Skip to main content

Transforming the understanding
and treatment of mental illnesses.

Celebrating 75 Years! Learn More >>

Improving Treatments for Mood Disorders and Depressive Symptoms in Women During Mid and Later Life


TAMARA LEWIS-JOHNSON: Good afternoon. Welcome to the Improving Treatments For Mood Disorders and Depressive Symptoms in Women During Mid and Later Life. My name is Tamara Lewis-Johnson and I am the chief of the Women’s Mental Health Research Program at the NIMH Office of Disparities Research and Workforce Diversity.

The purpose of the webinar series is to spotlight research on mental health disparities, women’s mental health, minority mental health, and rural mental health. This afternoon we are spotlighting the research of Drs. Barbara Parry and Susan Girdler. Dr. Vonetta Dotson is unable to participate in the webinar.

This research was funded by the NIMH’s Division of Translational Research. Please allow me to introduce our moderator for today’s webinar. Dr. Laura Roland serves as the chief of the Neuroscience of Mental Disorders and Aging Program in the Geriatrics and Aging Processes Research Branch in the Division of Translational Research. Her program supports translational neuroscience studies of risk, presentation, course, and outcome of mental illness in later life or in the relation to the aging process (recording ceased at this point) and with no further ado, Dr. Rowland.

LAURA ROWLAND: Thank you so much, Tamara. I am first going to go through some housekeeping notes. All participants have entered on mute and are on listen-only mode and your cameras are disabled. When we come to the question-and-answer portion, please submit your questions through the box. You can do this at any time during the seminar. And please address your question to the intended speaker. We are going to answer the questions at the end during the Q&A session. If you have technical difficulties hearing or viewing the webinar, please note these in the Q&A box and our technicians will work to fix the problem. You can also send an email to events at if you have additional issues.

The webinars in the series are being recorded and they will be made available for viewing at a later date. CEUs and certificates of attendance are not being offered for this webinar.

Today’s webinar is focused on the menopausal transition, which is a period of vulnerability for the development and worsening of mood disorders. Our presenters today are going to highlight current research findings on the underlying biobehavioral mechanisms that may increase depression and other mood disorders during the menopause transition.

Our first speaker is Dr. Barbara Parry. She will give the talk titled Circadian Rhythms, Sleep and Light Interventions in Menopausal Depression. Dr. Parry is a research professor of psychiatry at the University of California, San Diego, UCSD, where she has served as director of the Women’s Mood Disorders Clinic of the UCSD Outpatient Psychiatric Services and the La Jolla Psychiatry Specialty Clinics, director of the Women’s Mental Health Clinic at the San Diego Veterans Administration Healthcare Center.

The second presenter is Dr. Susan Girdler and her talk is titled Hormone Sensitivity and the Stressful Life Events: Predictors of Response to Transdermal Estradiol in Perimenopausal Women. Dr. Girdler is Professor and Vice Chair of Faculty Development in the Department of Psychiatry at the University of North Carolina at Chapel Hill.

And I now hand it over to Dr. Parry.

BARBARA PARRY: I will be speaking on circadian rhythms, sleep and light interventions in menopausal depression. The hypothesis is the disturbances in the phase or timing, the amplitude or temporal organization of circadian rhythms characterize menopausal depression, and that can be corrected by critically timed sleep and light interventions that we call SALI, thereby improving mood and sleep.

To illustrate it here, normally, melatonin is a rhythm that goes up a couple of hours before sleep time and is maintained throughout the night and comes back in the early morning hours and its in phase with the sleep time but estrogen and progesterone when they change during menopause or other reproductive events can shift melatonin later with respect to sleep or earlier. Progesterone tends to delay rhythms and estrogen to advance them.

Likewise, sleep can get out of phase with melatonin. This can often happen in jet lag so that the sleep time is later than the melatonin time or it is earlier than the melatonin and it is not in sync as we would expect with someone whose mood and sleep is in sync.

In the methods in this study, we took perimenopausal women who were depressed by DSM-5 criteria, the Diagnostic and Statistical Manual of Mental Disorders for major depression and then normal control women without major medical or psychiatric illness. They were off psychoactive medication for two weeks and for hormonal replacement therapy or the antidepressant fluoxetine for four weeks.

We measured plasma melatonin, serum cortisol, prolactin, and TSH but I will be focusing on melatonin today. We sampled every 30 minutes between 6 p.m. and 11 a.m. overnight and patients were kept in the dim light conditions under 30 lux, and we measured serum estradiol, progesterone, follicle stimulating hormone, routinizing hormone at 6 a.m. and 6 p.m. We also measured sleep by polysomnography and sleep logs by the Pittsburgh Sleep Quality Index.

For the statistical analysis, we looked at covariates of age, final menstrual period, body mass index, season when tested and morning-eveningness questionnaire, which we applied to the data.

When we look at the demographics of these women, the only really significant difference was that in the depressed patients, they had a higher history of personal and family history of psychiatric illness.

Also, with regard to their Hamilton and Beck, their depression rating scores, they were certainly higher in the depressed patients and normal controls were morning types on the Horne and Ostberg morning and eveningness scale than were the depressed patients.

We examined the plasma melatonin every 30 minutes, here is the normal control profile of melatonin and we found that the depressed patients actually had higher levels, and this was particularly in the early morning hours between 2 and 5 a.m.

