| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Center for Cognitive Medicine, Neuropsychiatric Institute, University of Illinois at Chicago, Chicago, Illinois 60612
Address all correspondence and requests for reprints to: Pauline M. Maki, Center for Cognitive Medicine, Neuropsychiatric Institute MC 913, University of Illinois at Chicago, 912 South Wood Street, Chicago, Illinois 60612. E-mail: pmaki{at}psych.uic.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Objective: Our objective was to determine whether supraphysiological testosterone influences verbal memory and brain function during a verbal memory task in healthy eugonadal older men.
Patients, Design, and Setting: Fifteen cognitively normal men, aged 66–86 yr, participated in a randomized, double-blind, placebo-controlled crossover trial involving 9 months of participation per participant at a hospital-based research facility.
Intervention: We used testosterone enanthate (200 mg im every other week for 90 d) crossed over with placebo (sesame oil vehicle im) with a 90-d washout between treatments.
Main Outcome Measures: Performance was assessed on a standardized verbal memory test, and brain activity (relative glucose metabolic rates) in medial temporal and frontal regions was measured with positron emission tomography during a verbal memory task.
Results: Treatment increased total testosterone by 241%. Behavioral results showed a significant decrease in short-delay verbal memory with treatment (P < 0.05, effect size = 0.59 SD) and a nonsignificant decrease on a composite verbal memory measure (P = 0.09, effect size = 0.48 SD). Positron emission tomography scans revealed decreased relative activity in ventromedial temporal cortex (i.e. right amygdala/entorhinal cortex) and increased relative activity in bilateral prefrontal cortex with treatment.
Conclusions: Decreased verbal memory and altered relative activity in medial temporal and prefrontal regions suggest possible detrimental effects of supraphysiological testosterone supplementation in elderly men. The results do not rule out potential benefits with other regimens, cognitive tests, or populations.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Five randomized, placebo-controlled clinical trials have investigated the effects of testosterone treatment on verbal memory in healthy elderly men (10, 11, 12, 13, 14). Results are mixed, although a finding of a beneficial effect of im injection of testosterone enanthate (TE) (100 mg/wk) in older men with low normal baseline testosterone levels was later replicated (12, 13). One of these two studies suggested that conversion of testosterone to estradiol modulated beneficial effects of TE on memory, because memory improved with im TE alone but not when TE was administered with an aromatase inhibitor (13). In contrast, no effects on verbal memory were observed with other testosterone regimens, including scrotal testosterone (15 mg/d) in men with normal baseline testosterone (15), testosterone cypionate (200 mg im every other week) in men with low baseline testosterone levels (11), or transdermal testosterone gel (75 mg total dose/d) in Alzheimer’s patients and healthy control subjects (14).
Basic science and neuroimaging studies suggest that any beneficial effect on verbal memory may relate to enhanced function and structure of the hippocampus and prefrontal cortex. Testosterone receptors are found in the hippocampus and prefrontal cortex in primates (16, 17), and physiological levels of testosterone help to maintain dendritic spine density in the cornu ammonis 1 (CA1) region of the rodent hippocampus (18, 19). Testosterone appears to enhance verbal memory through its conversion to estradiol (13), and our previous positron emission tomography (PET) studies show estrogen-related alterations in blood flow in hippocampus and parahippocampal gyrus during performance of verbal memory tests (20, 21). A recent study linked the maintenance of high endogenous free testosterone levels to enhanced blood flow in medial temporal and frontal areas in older men (6).
