs (paired-associates) presented in random order, six of which were clearly related (easy-associate; e.g., fruit-apple) and four of which had no apparent relationship (hard-associate; e.g., necktie-cracker). After hearing the list, she was asked to recall the word that was paired (e.g., apple) with each cue word (e.g., frui0. Three such trials of presentation and immediate recall were conducted. After the delay period, one final recall trial was administered. In scoring, two points were given for a correctly recalled hard-associate and one point for an easy-associate. d) Digit Span: The subject listened to a series of numbers and was instructed to verbally repeat them in the given order immediately afterwards. In order to ensure approximately identical test conditions across subjects, a manual of exact procedures was
488
S. M. PHILLIPS& B. B. SHERWIN
prepared, verbal test items were presented to subjects via a tape recorder, and verbal responses were recorded on tape and later transcribed. Protocols were scored independently by three coders who were blind to group assignment, Weehsler's (1945) original scoring criteria for the Logical Memory and Visual Reproduction subtests have been criticized as being poorly defined, which may lead to significant scoring discrepancies among coders (Mitchell, 1987). An improved scoring system therefore was developed for this study with welldefined criteria and high inter-rater reliability (r = 0.98 for Logical Memory; r = 0.96 for Visual Reproduction). In addition, alternate-form analyses indicated the two forms were equivalent under these scoring criteria (Phillips et al., unpublished data). Menopausal Index (Blatt et aL, 1953; revised by Sherwin, 1983): This 27-item paper-and-pencil questionnaire was administered pre-and postoperatively in order to monitor the frequency of various menopausal symptoms on scales ranging from 0 (never) to 7 (very often). In addition to an overall menopausal symptom score, this instrument also yields scores for three subscales: somatic, psychosomatic, and psychological symptom clusters. Mulfiole Affect Adjective Check List (MAACL) (Zuckerman & Lubin, 1965): In this brief, self-administered inventory, the subject indicated which of 132 adjectives best reflected her present mood state at the time of each of the test sessions. This instrument yields scores for current levels of anxiety, depression, and hostility, and has been shown to be valid and reliable. It has been used previously in psychoendocrine studies to measure the relationship between affect and concurrent hormone levels (e.g., Sherwin & Gelfand, 1985).
Statistical Analyses Memory, mood, and symptom data were analyzed with the SPSS-X multivariate analysis of variance (MANOVA) program. The study had a repeated-measures design, in which treatment group (E or PL) was the independent factor and time (pre- or postsurgery) was the repeated-measures factor. If a significant multivariate F statistic (p < 0.05) was found for the treatment x time interaction, the Hummel-Sligo procedure was employed to examine each dependent variable with univariate F tests. When a univariate F was significant, post-hoc pairwise comparisons were carried out with Tukey's HSD procedure. RESULTS
Baseline Subject Characteristics Student t-tests were carried out on subject age and education, and on baseline measures of memory, mood, and menopausal symptoms o f the treatment groups. There were no significant group differences on any o f these variables, suggesting that the E and PL groups were comparable on these dimensions before they underwent surgery.
Plasma Hormone Measures There was a significant group x time interaction for both E 1 (F1,17=217, p < 0 . 0 0 1 ) and F_q (F1A7=198, p < 0 . 0 0 1 ) . Preoperatively, mean plasma E 1 and E 2 levels fell within the normal physiological range and did not differ between the E and PL groups. By the second month after surgery, however, the supraphysiological levels of E 1 and F~ in the E group exceeded both their own preoperative levels ( p < 0 . 0 1 ) and the postoperative values o f the PL group ( p < 0 . 0 1 ) . In addition, the PL group experienced a significant decrease in plasma E 1 and F_q levels postoperatively, to values within the postmenopausal range (p < 0.05) (Table I).
Memory Measures There were significant group x time interactions for the f o l l o w i n g m e m o r y measures: i m m e d i a t e p a r a g r a p h recall (Fl.17 = 4.83, p < 0.05), i m m e d i a t e recall o f p a i r e d - a s s o c i a t e s (Fl.17 = 6.55, p < 0.03), and delayed recall of paired-associates (F1,17= 4.67, p < 0.05). Post.hoc pairwise comparisons indicated that the E group showed a significant increase in immediate paragraph recall scores after h o r m o n e replacement therapy c o m p a r e d to their preoperative
489
EFFECTS OF ESTROGENON MEMORY FUNCTION
scores (p < 0.05), whereas the PL group maintained their baseline scores. On the other hand, scores of the PL group deteriorated significantly pro- to postoperatively on immediate and delayed recall of paired-associates (p