We looked at those women who had a family history of depression and that was even more marked in terms of them having higher melatonin levels in the early morning hours, which potentially can delay your sleep end time and then prevent you from getting morning light to help synchronize rhythms.

We measure melatonin by timing measures, the time of its onset, by the synthesis offset when the synthesis has stopped, and by its return to baseline, the offset, and the timing between onset and offset, the duration, and then the amplitude measures, there are peak levels in the area under the curve.

And then we defined these melatonin parameters. We fit a slope to it. And when the log transformed melatonin concentration curve first become steeply positive for at least three consecutive points, we call that onset. The synthesis offset is when the slope of the curve becomes steeply negative for at least three consecutive time points. Baseline offset is when it returns back to zero. The duration is the difference between onset and offset. The peak is the highest concentration and the area under the curve is the integrated melatonin between onset and offset.

When we look at these variables with menopausal depression versus normal controls, we find that the peak levels are higher in the depressed patients. The area under the curve is higher in the depressed patients.

When we look at those who have a family history of depression, we find that the timing measures are significantly different. The depressed patients with a positive family history have a later melatonin offset time and a longer duration. The melatonin duration in non-human animal species is the critical melatonin parameter that regulates changes in behavior like with the seasons and hibernating.

We also measure urinary measures of melatonin, and we collect overnight samples because that is when the melatonin is secreted, 2 overnights, 36 hours, starting at 6 p.m. and going until 12 p.m. the following day. We look at the onset time, which we define by the upward crossing of the cosine mesor after we fit a cosine curve to it. The offset is the downward crossing and the acrophase is the time of the peak melatonin of 6-SMT, 6-sulfatoxymelatonin quantity of the cosine fitted curve.

We look at the amplitude measures here, the mesor, the mean, and the acrophase is the timing of the highest point. When we examine this in menopausal depressed women versus normal controls, the acrophase or the timing measure is later in the depressed patients versus the normal controls.

The only difference we find between the reproductive hormones is that in the depressed patients, the follicle stimulating hormone is higher, which has been found by other groups as well compared to the normal controls and that may reflect their time since menopause.

Now, we also look at the sleep variables and these are the results from the polysomnography and actually depressed menopausal women sleep as well as normal controls. There are not significant differences in sleep latency, which is the time to sleep onset, sleep efficiency, the rapid eye movement, latency or REM sleep when most dreaming occurs or the total sleep time.

If we look at those women who have a family history of depression then we find that the – actually, the sleep of the menopausal depressed patients – their sleep deficiency is actually higher. And their lighter stages of sleep are slightly lower.

However, if we look at the subjective ratings of sleep when there is often a mismatch between object of measures of sleep and subjective measures of sleep. In this case, even though the menopausal depressed patients by polysomnography or objective measures of sleep are sleeping as well as the normal controls, the subjects, the depressed patients report more awakenings compared to the normal controls.

And when we look at the Pittsburgh Sleep Quality Index, the depressed menopausal women report poorer sleep compared to the normal controls, again, even though their objective sleep is very similar.

We also try to measure hot flashes by skin conductive. It is pretty hard to get women to participate in this. But I think the bottom line is that there were not more hot flashes by skin conductance measures in the depressed patients compared to the normal controls that you could account for more subjective sleep disturbances even in terms of frequency, duration, or severity.

However, again in terms of subjective assessment of the hot flashes, it is not a significant difference but there – for the depressed patients to report higher frequency and severity of hot flashes.

Based on this component of the study, we found that the plasma melatonin secretion was increased in the early morning hours between 2:30 and 5:30 in the menopausal depressed patients versus the normal controls. And those who had a family history of major depression, there was more of a delayed offset time of melatonin secretion. Urinary melatonin measures also supported the plasma measures in that secretion had a delayed offset in the acrophase. I did not think I had time to show these data but cortisol, prolactin, and TSH were not significantly different between depressed patients and normal controls. Follicle stimulating hormone levels were higher in the menopausal depressed patients. Although the polysomnography was not different, those women who had a family history of depression tended to have deeper sleep and less lighter sleep but they rated their subjective sleep by logs as having poorer sleep quality.

That was the basis for our looking at the sleep and light interventions. We can shift circadian rhythms by shifting sleep and by shifting rhythms with light. To give a little background and rationale for this, one of the great paradoxes in life is that if you take a depressed patient and you keep them all night, the majority of them will improve their mood, have anti-depressant effects of that intervention by the next day. The history of this is that nurses in Europe and at the NIH observed in patients who had manic depressive illness or bipolar illness that they did not sleep the night before switching from a depression into mania. And subsequently, studies showed that just one night of staying up all night induced rapid antidepressant effects in one day and also found that only a partial night of sleep restriction of four hours was required to induce this antidepressant effect to benefit mood.

There have now been over 30 studies in over 1000 patients that showed that a majority of depressed patients respond to one night of what we call wake therapy or therapeutic sleep restriction or deprivation. Some but not all patients may relapse when they go back to sleep the following night and even with a nap the following day. There is a different response in hormonal profiles between those who respond after the first night and those who respond after they have gone back to sleep, what we call recovering night of sleep. One seems to be a noradrenergic mechanism and the second night responders, which we have found characterize as many women with reproductive depressions, they have more serotonin deficiency and it enhances serotonin function. Whether it is being awake in the first half, or the second half of the night seems to depend on someone’s underlying circadian phase.

