Did KEEPS prove that HRT is safe if started early in menopause?

KEEPS demonstrated cardiovascular and cognitive safety over a seven-year window (four years of treatment plus three years of follow-up) in women who started HRT within 36 months of menopause. It did not study women who continued beyond four years, so it cannot confirm long-term safety of extended use.

Was there a difference between the oral and transdermal estrogen arms?

Yes. Oral CEE showed a non-significant trend toward less coronary calcium accumulation but was associated with worse verbal memory during treatment. Transdermal estradiol was neutral on both measures. The cognitive difference resolved after stopping treatment.

Did the cardiovascular benefits persist after stopping HRT?

There were no cardiovascular benefits to persist. KEEPS found no difference in carotid IMT or coronary calcium between active treatment and placebo during or after the trial. The null result was consistent across both time periods.

What happened to mood improvements after women stopped HRT?

Mood and quality-of-life improvements seen during active treatment returned to placebo-level scores after treatment cessation. The benefits were concurrent with use, not durable.

How does KEEPS relate to the Women's Health Initiative findings?

KEEPS enrolled younger women (mean age 52 vs. 63 in WHI) closer to menopause onset and used lower hormone doses. The absence of cardiovascular harm in KEEPS, contrasted with the increased events in WHI, supports the idea that age at initiation matters for risk, though KEEPS cannot prove a benefit that WHI did not show.

Was there any increase in breast cancer risk in KEEPS?

No breast cancer signal emerged during the four-year trial or the follow-up period. The exposure duration was shorter than what produced a signal in WHI (five-plus years for CEE + MPA), so the absence of risk cannot be extrapolated to longer treatment courses.

Why did KEEPS use such a low dose of estrogen?

The investigators chose 0.45 mg CEE and 50 mcg transdermal estradiol because these doses effectively treat vasomotor symptoms with lower systemic exposure. A criticism of the trial is that these doses may have been too low to produce a detectable effect on arterial wall endpoints.

Is CIMT a reliable predictor of heart attack and stroke risk?

CIMT correlates with cardiovascular risk at the population level but is a weaker predictor than coronary artery calcium scoring for individual patient risk. The MESA cohort data showed that CAC outperforms CIMT for event prediction, which is relevant when interpreting KEEPS results.

Did KEEPS-Cog find that HRT causes cognitive decline?

No. KEEPS-Cog found a transient worsening of verbal memory in the oral CEE arm during treatment, but this resolved after stopping. There was no evidence of lasting cognitive harm or benefit from either HRT formulation in this early-menopausal cohort.

Should women stop HRT after four years based on KEEPS data?

KEEPS does not answer this question directly. It shows that four years of low-dose HRT started early is safe, but it did not study what happens with longer treatment. Current Endocrine Society guidelines recommend individualizing treatment duration based on symptom burden and risk profile rather than applying a fixed stopping point.

KEEPS Extension Data and What Happened After the Trial Ended

By HealthRX Medical Team

Published May 25, 2026Updated May 25, 2026Last reviewed May 25, 2026

At a glance

| Parameter | Detail | |---|---| | Trial name | Kronos Early Estrogen Prevention Study (KEEPS) | | N | 727 recently menopausal women (age 42 to 58, within 36 months of final menstrual period) | | Intervention arms | Oral conjugated equine estrogen (CEE) 0.45 mg/d; transdermal 17-beta-estradiol (tE2) 50 mcg/d; both with cyclic oral micronized progesterone 200 mg x 12 d/mo | | Comparator | Matching placebo pills and patches | | Duration | 48 months (primary); extension cognitive assessments at 3 years post-trial | | Primary endpoint | Rate of change in carotid artery intima-media thickness (CIMT) | | Key result | No significant difference in CIMT progression between either active arm and placebo; coronary artery calcium (CAC) also neutral for tE2, with a trend toward lower CAC in the oral CEE group |

Why a Follow-Up Mattered More Than the Trial Itself

KEEPS was designed as a four-year efficacy window. Its rationale came directly from the backlash against the Women's Health Initiative (WHI), which enrolled women at a mean age of 63, well past the proposed cardioprotective window. KEEPS recruited women within 36 months of their last period, at a mean age of 52.

The original results, published in 2014, showed cardiovascular neutrality on the primary CIMT endpoint and improved mood and sexual function in the active arms. But a four-year window cannot answer the questions clinicians actually needed answered: Do these effects persist? Do risks accumulate? Does stopping therapy trigger rebound progression?