Here we report data from a double-blind, placebo-controlled crossover study of TE (200 mg im every other week) in a group of healthy, eugonadal older men that includes some individuals with low-normal total testosterone. Primary outcome measures were verbal memory performance and regional brain function measured with PET from the radiotracer [18F]fluorodeoxyglucose (FDG) during performance of a verbal memory test. These outcomes measures are clinically relevant, because they change early in the course of Alzheimer’s disease (22). We hypothesized that supraphysiological TE would enhance verbal memory and that this enhancement would be associated with an increase in relative activity in the medial temporal lobe (i.e. hippocampus, parahippocampal gyrus, perirhinal, and entorhinal cortices) and prefrontal cortex during performance of a verbal memory test.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Fifteen elderly men participating in a larger pilot study of the effects of TE on mood and cognition agreed to participate in an optional PET neuroimaging study. They were recruited through advertisements in local newspapers and local retirement communities. The first 15 men to consent were enrolled. Each participant underwent a medical screen with a study physician before enrollment. Inclusionary criteria were age greater than 65 yr; normal prostate-specific antigen (PSA), hepatic function, hematocrit, and lipid values; normal digital rectal exam; and a total testosterone level greater than 240 ng/dl. Exclusionary criteria were prostate cancer, allergy to sesame oil vehicle, serious systemic disease, conditions that affect gonadal hormone levels (e.g. primary hypogonadism), use of medications affecting central nervous system function, current Axis I psychiatric disorder, and traumatic brain injury. Standardized instruments were used to rule out clinical depression, prostate symptoms (23), alcohol abuse (24), and cognitive impairment (i.e. Mini Mental State Exam < 25th percentile for age and education). Before and after the PET study, participants underwent a routine medical examination, including determination of blood pressure and pulse. Participants did not receive any central nervous system-active drug and did not suffer any significant medical or psychiatric illness during the study. The Johns Hopkins University Institutional Review Board approved the study, and informed consent was obtained from all participants after they received a full explanation of the procedures. The participants received no compensation for participation.
Randomization procedures
An investigator (J.B.) determined the treatment order by coin toss and conveyed that information to a study nurse. Neither the person responsible for randomization nor the person responsible for dispensing the treatment participated in data collection or analysis. At the beginning of each treatment phase, blinded medical personnel administered the injection. The participants received each of two treatments for 90 d, one consisting of TE (200 mg every other week) and the other being an identical placebo (sesame oil vehicle), with a 90-d washout between phases. Neuropsychological, neuroimaging evaluations, and hormone measures were obtained on the same day at the end of each of the two phases. Evaluations took place 7–10 d after the final biweekly injection, corresponding to postpeak values. There was a 6-month interval between the two (i.e. placebo and TE) test sessions.
Image acquisition
PET and FDG were used to measure relative regional cerebral metabolic rate for glucose during performance of a verbal memory task. Brain function was measured with regional decay-corrected raw counts normalized to whole-brain values as a surrogate index of glucose metabolism. Relative activity, as used here, refers to this measure in brain regions of participants while they performed a verbal memory test. PET scans were performed on a Siemens ECAT EXACT HR+ scanner, which provided 63 slices (voxel size 1.87 x 1.87 x 2.42, resolution 4–5 mm). A flexible iv catheter was placed into the antecubital vein for the radiotracer administration. Approximately 5 mCi FDG was injected as a bolus. Reproducible head positioning and immobilization were maintained by a thermoplastic mask that was custom-fit to each individual. Attenuation correction was performed using a transmission scan acquired before the emission scans. Images were acquired in three-dimensional mode for 30 min, and the images were summed for analysis.
PET activation task
The verbal memory task included an encoding phase that was followed 30 min later by a retrieval (i.e. recognition) phase. In the encoding phase, participants viewed 20 abstract words (e.g. hindrance and suffrage), five to nine letters in length, displayed individually on a laptop computer for 3000 msec each. The recognition task began 30 sec before injection of FDG and required recognition of studied words as well as monitoring and encoding of novel distractor words that might reappear during the task. The participants responded yes and no to stimuli by pressing hand-held trigger buttons (left for novel words seen for the first time, right for reappearing words regardless of whether words first appeared during the encoding or recognition phase). The recognition task comprised 320 trials divided into 16 blocks of 20 words each. One set of 10 targets (i.e. words studied in encoding phase) appeared randomly in even-numbered blocks, and the other set appeared randomly in odd-numbered blocks. Half of the items in blocks 1–8 were targets, and half were distractors. To increase task difficulty, two distractors from blocks 1–8 reappeared eight blocks later, such that in blocks 9–16, 10 items were targets from the study phase, two were previously seen distractors, and eight were novel. The primary outcome measure was the number correct of 320. There were two parallel forms of the task, each matched for imagery, concreteness, frequency, and length. One form was administered at the first imaging session, and one was administered at the second imaging session.