How does this work? We think that sleep in circadian rhythms that are realigned – they have been out of phase with each other. It also could work by rapid eye movement sleep or dream sleep occurs in the second half of the night and is often depressogenic. And most antidepressants suppress REM sleep. This being awake – patients are sleeping at the wrong phase of the circadian cycle; the suppression of REM sleep may be therapeutic.

It also may be – what we call an increase in homeostatic drive. The longer you have been awake, the more you are going to sleep better the following night. As people age, this homeostatic drive tends to decrease. If you are up for a longer period of time, you are going to sleep better the next night, which can also improve your mood. There is documented evidence of a change in thyroid function. Many people – for those who have been on call, for example, if you have been up all night, you feel a buzz the next day. That is probably your thyroid hormone kicking in.

Now, how do you maintain this effect if you lose it so rapidly? One way is to give light treatment, which we have chosen to do. But it needs to be bright light like 10,000 lux for 30 minutes a day at least 5 times the intensity of normal room light. Morning light, bright light will advance circadian rhythms whereas evening light will delay them. You can also phase-shift the sleep/wake cycle and advance it 30 minutes to 5 p.m. a day but that is not very practical for many people.

You can also enhance the effect with lithium or antidepressants but particularly in pregnant or postpartum women, that is not really always the best option, and it may also be true for menopausal women who may need estrogen to enhance their effects of antidepressants.

We wanted to explore combining the wake therapy and the light therapy. And the rationale is that the wake therapy will hasten and potentiate the effects of light therapy, which otherwise may take at least six weeks to exert its efficacy and in older people like eight weeks if just using light treatment alone in a non-seasonal depression.

On the other hand, light therapy can prevent the relapse after the wake therapy occurs. We can use both this wake and sleep adjustment to advance or delay, shift it earlier or later these wake and light interventions and help to realign disturbed sleep-in melatonin circadian rhythms and there is a more potent effect of combined therapy.

Our conceptual model is that in menopausal depression that we have shown from the preliminary baseline data that the melatonin profiles are delayed in depressed patients compared to the normal controls. And then if we intervene with the phase of what we call a phase-advance probe or intervention where we advance sleep and use morning light to advance rhythms, we hypothesize that those rhythms will become realigned. In contrast with the phase-delay probe or phase-delay intervention, we anticipate that there will not be that much difference in the rhythm they will actually be a little bit more delayed.

What we did in this study is we took perimenopausal women determined by their final menstrual period, using DSM-5 criteria for depression and normal control women. They were randomized in a parallel trial to the active phase-advance intervention or probe, which we use late night wake therapy. That means they are awake in the latter part of the night, and they sleep in the early pat of the night from 9 p.m. to 1 a.m. And they just do that one night. And then we add morning bright light, which phase advances or shift circadian rhythms of melatonin earlier. We started with 60 minutes daily, starting within a half an hour of their habitual wake time. And we started using initially eight weeks of treatment.

We compared that to the control phase-delay intervention or probe where we used early night wake therapy where participants were awake in the early part of the night and they slept in the late part of the night for four hours. We shifted and restricted sleep from 3 to 7 a.m. just for one night. And then we gave the evening bright white light, which shifts rhythms earlier for 60 minutes daily starting 90 minutes before their habitual bedtime and did that for 8 weeks.

This is just a schematic illustration after baseline screening and diagnostic evaluation and collecting urine for melatonin and actigraphy. They either got the phase-advance intervention where they slept from 9 to 1 a.m. one night and then they were given morning light for two weeks. We initially started with eight weeks but then we subsequently found that the two weeks was as effective as the eight weeks so then we reverted to the two weeks to reduce the burden on the patients. That was compared to the phase-delay intervention, which they sleep from 3 a.m. to 7 a.m. one night and then get evening light to delay their rhythms for 30 minutes a day within 60 minutes of the onset of sleep.

For this particular study, we used the Litebook just because it was more portable. We had started initially with very much larger light boxes but those proved to be rather cumbersome. We now have moved to using an intermediate sized light box just because we want to be sure the light box covered the full visual field. But it is a bright light and it is in the blue-green spectrum.

We found that just the baseline, first the higher the depression score, the more depressed the subjects were, the worst their sleep was. Mood and sleep at baseline were correlated.

And then consistent with our hypothesis that with the phase-advance – let me just get before I get to the treatment effects that we found that before the interventions, the acrophase in the depressed patients was basically shifted later compared – the acrophase was shifted somewhat I should say different in the pre-probe condition, the depressed patients compared to the normal controls.

With the treatment, however in the phase-advance intervention, we found that the offset time, which is a critical parameter in melatonin synthesis particularly in humans, onset and offset, move differently in response to light. The offset time was significantly advanced from before the intervention until after the intervention whereas in the control condition, there was really no significant movement in the offset time.

We found that the more the mood improved, the more there was a phase advance. The phase advance is illustrated by positive numbers here. The greater mood improvement correlated with the greater phase advance of the melatonin rhythm.

Just to show that the early wake therapy and the evening light did not worsen mood. I show this slide. But it can also have beneficial effects and we think that may be because of its effects on sleep that the early wake therapy in the evening light significantly improved sleep, their sleep quality both compared to normal controls and to their baseline measures whereas the phase-advance intervention did not have significant effects on sleep. We find that in other groups as well.

But both treatments work within one or two weeks by the late wake therapy, shifting circadian rhythms, and the early wake therapy improving sleep combined with their respective light treatments.