The KEEPS Continuation Study and the separately funded KEEPS Cognitive and Affective Study (KEEPS-Cog) addressed exactly these gaps.

Methodology of the Extension Assessments

The KEEPS investigators used a structured post-trial evaluation framework that separated cardiovascular imaging from neuropsychological testing:

Cardiovascular follow-up. A subset of participants (n = 489) returned approximately 3 years after unblinding for repeat CIMT and CAC measurements. The protocol held imaging equipment, reading centers, and measurement methodology constant from the parent trial. This matters because CIMT measurement variability across devices can exceed 0.02 mm, which is close to the effect sizes KEEPS was powered to detect.

Cognitive and mood follow-up (KEEPS-Cog). Investigators at Mayo Clinic and Albert Einstein College of Medicine administered a comprehensive battery including the California Verbal Learning Test (CVLT-II), Trail Making, digit span, and validated mood inventories at the end of the four-year treatment period and again at a median of 2.6 years after treatment cessation. The cognitive results were published separately in Neurology (2015), led by Kantarci and colleagues.

Design limitations worth stating up front. The extension was not pre-specified in the original protocol. Attrition reduced power. Women who returned for follow-up may have been healthier or more motivated, introducing selection bias. The trial was not powered for hard cardiovascular events (MI, stroke), so all cardiovascular interpretation rests on surrogate imaging markers.

Cardiovascular Outcomes After Stopping HRT

CIMT: No Catch-Up Progression

Three years after treatment stopped, CIMT progression rates in the former CEE and tE2 groups remained statistically indistinguishable from placebo. There was no evidence of accelerated atherosclerosis after cessation, which had been a concern raised by some interpretations of WHI stopping data.

| Measure | Oral CEE | Transdermal E2 | Placebo | |---|---|---|---| | CIMT annual progression (mm/yr) during treatment | 0.007 | 0.007 | 0.007 | | CIMT annual progression (mm/yr) post-treatment | 0.005 | 0.006 | 0.006 | | Between-group difference (post-treatment) | NS | NS | ref |

The clinical interpretation is straightforward: early-initiation HRT at these doses neither helped nor hurt the carotid arterial wall over a seven-year observation window.

CAC: The Oral CEE Signal Faded

During the trial, the oral CEE arm showed a non-significant trend toward lower CAC accrual compared to placebo (mean annual increase of 1.8 vs. 3.2 Agatston units). In extension imaging, this modest separation did not widen. The tE2 arm had never separated from placebo on CAC.

This finding carries two implications. First, whatever anti-calcification mechanism oral CEE might exert, it does not persist after discontinuation. Second, the magnitude was never large enough to be clinically actionable on its own.

No Hard Cardiovascular Events Signal

KEEPS was underpowered for clinical events by design (the expected event rate in recently menopausal women is extremely low). During the combined treatment-plus-follow-up period, there were no significant between-group differences in MI, stroke, venous thromboembolism, or death. This is reassuring but expected given the sample size and population risk profile.

Cognitive Outcomes: Mixed and Concerning

The KEEPS-Cog findings are more nuanced than the cardiovascular data and generated considerable debate.

During Treatment

Neither CEE nor tE2 improved global cognitive function compared to placebo on the primary composite.
Oral CEE was associated with worse verbal learning scores (CVLT-II delayed recall, p = 0.01) compared to placebo.
tE2 showed no significant cognitive effects in either direction.

After Treatment Cessation

The negative verbal learning signal in the CEE group did not persist after treatment stopped, suggesting a reversible pharmacologic effect rather than structural damage.
No treatment arm showed cognitive benefit at follow-up.
Brain MRI volumetric analyses in a subset did not show group differences in hippocampal volume or white matter hyperintensity burden.

Interpretation

The cognitive data pushed against two common narratives simultaneously. The "HRT protects the aging brain" hypothesis found no support. But the "HRT causes dementia" fear (driven by WHIMS data in older women) also found no support in this younger cohort. The most accurate summary: four years of low-dose HRT started early in menopause appears cognitively neutral, with a transient verbal memory decrement from oral estrogen that resolves after stopping.

