Estrogen Treatment for Urinary Incontinence: Title and subTitle BreakNever, Now, or in the Future?

Deciding whether to treat postmenopausal women with urinary incontinence (UI) with estrogen has become increasingly difficult. The literature is divided but generally supportive. A Cochrane systematic review concluded that “estrogen treatment can improve or cure incontinence,” especially urge UI.¹ Similarly, the International Consultation on Incontinence agreed in 2004 that “it seems reasonable to assume that estrogen therapy may be of benefit for the irritative symptoms of urinary urgency, frequency, and urge UI,” but noted that estrogen had little role in the treatment of stress UI.²

However, in the Cochrane combined meta-analyses of randomized controlled trials, the overall number of participants was small (for example, for the outcome of subjective cure and improvement of UI, the analysis included 179 women assigned to estrogen and 177 women assigned to placebo from a total of 9 trials).¹ Although the large prospective Heart and Estrogen/Progestin Replacement Study (HERS) (n = 1525) found that estrogen plus progestin increased UI rates, these findings could be applied only to women with ischemic heart disease.³ Both the Cochrane collaborators and the International Consultation on Incontinence admitted that specific recommendations could not be made regarding which estrogen formulation to use, optimal route of administration, duration of therapy, and which women with urge UI would be most likely to benefit.

Media coverage about cardiovascular and other risks with estrogen from HERS and the Women’s Health Initiative (WHI) have cooled enthusiasm among clinicians and patients for estrogen use in general—and particularly for UI, for which other treatments are available. To address all of these uncertainties and questions, large, well-designed, randomized controlled trials that could be broadly generalized are needed.

In this issue of JAMA, Hendrix et al⁴ present such a study examining UI outcomes among the thousands of women participating in the WHI. This study has the clear advantages of larger size, multicenter design, and longer follow-up (1-3 years) than most previous randomized controlled trials, and unlike HERS is generalizable to the great majority of postmenopausal women. Their finding of higher rates of UI in women treated with estrogen with or without progestin suggests that estrogen has no place in the UI treatment armamentarium. But is this really the end of the story regarding estrogen treatment for UI? The answers appears to be both yes and no.

The authors evaluated the incidence of new and worsening UI in the 2 WHI studies: conjugated equine estrogen (CEE) treatment alone compared with placebo (estrogen alone trial), and combined hormonal treatment of CEE plus medroxyprogesterone acetate (MPA) compared with placebo (E + P trial). In women who were continent at baseline, estrogen significantly increased the risk of stress UI (CEE + MPA: relative risk [RR], 1.87 [95% confidence interval {CI}, 1.61-2.18]; CEE alone: RR, 2.15 [95% CI, 1.77-2.62]) and mixed UI (CEE + MPA: RR, 1.49 [95% CI, 1.10-2.01]; CEE alone: RR, 1.79 [95% CI, 1.26-2.53]) at 1 year. An increased risk of urge UI was found in the estrogen alone trial (RR, 1.32; 95% CI, 1.10-1.58). Women who were incontinent at baseline developed larger and more frequent amounts of leakage with both CEE alone and CEE + MPA. Subgroup analyses demonstrated that CEE alone and CEE + MPA particularly increased the amount and frequency of UI in older women, and CEE + MPA increased the risk in women who were 15 years or more from menopause. Limitation of activities by UI and the degree of bother of urine leakage also increased in the groups randomized to CEE alone and CEE + MPA. Moreover, the authors show that the risk of UI with CEE + MPA did not diminish over 3 years, although the risk with CEE alone was no longer significant. Incident UI at 1 year persisted for the majority of women (about 70%), regardless of treatment group assignment. The authors strongly conclude that “conjugated equine estrogens with or without progestin should not be prescribed for the prevention or relief of UI.”

One notable aspect of their findings is the high baseline rate of UI in WHI participants (64%), which is at or above the upper limit of median rates found in systematic reviews of UI epidemiology (30%-40% for middle-aged women and 30%-50% for elderly women⁵ ). This raises a concern about possible overascertainment: one reason for this high rate could be the definition of UI that was used (“ever leaked even a very small amount in the past year”). Although the WHI question appears similar to the UI screening question recommended by the International Continence Society⁶ and the International Consultation on Incontinence⁷ (ie, “ . . . leaked any urine in the past X [typically 6 or 12] months?”), the emphasis on ever and even a very small amount could have rendered the WHI question less specific. Indeed, a majority of women incontinent at baseline reported only very mild leakage (none or barely noticeable on underpants in 81%-86% of the women with stress UI and 75%-80% of the women with urge UI). On the other hand, a sensitivity analysis (Table 5 in Hendrix et al⁴ ) demonstrated that altering the frequency threshold in the definition of UI did not markedly affect the RRs for UI in women treated with estrogen.