Menopausal depressed patients versus normal controls have phase-delayed urinary melatonin rhythms. And the sleep and light probes that advance melatonin rhythms improve mood more than the control probes that delay their rhythms.

The two weeks has equal efficacy to eight weeks of intervention. That sleep improves more with the phase delay probe or intervention in the depressed versus the normal controls. We target both melatonin rhythms and the sleep to improve mood.

The advantages of sleep and light therapy is that it is a nonpharmacologic treatment. It is well tolerated. It has few side effects. It is rapid acting. It can be done at home. It can be done independent of hospital and clinical visits, certainly an advantage during a pandemic. It is affordable and repeatable. It can be administered by paraprofessionals.

We have also explored this in other reproductive-related mood disorders such as premenstrual, postpartum depression. In those women who had phase-delayed melatonin and sleep rhythms, we advance and restrict sleep and then give morning bright light. In those women who have phase-advanced melatonin sleep rhythm, for example, during pregnancy, we delay and restrict sleep and give evening light. And we find the efficacy within one or two weeks in all conditions.

Our body is like a clock. If one wheel be amiss, all the rest are disordered, the whole fabric suffers. With such admirable art and harmony is a man composed, as Robert Burton said in the Anatomy of Melancholy.

And as Francis Crick said, if a breakthrough in the study of the brain does come, it is perhaps likely to be at the level of the overall control of the system. If the system were as chaotic as it sometimes appears to be, it would not enable us to perform even the simplest task satisfactorily. To invent a possible, although unlikely, example the discovery that brain processing was run phasically, by some kind of periodic clock, as a major computer is, would probably constitute a major breakthrough.

If we are ever going to take over the world, we have to synchronize our biological clocks.

I would not do this work without my dedicated lab. As Helen Keller said, alone we do so little, but together we can do so much.

I am indebted to my collaborators as listed on this slide. Thank you and I will turn it over to my esteemed colleague, Dr. Girdler.


SUSAN GIRDLER: Thank you so much. A little daunting to follow an icon like Dr. Parry. I am honored to be able to do so. I really appreciate being here today. I have no conflicts of interest to report. But I do want to express my gratitude to the National Institute of Mental Health that has provided support for a lot of the research that I have done in my career, including some of the research that I will present today.

A couple of objectives. I want to touch very briefly just on some evidence that mood susceptibility to hormonal fluctuations really represents an abnormal response to normal hormone change.

Transdermal estradiol, which is estradiol delivered through a skin patch – we know that it has been shown to treat perimenopausal depression, but I will present some evidence today that it might actually prevent the emergence of depression during the menopause transition.

Recent stressful life events and hormone sensitivity to that fluctuation will actually potentially predict the benefit of transdermal estradiol for affective symptoms in the menopause transition.

Lastly, I would like to suggest that the hormone susceptibility to hormonal flux may represent a transdiagnostic mechanism of risk for dysphoric mood and dysphoric mood state changes, which are common across almost all psychopathologies.

We know that reproductive events are a time of increased risk for mood disorders in females. As hormones begin to rise with puberty and begin to cycle, there is an increased risk for first episode depression, the development of premenstrual dysphoric disorder or menstrual magnification of a chronic affective illness.

But risk is increased even greater during times of tremendous hormonal change such as occurs when there is a precipitous drop in high pregnancy levels and to the postpartum period as well as during the menopause transition when we know there is also very rapidly fluctuating changes in reproductive hormones but against a backdrop of estrogen withdrawal.

It would not be surprising that these fluctuations of invariant hormones might influence mood because we know that these ovarian steroid hormones regulate virtually every neurobiological system that has ever been implicated in the disorders.

But not all women are sensitive or susceptible to mood impairment associated with these reproductive events. In fact, far fewer than 50 percent of women are sensitive to menstrual cycle fluctuations or postpartum hormone withdrawal or perimenopausal depression.

Who are those women? Can we predict those women who are susceptible to hormone flux and does that have implications for treatment?
We know that women in the menopause transition are at elevated risk for depression effect even those who have never had a depression prior to the transition have a twofold increase in risk of developing depression during menopause transition. And those with a prior history have up to a fourfold increase in risk.

The endocrine experience of the menopause transition is summarized in this slide whereas women approach their final menstrual period indicated by time 0 here, you see a gradual decrease in estradiol or estrogen levels at the primary estrogen estradiol and a gradual increase in follicle stimulating hormones so that by about two years from a woman’s final menstrual period, her estrogen levels are nice and low and stable. Her FSH levels are nice and high and stable.

But these average data very unlikely that they characterize the experience of any particular woman during the menopause transition. In fact, we know from these data of O’Connor and colleagues, which summarize daily changes in progesterone, this upper graph, estrogen in the middle graph, and follicle stimulating hormone in the lower graph. These are daily changes in these hormones in a single woman in the menopause transition.

The dark shaded patches on the X-axis here represent menstruation. And indeed the hallmark of the menopause transition is irregular menstrual cycle length. And bleeding patterns are the primary criterion for diagnosing a woman in a menopause transition, not her hormonal levels, not her menopausal symptoms, and not her age, but bleeding patterns.

We know that particularly when follicle stimulating hormone levels are variable that can cause periods, which occurs, for example, times of longer cycles and ovulatory cycles or skipped cycles. That variability can produce both episodes of hypo-estrogenism and hyper-estrogenism. In fact, estradiol levels can be higher on any day in the menopause transition than they ever are during a normal menstrual cycle.