Neuropsychological testing
A technician administered a 75-min neuropsychological test battery in conjunction with each neuroimaging evaluation. The primary outcome measure was the California Verbal Learning Test (CVLT) (25), a measure of verbal learning and memory. Secondary outcome measures included the following instruments: 1) Benton Visual Retention Test, Form D (26), a measure of short-term memory for geometric figures; 2) Digit Span, a measure of attention and working memory; 3) 
-Span (27), a test of working memory and attention; 4) Grooved Pegboard, a speeded manual dexterity test; 5) Trail Making Test, a measure of attention, visuomotor scanning, and cognitive flexibility; and 6) Positive and Negative Affect Schedule, a measure of mood (28). A modified version of the Primary Mental Abilities Vocabulary Test (maximum score = 75) estimated verbal knowledge (29) but was not an outcome measure. Parallel forms of the CVLT were used to minimize practice effects.
Hormone assays
Blood samples were drawn 7–10 d after injection of TE or placebo, in conjunction with imaging assessments. Serum levels of estradiol and total testosterone were measured by Quest Diagnostics Inc. (Baltimore, MD). Estradiol was extracted using a RIA technique (normal range 0–35 pg/ml). Total testosterone was extracted using a combination of competitive chemiluminescent assay and RIA, with normal ranges of 194–833 ng/dl.
Data analysis
The primary neuropsychological outcome was the CVLT, which, like the PET activation task, measured verbal memory and was administered twice with a 6-month interval between test sessions. We first calculated standardized z-scores for each of three CVLT scores (learning in trials 1–5 and short- and long-delayed free recall) using means and SD based on performance during placebo. We then averaged the scores within each phase to create a composite score for each phase for each subject. We next computed repeated-measures ANOVA with treatment (TE or placebo) as the independent variable and composite scores as dependent measures. Effect sizes were calculated with Cohen’s d, and small, medium, and large effects were evaluated according to Cohen’s standard definition (30).
PET image processing and analysis
Statistical parametric mapping (SPM99) (31) was used to compare patterns of task-associated brain activity during TE treatment and placebo. Voxel-based image analysis involved four steps: image realignment to adjust for head motion and variability in positioning between scans, stereotaxic spatial normalization to a standardized brain atlas, image smoothing, and statistical analysis, including adjustment for differences in mean global decay-corrected radioactivity between scans. The second PET image for each subject was first coregistered to the first image to correct for intersession movement. Images were then spatially normalized to the Montreal Neurological Institute template and smoothed with a 12-mm3 filter. Global values of decay-corrected brain radioactivity were obtained, and proportional scaling was used to adjust for the global means. The WFU Pickatlas Automated Anatomical Labeling tool (32) was used to define a priori regions of interest, including medial temporal (i.e. hippocampus, parahippocampal gyrus, entorhinal, perirhinal cortex) and prefrontal regions (i.e. superior, middle, inferior, and medial frontal gyri). Statistical parametric mapping implements the general linear model and paired t tests to evaluate within-subject, treatment-related change. This produces a brain map showing significant changes between hormone and placebo conditions. In exploratory analyses, we carried out voxel-wise, whole-brain analyses. Clusters defining changes in specific brain regions were viewed as significant when the peak voxel-based change was significant at P < 0.01 (uncorrected), with a minimal cluster size (k) = 10 voxels. Anatomical localization of significant activations of interest was determined by overlaying activations on structural magnetic resonance imaging scans and consulting brain atlases.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The mean age of participants was 73.9 yr (range 66–86, SD = 6.75), mean education was 17.4 yr (SD = 1.87), mean Mini Mental State Exam was 29.1 (SD = 1.14), and mean Primary Mental Abilities vocabulary was 42.0 (SD = 11.0). Baseline total testosterone and estradiol levels were, respectively, 10.21 ± 3.19 pg/ml, 475.64 ± 165.41 ng/dl, and 24.71 ± 24.71 pg/ml. Nine of the 15 were randomized to receive testosterone first (binomial test, P = 0.60). All participants received the study interventions as planned, averaging about 14 ± 1.4 d between successive injections. One man fell asleep during the recognition task and was excluded from the imaging analysis. As shown in Table 1