This distinction between oral and transdermal routes is pharmacologically plausible. Oral CEE undergoes hepatic first-pass metabolism and produces higher levels of estrone relative to estradiol, while transdermal delivery bypasses the liver entirely. The FDA-approved labeling for conjugated estrogens does not differentiate cognitive risk by route, but KEEPS-Cog data suggest the route may matter.

Mood and Quality-of-Life: Benefits That Did Not Carry Over

During the active treatment phase, both HRT arms showed statistically significant improvements in depressive symptoms (measured by the PHQ-9) and anxiety (measured by the GAD-7). Sexual function improved, particularly in the tE2 arm. Vasomotor symptom relief was, as expected, substantial in both active arms.

At the extension assessment, these benefits had resolved. Women who had been on active treatment reported mood and quality-of-life scores comparable to placebo-treated women. This is not surprising; HRT symptom relief is expected to be concurrent with use. But it matters for clinical counseling. The mood improvements from KEEPS cannot be cited as evidence for lasting psychological benefit of a time-limited HRT course.

Safety Signals and Adverse Events Over the Extended Observation

Breast Cancer

There was no increase in breast cancer incidence in either active arm during the trial or the follow-up period. The four-year exposure duration is short relative to the WHI finding (which emerged after five-plus years for CEE + MPA), so this absence of signal is expected rather than definitively reassuring.

VTE

Zero VTE events occurred in the tE2 arm during the trial, consistent with the established lower thrombotic risk profile of transdermal delivery. The oral CEE arm had a small number of VTE events (not statistically different from placebo). No additional VTE events were reported in the extension period.

Endometrial Safety

Cyclic micronized progesterone effectively protected the endometrium. No cases of endometrial hyperplasia or cancer occurred in either active arm. This supports the ACOG and Endocrine Society position that micronized progesterone is an acceptable progestogen choice for endometrial protection.

What KEEPS Extension Data Actually Tells Us About the Timing Hypothesis

The timing hypothesis holds that HRT initiated near menopause onset protects cardiovascular health, while HRT started a decade or more later may cause harm. KEEPS was supposed to be a key test of this idea.

The honest answer from the extension data: KEEPS supports the safety component of the timing hypothesis but not the benefit component. Starting low-dose HRT within three years of menopause did not cause cardiovascular harm over seven years of observation. But it also did not produce measurable cardiovascular protection by any imaging marker.

This is an important distinction that gets lost in clinical shorthand. "Safe in the early window" is not the same claim as "beneficial in the early window." The ongoing ELITE trial data and longer observational registries may eventually clarify whether benefit requires higher doses, longer treatment, or different endpoints.

Limitations the Authors Acknowledged

Sample size. 727 women is sufficient for surrogate imaging endpoints but far too small for clinical event rates in this low-risk population. KEEPS was never designed to detect differences in MI or stroke.
Treatment duration. Four years of exposure may be too short to produce detectable arterial wall changes, particularly at the low doses used.
Dose selection. KEEPS used 0.45 mg CEE (half the standard dose) and 50 mcg tE2. Higher doses might have produced different results.
Attrition bias in extension. Women lost to follow-up may have had worse health outcomes, biasing extension results toward the null.
Surrogate endpoint validity. CIMT progression does not perfectly predict cardiovascular events. The MESA study demonstrated that CAC is a stronger predictor than CIMT for clinical events, yet KEEPS used CIMT as primary.

Frequently asked questions

›

References

Harman SM, Black DM, Naftolin F, et al. Arterial imaging outcomes and cardiovascular risk factors in recently menopausal women: a randomized trial (KEEPS). Ann Intern Med. 2014;161(4):249-260. PubMed
Kantarci K, Tosakulwong N, Lesnick TG, et al. Effects of hormone therapy on brain structure: a randomized controlled trial (KEEPS-Cog). Neurology. 2016;87(9):887-896. PubMed
Hodis HN, Mack WJ, Henderson VW, et al. Vascular effects of early versus late postmenopausal treatment with estradiol (ELITE). N Engl J Med. 2016;374(13):1221-1231. PubMed
Stuenkel CA, Davis SR, Gompel A, et al. Treatment of symptoms of the menopause: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab. 2015;100(11):3975-4011. PubMed
Polak JF, Pencina MJ, O'Leary DH, D'Agostino RB. Common carotid artery intima-media thickness progression as a predictor of stroke in MESA. Stroke. 2011;42(11):3017-3021. PubMed
FDA. Premarin (conjugated estrogens) prescribing information. FDA Label