What is the evidence that that kind of variability might be relevant to mood, this kind of experience? I should say it is not to be taken lightly either because exposure to that kind of endocrine environment can last on average five years. It is on average. It is five years that women are in the menopause transition. And other women can suffer from menopausal symptoms for many years beyond that.

But the evidence of that variability might be relevant to the depression came from early research by Ellen Freeman and her colleagues. They followed women in the menopause transition, none of whom had a prior depression. And they looked at their estradiol levels on an annual basis over eight years and found that while a woman’s average estradiol level over eight years did not predict the development of depression, the variability, the change in estradiol around that woman’s average level over eight years did increase the odds of her developing clinically significant depression as well as variability in follicle stimulating hormone.
Subsequent experimental research conducted by Peter Schmidt at NIMH and his colleagues – they recruited women who were currently free of any depression but some of these women had a past history of perimenopausal depression. Others were unaffected controls. Schmidt and colleagues put all of these women on three weeks of transdermal estradiol. You see these first three weeks here with the shaded symbols. And all women showed nice low levels of depression. This is measured using the Center for Epidemiological Scale Depression Score, a very common measure of depression in the field. Scores of 16 or above on this scale are indicative of major depression. You can see during the three weeks of transdermal estradiol, everybody had very low stable depressive symptoms.

He withdrew abruptly – took them quickly off of their transdermal patch to a third of them. And those women who had that history of prior perimenopausal depression showed an immediate increase, a rapid increase in the depressive symptomatology with the withdrawal of these hormones, suggesting a hormone-sensitive phenotype in perimenopausal depression. Unaffected controls here in the open squares had the abrupt withdrawal from estradiol but it did not increase the depressive symptoms. Some evidence of an abnormal response to hormonal change in those who were vulnerable to perimenopausal mood disorders.

I do want to also note that the women in the triangle here – they also had a past history of perimenopausal depression. They were left on the estradiol for the full six weeks and their depression symptoms remained low and stable.

These data among other data led by colleague David Rubinow and I to ask whether or not – we know that it is effective for treating current depression in perimenopausal women. But could transdermal estradiol actually buffer against the emergence of depressive symptoms?

In this study that we led, we recruited medically healthy non-depressed women. We were very specific – we recruited women who had very low depressive symptomatology, an average of five on that CES-D score. They were 45 to 60years of age and they all met a standard criterion for being in the early menopause transition, in other words, variable menstrual cycle lengths, late menopause transition. They were starting to have skipped periods or within the first year of their postmenopausal time.

We randomized 172 women to 12 months of transdermal estradiol at 100 micrograms per day and gave intermittent oral micronized progesterone every two to three months just to prevent endometrial thickening or hyperplasia or women were randomized to 12 months of transdermal placebo and oral placebo. You can see 63 and 69 completed the study.

We were very interested also in characterizing the psychosocial stress profiles of participants because we know the menopause transition is a time of not just of hormonal change but of tremendous change in psychosocial stress experiences. These were women who were recruited from the Chapel Hill, North Carolina community. Sixty percent of them had had at least one significant life event in the six months prior to enrollment and over 30 percent of them had had two or more significant stressful life event. You can see the nature of these stressors here on the slide.

We were also interested as Freeman and others had done in looking at this issue of variability in estradiol and does it predict depression. We did that. We took advantage of the placebo arm of this study. We had assessed estradiol from plasma from serum specifically from blood over 4 months during a 14-month period. And those women who had been randomized to placebo, over this 14-month period and remembering each of them had very low depressive symptom levels on this scale in enrollment. We found that by month 14, those who had the higher estradiol variability of that 14-month period had higher depressive scores but especially if they had been exposed to multiple recent stressful life events. The combination of estradiol variability with stressful life events was the strongest predictor of who developed an increase in depressive symptomatology over that period.

These are the data on the randomization effects that we found. Those women randomized to the active estradiol for 12 months are in pink and the placebo in the yellow/orange color. We found that over 12 months of assessment of daily mood and interaction between treatment with stressful live events. On average, those women at the end of 12 months who had been given the active estrogen did have lower depressive symptomatology. But that was especially true if women who had been exposed to two or more stressful life events prior to enrollment. The benefit of estradiol on depressive symptomatology was significantly greater in women with recent stressful life events.

Moreover, we found when we looked at episodes of clinical depression during that 12-month period and then on this scale, we know that a score of 16 or greater is predictive of major depression. Again, these women were all enrolled to be very low in depressive symptoms at the beginning. But by 12 months, those who had been randomized to placebo, over 30 percent of them developed one or more clinically significant depressive episode and that risk was cut by two and a half times in those who were randomized estradiol.

And that risk was – that risk benefit was especially evidence in women in the early perimenopause transition compared with those in the late or the early to postmenopause transition. The risk – those in the early perimenopause transition showed especially beneficial risk for cutting the risk of those depressive episodes.

The reason behind that – really, we do not know. It is not simply a matter that transdermal estradiol raises estrogen levels because early perimenopausal women naturally have higher mean estrogen levels than do late or early postmenopausal women. The explanation for this effect remains unexplained at present.