, mean total testosterone levels during placebo were 402 ng/dl (SD = 138) and increased significantly with TE treatment by 241% (range 47–804%, P < 0.001). During placebo, four of the 15 participants had total testosterone levels less than 300 ng/dl, the common clinical cutoff. Estradiol also increased significantly, by an average of 203% (range 56–492%, P < 0.001). Hematocrit increased significantly from the beginning to end of the study (from 43 to 46%) but was in the normal range (i.e. 40–53%) for all subjects and was no higher at the end of the 9-month study for those randomized to testosterone (45%) vs. placebo (45%) in the final treatment phase. There were no adverse events associated with testosterone treatment and no change in American Urological Association Symptom Index score, prostate-specific antigen, or any measured health-related outcome across the study. Treatment-related changes in cognition and brain function were not significantly correlated with any health outcomes, including mood.
|
Table 1
shows the performance measures for each phase, and Fig. 1 shows the effect sizes for the primary outcomes. The test-retest correlations for the primary outcome measures on the CVLT were 0.91 or greater between the two assessments (P < 0.001). The test-retest correlations for the secondary outcome measures ranged between 0.64 and 0.83. The composite verbal memory score tended to be lower with testosterone compared with placebo (P = 0.086), with a medium effect size (Cohen’s d = 0.48). Repeated-measures ANOVA on the individual verbal memory measures revealed significantly worse performance on the short-delay free recall trial during treatment compared with placebo (P < 0.05), with a medium effect size (Cohen’s d = 0.59). Short-delay free recall also improved over time (P < 0.05) for the main effect of time. Table 1 also shows scores on secondary cognitive outcome measures for the two treatment phases. No treatment effects were observed on any secondary cognitive measure. Positive mood decreased over time, independent of treatment (P < 0.05). There was no effect of treatment on performance of the PET activation task, although performance decreased over time, independent of treatment (P < 0.01). Overall, these data suggest a significant decrease in verbal memory with testosterone treatment, with no effect on working memory and attention, psychomotor speed, figural memory, or mood.
|
Figures 2 and 3 show the analyses addressing the primary hypotheses for the medial temporal and prefrontal regions of interest, respectively. Measures of regional cerebral metabolic rates for glucose within the primary regions of interest were reliable between placebo and testosterone assessments, with correlations ranging from 0.80–0.94. Testosterone treatment led to a decrease in relative activity (representing regional glucose metabolism) in a medial temporal region encompassing right dorsal entorhinal cortex/amygdala. There was also a trend (P < 0.05) toward an increase in relative activity in the left entorhinal cortex, right posterior hippocampus and right parahippocampal gyrus. Testosterone treatment led to widespread increases in relative activity in multiple prefrontal regions, including left medial, superior, and inferior frontal regions and right middle and orbital regions.
|
|
Table 2 presents the results of exploratory analyses examining significant treatment-related changes in relative activity across all brain areas. Increases in relative activity were observed in the left inferior parietal lobe/supramarginal gyrus, an area extending from the left subgenual area of the anterior cingulate inferiorly and laterally to Broca’s area, the right cerebellar culmen/fusiform gyrus/pons, right middle frontal gyrus, left cerebellum, left parietal association cortex, left substantia nigra, left superior frontal gyrus, right thalamus, right cingulate gyrus, and left orbital gyrus. Significant decreases were observed in the right postcentral gyrus and right entorhinal cortex/amygdala.