That was the first study to show that maybe you can prevent the depression in perimenopausal women with transdermal estradiol. But what are the mechanisms that might underlie the beneficial effects for affective symptoms? Our recently completed study was designed to look at that. We went a little bit beyond sort of a gross measure of just looking at estrogen variability. But in this study, we followed the methods of Elizabeth Andersen and Jennifer Gordon. We created an individual difference score for estradiol mood sensitivity strength.

This is how it worked. Over the eight weeks at baseline in this study, we took weekly measures of serum estradiol. WE also did weekly measures of this time – looking at gross measure of depression. We were interested in specific symptoms of anxiety and anhedonia because they are both core features of depression but also transdiagnostic across most psychopathologies.

And basically, we looked at the correlation within each woman of the magnitude of change from Week 1 to Week 2 in her estrogen and how that correlates with her anxiety and anhedonia symptoms at the index week. The correlation between this change and symptoms at this week, the correlation between this change and symptoms at this week and so on over the eight-week baseline period, yielding an individual index of hormone sensitivity strength.

In this study – it is a smaller study, as you can see. Only 82 women were randomized. We had intended on a larger sample size, but the pandemic ceased a lot of clinical research for quite a while. Within NIH approval, we terminated the study at this point because we had demonstrated – we did not have the power at least to look at main effects, statistical power look at our main effects and to look at our primary hypotheses. In this study, 42 were randomized, 42 women randomized to 8 weeks of transdermal estradiol and 40 randomized to 8 weeks of transdermal placebo. Again, these are women in the menopause transition, medically healthy.

Unlike our prior study though, we did allow women who had elevated mood symptoms – any psychotropic medications. You can see the number of completers for this smaller, randomized controlled trial.

Our protocol was basically after enrollment. Every woman went through Trier social stress test. In this one, we did not look at the biological responses. It was meant to control for the novelty effect of first exposure of this type of test.

Here is our baseline period I have already mentioned so eight weeks of estradiol assessment as well as anxiety with the Spielberger State Anxiety Inventory and anhedonia with the Snaith-Hamilton anhedonia Scale. Then they came in at the end of the eighth week for stress testing protocol where we looked at cortisol and pro-inflammatory cytokine IL-6 response to stress. Then women were randomized for another eight weeks to either active treatment with estradiol and placebo and had another stress test at the end of that period as well as weekly mood assessments to look at intervention effects.

By the way, I put that in for myself otherwise I knew I would forget. We did give the Greene Climacteric Scale at enrollment at Week 8 and Week 16, which allowed us to look at changes in vasomotor symptoms over time because we know estrogen will drastically reduce the symptoms and to control for that in terms of looking at our treatment effects.

We really replicated – always love seeing replication is the cornerstone of science. Very similar findings to our previous study although instead of just the gross measure of depression, this is showing the anhedonia symptom score. Again, in this study during the baseline eight-week phase, we found that those women who had high estrogen variability and had been exposed to more recent stressful life events showed a higher anhedonia symptom score. Anhedonia, by the way, is that kind of loss of pleasure and enjoyment that is very characteristic of depression that we don’t see in women with low variability. The same effects were seen in terms of high E2 variability, predicting anxiety as well but only in a statistical trend level.

We did find for the first time, the first study that we are aware of that has looked at the degree to which this estrogen variability modifies the stress response. This is in response to the Trier social stress test. Women who had higher estrogen variability at baseline showed a much greater cortisol response to stress relative to women with lower variable. And they showed a blunted interleukin-6 response to stress relative to women with lower variability. This inverse relationship between cortisol and this pro-inflammatory cytokine has been documented previously. And because we know that cortisol regulates the immune response to stress, this provides a biologically plausible explanation for why we see this divergent direction between cortisol stress response and IL-6. But the real point is that estrogen variability was modifying a woman’s stress response.

Now, to our treatment effects or intervention effects, we did find a main effect of treatment. This is controlling for baseline symptom levels as well as any change in vasomotor symptoms over time. We found that on average, women who were randomized to the active transdermal estradiol intervention had lower anhedonia symptoms and lower anxiety symptoms at the post-randomization period relative to women who had been randomized to placebo.

But – this beneficial effect was really most evidence in those women who had baseline had this higher estrogen anxiety sensitivity strength score. In other words, those women who had the baseline period over those eight weeks showed a really strong connection between the changes in hormones and the change in anxiety symptoms. Those women appeared to show the greatest benefit of transdermal estradiol in terms of their post-randomization anxiety levels.

We could not explain this effect by either any kind of treatment effect on the stress reactivity or by modulation by stressful life events. We did explore in the preliminary fashion with post-hoc analyses, what else do we see in our data set that might explain this greater beneficial effect of transdermal estradiol in women who were sensitive to estradiol fluctuations in terms of their anxiety.

We found that absolute parallel finding if we looked at somatic symptoms based on the Greene Climacteric Scale that has a subscale for somatic symptom bother. How much are you bothered? How much impairment is there associated with their somatic symptoms? These findings parallel with what I just showed you for anxiety. The subscale of the Greene Scale that measures somatic symptoms includes reports of dizziness, feeling faint, breathing difficulties, pressure or lightness in the head, numb and tingling in body parts. That comprises the somatic scale of the Greene and those are exactly very similar symptoms that are associated with clinical anxiety.

It is speculative but we speculate that one mechanism by which transdermal estradiol may be particularly effective or beneficial in women who are sensitive to hormonal flux and anxiety is by beneficially modulating the somatic symptoms of that anxiety.