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Previous trials of testosterone supplementation produced mixed results with respect to verbal memory (10, 11, 12, 13, 14, 33). The reasons for the mixed results are difficult to determine, because treatment type and duration, cognitive tests, timing of cognitive tests in relation to treatment, and subject characteristics differed across studies. In previous positive trials, cognitive testing occurred 14–48 h after injection of 100 mg testosterone ester, coinciding with peak testosterone levels (12, 13, 33). Conversely, testing in the present trial occurred 7–10 d after injection, coinciding with postpeak (mid-interval) values. Baseline and posttreatment total testosterone levels were similar across studies, although the validity of using peripheral hormone levels as a marker of central hormone levels is questionable (34). Taken together then, the trials to date suggest that lower dosages of weekly TE injections are superior (at least in relation to verbal memory) to higher dosages of biweekly injections in older men. Supraphysiological testosterone administration may be beneficial for verbal memory in older men with low testosterone levels when administered at lower dosages and over shorter intervals, as in studies with weekly injections (12, 13, 33). This benefit may be most pronounced when testosterone levels are rising or at a peak and when the frequency of testosterone administration (weekly) induces smaller fluctuations in circulating testosterone levels. Conversely, in a group of eugonadal elderly men that includes some individuals with low-normal testosterone levels, supraphysiological testosterone administration may be detrimental when administered at high dosages over long intervals, as in the biweekly regimen. This detrimental effect may be most pronounced in a regimen that induces appreciable fluctuations in circulating testosterone levels, particularly when testosterone levels are dropping from peak.
Neuroimaging data revealed changes in medial temporal and frontal regions with biweekly testosterone administrations. Although higher endogenous free testosterone levels are associated with increased blood flow in these areas (7) and memory performance (6), 200 mg TE biweekly appears to have opposite effects. The overlap between receptor locations (16, 17) and brain areas showing PET changes is consistent with direct effects of testosterone on androgen receptors, direct effects of aromatized testosterone on estrogen receptors, or both. Exploratory whole-brain analyses replicated the region of interest findings of significant increases in prefrontal cortex and decreases in the right amygdala/entorhinal cortex. Other exploratory findings parallel findings from other neuroimaging studies examining testosterone effects during rest or other conditions, including changes in the left parietal lobe (35), subgenual cingulate (BA 25) (36), and midbrain (37). These findings may relate to general effects of testosterone on brain function that are not specifically related to memory.
The present study has several limitations. Because of the long half-life of FDG, we could not administer a control neuroimaging task on the same day as the memory task. We could have administered a control task on a different day, but the validity of this would be questionable given the pharmacokinetics of im TE. A control task would have helped to identify specific cognitive processes/systems underlying the pattern of findings during verbal memory tasks. Instead, we carried out correlational analyses to understand brain changes in relation to changes in memory performance. Second, treatment effects on hormone levels were highly variable, possibly reflecting true clinical variability. Third, the findings were based on a small sample size (because of PET costs) and do not address potential beneficial effects of testosterone on a range of spatial abilities.
In summary, the present study suggests that supraphysiological doses of biweekly im TE in a group of eugonadal elderly men may lead to declines in verbal memory and impaired function of medial temporal and prefrontal brain regions. The study does not address the impact of physiological testosterone upon cognition and cannot be used to definitively exclude a neuroprotective impact of testosterone upon central nervous system functioning in elderly men.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Disclosure Summary: The authors have nothing to declare.
First Published Online August 28, 2007
Abbreviations: BA, Brodmann area; CVLT, California Verbal Learning Test; FDG, [18F]fluorodeoxyglucose; PET, positron emission tomography; TE, testosterone enanthate.
Received August 16, 2006.
Accepted August 16, 2007.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