To wrap things up, we and others have shown that estradiol variability predicts depression and in our most recent study, preliminary results granted, we also show that it predicts anhedonia and anxiety symptoms, which are transdiagnostic. This relationship appears to be strengthened. We show this in two studies now by stressful life events.

We also showed in this most recent study that the variability predicts a specific dysregulation in hypothalamic pituitary and IL-6 for inflammatory responses to stress. But whether that stress response dysregulation serves to mediate or serves to explain the relationship between estradiol variability and affective symptoms. We were unable to test in the study because of the early termination of the research due to the pandemic. We did not have the statistical power to test for that mediation, which would be an important next step.

We and others have shown that transdermal estradiol is effective in treating current depression and maybe effective in preventing the emergence of depressive symptoms of perimenopausal women. And that most stressful life events and sensitivity to this hormonal flux may in fact be predictors of the beneficial response to transdermal estradiol in terms of affective syndrome.

Wouldn’t it be cool if items from the clinical interview such as stressful life event exposure, sensitivity to hormonal flux over the course of a woman’s reproductive life and somatic symptoms could actually be predictive of beneficial response to interventions and therefore advance our precision medicine of treating perimenopausal mood disorders.

Thank you very much. It has been a pleasure to be with you all today.

LAURA ROWLAND: Thank you so much, Dr. Girdler and Dr. Parry. Now, we are going to start the Q&A session. Dr. Parry, could you please turn on your camera and unmute yourself?

BARBARA PARRY: I cannot turn on my video because the host has stopped it.

LAURA ROWLAND: We will get that fixed. We have lots of questions in the Q&A box. What I am going to do is I am going to start with some methodological questions that came up. First, I am going to ask Dr. Parry a couple of questions. One is with the bright light therapy, are there any adverse effects with the eyes?

BARBARA PARRY: Some women may get some eye strain. They feel like it is too close to them. We can be flexible and have them instead of being a foot and a half away from the light box, move further away or do 20 minutes instead of 30 minutes. But that has not really been a major problem.

LAURA ROWLAND: Great. That is wonderful to hear.

The other methodological question was, were your controls age matched to the depressed groups. And a second was, did you use standard deviation or standard error bars in your graphs.

BARBARA PARRY: We used standard error bars in the graphs. The depressed versus normal controls were selected on symptomatology and in relation to their FSH. We looked at whether there was – was age as covariate – not significant differences between the groups.

LAURA ROWLAND: Wonderful. Thank you so much.

Turning to Dr. Girdler, there was a question about what instrument or measures you used to assess dysphoria or depression. I think you mentioned it briefly. Could you mention it once again for the audience?

SUSAN GIRDLER: Absolutely. The primary measure was the Center for Epidemiologic Studies Depression Subscale, the CES-D. It is probably one of the most commonly used measures of depression in the perimenopausal research studies. But then we also in our most recent study used other scales to measure specifically the symptoms of anhedonia and anxiety. We used the Snaith-Hamilton anhedonia scale. And we used the Spielberger State-Trait Anxiety Inventory.

LAURA ROWLAND: Great. Thank you so much.

Dr. Parry, there was a question regarding the improvement effects. You measured up to eight weeks, I believe. How long do you think these effects can be sustained?

BARBARA PARRY: We tried to do a follow-up study for three months afterwards. But we did not have the funding to pay for that and many of the participants did not continue. We could offer them to come back and have the other treatment if it did not work. But we are trying to do that now but with limited resources.

But the advantage is you can always restart this. You do not have to – we do that for other – the data, for example, in other mood disorders we have followed mood disorder patients and find that these interventions last for at least up to nine months afterwards.

LAURA ROWLAND: That is great. You did mention that this is something that could be repeated as well. I look forward to additional research on this.

Dr. Girdler, there has been a few comments regarding hormone replacement therapy. Many clinicians are apprehensive of prescribing hormone replacement therapy. Do you think that your results of your study will help improve the prescription of hormone replacement therapy? Perhaps, discuss other barriers.

SUSAN GIRDLER: Sure, sure, and I know Dr. Parry could very much contribute to this discussion as well too. I invite her to make comments after I make a few.

We certainly know that subsequent to the original release of the findings from women’s health initiative that scared everybody to death that if you gave women hormones, they were going to develop breast cancer and they were going to have heart attacks and strokes and all of that. We know there is a lot to unpack from that study. And what we did learn from that is that giving older women who have been many years from their last menstrual period or women who – a lot of the women in that study were also either overweight or obese had established hypertension, had a history of hormone exposure, which we know can increase our risk for breast cancer. In that sample, certainly we learned a lot about maybe the type of women who would not be good candidates for hormones.

But there is so much with secondary analyses of those same data, we know that there is this timing hypothesis that women who start hormones early in the menopause transition and those were the women who were always intended for hormones. In fact, there are decades of observational evidence that show that women who were prescribed hormones in the menopause transition by their clinicians, they showed a 50 percent reduction in risk of coronary heart disease and all-cause mortality.

In fact, that subgroup of the women’s health initiative if you looked at the younger women in their 50s, they also showed cardiac protection. They showed a reduction in cardiac events with the active estrogen.

But it is a complicated question because it depends on timing where a woman is with respect to her last menstrual period. The effects of estrogen are beneficial if started early or detrimental potentially if started later. The route and the dose of the hormones are very important so transdermal estradiol, the skin patch, for example, which we use in our studies. We know that that is the recommended route for particularly women who are at high metabolic risk. It is also associated with much lower risk of embolism or blood clots. The route is important. The dose is important. The age of the woman is important. Her medical health is important. Her family history is important. All of these decisions should be made personally with your health care provider.
But I want to also point out that the benefits – there are risks. There is no hormone that is completely safe. But we also want to consider what are the benefits. We know that there is tremendous benefit for hip fraction, other bone fractures associated with using hormones – effective for sleep – Dr. Parry – relevant to all your work. It helps sleep. It improves sexual functioning as women age. It is really a matter of a risk benefit analysis to determine for which women and that is what some of our research is trying to look at. Which women will show the greatest benefit of using hormones? But it is definitely complicated. A lot of factors to consider.

But as David and I wrote a paper – I give him credit – David Rubinow and me. Don’t throw the baby out with the bath water because there is a lot of benefit if it is done in an informed fashion in terms of using hormones.

Barbara, would you like to add anything to that?

BARBARA PARRY: Very comprehensive answer. I just underscore the timing, the early menopause versus ten years after and the other -- medical conditions also. Absolute versus relative risk. The absolute risk in 4 in 10,000 women. I think that – take into consideration.

And also, it is the estrogen versus the progesterone too that in the women who had hysterectomies and could be given estrogen alone. We did not see the risk factors.

That spurs a question if I may. You had to give the progesterone to prevent the endometrial hyperplasia. Did you notice any significant effects on mood during the short time period you had to give the progesterone?

SUSAN GIRDLER: That is a good question. No, we did not. We used oral micronized Prometrium, which is a bioidentical. It is a different progesterone than was used in the Women’s Health Initiative. We only gave it for ten days, every two to three months. We did not notice – I have to be honest – we were not really focused on that. There was nothing evident in terms of – we were careful to do most of our measures kind of disentangled from the timing of the progesterone.

SUSAN GIRDLER: Barbara, you are the clinician. Would you like to take that one?

BARBARA PARRY: I think earlier the better. And some gynecologists often may put women on oral contraceptives during the transition although they are higher dose.

As Dr. Girdler pointed out early on, I think early in the menopausal transition is better. This seems to be protective for at least the next ten years. And then paying attention to the other effects of health, exercise, and sleep.

But there is supportive evidence that the antidepressants work better when estrogen is given with them. That the antidepressants alone in perimenopausal women and early postmenopausal women are not as effective. They do not have the same downregulation effect on the serotonin receptors postsynaptically. That can be an additional treatment.

The other additive treatment that seems to be very helpful is thyroid replacement, very low dose thyroid replacement, which can enhance the effect of antidepressant and estrogen and thyroid work synergistically.

LAURA ROWLAND: Interesting. Thank you so much. Again, it is just not one particular sign. It is not measuring a hormone level. It is really early in the perimenopausal transition period. Thank you.
Do you have any questions for each other before we close this session?

No. Okay.

I will end by asking one question from our audience member. What is the take-home message for the clinicians listening to these talks? Do you have any last remarks?

BARBARA PARRY: I will step forward if I can. I just think that these sleep and light interventions have been overlooked. They are very widely used in Europe. I think that for the often circadian rhythm disruption that occurs whether it is during the menstrual cycle, pregnancy, postpartum or menopause, I think these are rapid-acting treatments that have been very well tolerated and with few side effects. It can be done at home. Done by a paraprofessional. I think it is certainly worth a try. They do not have the side effects of medications or hormonal treatments. I think they have been overlooked and it is certainly dependent on the circadian phase.

Light in the middle day is not necessarily that effective unless you have patients with bipolar illness who both light and sleep interventions can induce mania. Then light in the middle of the day is important. But light at dawn and dusk and shifting rhythms with light and sleep I think has an important potential therapeutic – it should be part of our tools, I think. We are trying to move it out into the community with that in mind.

SUSAN GIRDLER: I totally agree that it should be part of our tool – incredible work that you are doing and done, Dr. Parry, over your career.

I might just add that I think a lot of Dr. Parry’s work and what we are trying to do is really meant to ultimately inform clinicians about how they might think about precision medicine or individualized approach to treating women in midlife in terms of depression. Interviewing, finding out what else is going on in their lives, what stress are they experiencing.

And also, I think it is worth the time to – if you really want to know is this perimenopausal depression, is this really probably triggered by the hormone fluctuations before you decide to prescribe transdermal estradiol, doing a history in terms of when did symptoms onset. When did anxiety or depressive symptoms onset? Were they correlated with changes in menstrual bleeding or has there always been sort of history of depression and anxiety and now because there is also greater stress such as empty nesting and death of loved ones and divorce and all these horrible things that happen to us midlife women. Are those events more of an explanation? I think it is a matter of just really some careful clinical interviewing to help inform a treatment approach for any individual woman.

BARBARA PARRY: I would just like to echo the fine work of Dr. Girdler and bringing the stress picture in and how that interdigitates with past history with depression. I think that is a major contribution to the field.

LAURA ROWLAND: Thank you. I want to thank you both for these wonderful presentations. We look forward to seeing what you produce in the future. On that, the Q&A session has ended.

I want to remind the audience of the upcoming seminars in the series. We have four upcoming seminars in the near future. Please go to to register for them.

Next slide please, I want to thank everybody for attending and for programmatic questions and information on webinar recordings, please contact us at Thank you again.