There is a small chance that some women with UI could have been considered continent (false-negative) if they responded that they “no longer leak[ed] urine” because in fact they were receiving behavioral therapy for UI. Drugs that could be used to treat UI were not found to be associated with the risk of developing stress UI; only over-the-counter phenylpropanolamine, still available for stress UI treatment during the study years 1993-1999, could not be examined.

Detection bias and dropouts are other potential problems in this study. These issues have been discussed in relation to other WHI analyses.⁸ Discontinuation rates of study medications at 1 year in the E + P trial were 9.7% for CEE + MPA and 6.6% for placebo, and in the estrogen alone trial 8.4% for CEE alone and 8.0% for placebo. A major reason for discontinuation in both WHI trials was vaginal bleeding; this could have occurred as well in the estrogen alone trial, although specific reasons for discontinuation have not been reported.⁹ This obvious adverse effect could have led to unblinding of the participants and their primary clinicians to their randomization,¹⁰ with potential effects on UI reporting. Moreover, vaginal bleeding likely would have prompted a participant’s gynecologist to ask about other genitourinary symptoms (including UI) and perform pelvic examinations, thereby leading to increased discussion and detection of UI, particularly in the CEE + MPA group. Such a discussion of UI symptoms by physicians, who often do not routinely ask about UI,¹¹ could have provided validation that led women to report not only new UI but greater severity of existing UI.

Quality of life impact is a key concern with UI, which otherwise has little and infrequent resultant morbidity. Quality of life assessment is increasingly considered to be one of the primary outcomes that should be used in UI treatment trials.¹² In the WHI trials, both CEE alone and CEE + MPA increased the likelihood of worsening UI affecting the degree of bother and limitations in daily activities. However, this effect was attenuated and did not reach significance in younger women (aged 50-59 years). One possible reason is that degree of bother is only one dimension of UI-related quality of life, and limitations in activity (even among older women) may not be perceived as contributing to the effect of UI on quality of life.¹³ It would be interesting to know the impact of incident UI on general health-related quality of life, which was collected in the WHI using the RAND 36-Item Health Survey.¹⁴

The biological plausibility of estrogen causing or worsening UI is another issue raised by this trial. The WHI has illustrated the limitations of using biological plausibility as sufficient evidence of the benefits of estrogen (eg, on cardiovascular disease). Is it the same now for the effect of estrogen on the lower urinary tract and incontinence?

Estrogen is trophic for much of the lower urinary tract in women, and estrogen receptors are found in the vagina, vestibule, distal urethra, bladder trigone, pelvic muscles, and ligamentum rotundum.¹⁵ Changes observed in postmenopausal women include thinning of the vaginal epithelium and unwinding of the urethral submuscosal venous plexus with a decrease in vessel number and vascular pulsations¹⁶ ; increased volume of connective tissue; decreased ratio of proteoglycans to collagen; and decrease in nerve density in the urethral stroma.^{17 - 18} These changes have been assumed to contribute to the observed decrease in urethral closure pressure in older women,¹⁹ and therefore to the development of stress UI, especially that associated with intrinsic sphincter deficiency.

Although the role of estrogen effects in efferent and afferent signaling in the bladder has yet to be extensively investigated,²⁰ low estrogen levels are associated with thinning and irritation of the trigonal area in the bladder (similar to that seen in the vaginal mucosa), which may precipitate urge UI. In addition, recent animal models have shown that ovariectomy causes ultrastructural changes in the detrusor that have been associated with impaired contractility,²¹ and that estadiol treatment in these animals up-regulates detrusor caveolin-1 protein, which could lead to increased calcium fluxes important for contractility.²² Whether these observed changes are due to low estrogen levels, changes in estrogen receptor number and activity, aging per se, coexistent vascular disease, or other age-associated factors has not been determined. However, decreased estrogen and its trophic effect on the lower urinary tract appear to contribute to the development of postmenopausal UI. This being the case, it is difficult to explain why UI would worsen with oral estrogen and why the addition of progestin would mitigate the impact of estrogen on the development of urge UI, as was found in the WHI.

One biological reason for the WHI findings, as the authors note, comes from the study by Jackson et al,²³ who found that 6-month treatment with estradiol caused loss of collagen in vaginal epithelium and did not improve stress UI. However, all of the individuals in that study had urodynamically proven stress UI, which has been associated with alteration in collagen type and cross-linkage.²⁴ Jackson et al postulate that estradiol disrupted already biomechanically altered collagen, leaving open the possibility that estrogens have a role in stress UI prevention, and also the treatment of urge UI.

The issue of total estrogen effect further complicates the arguments regarding biological plausibility. The same dose of exogenous estrogen may have led to striking variations in interindividual serum estrogen levels and end-organ effects, given the wide variation in estrogenic status of postmenopausal women.²⁵ Some of this variability is associated with higher estrogen production by adipose tissue in obese women. There were substantial differences in baseline body mass index between the 2 WHI trials, with higher rates of obesity in the estrogen alone compared with the E + P trial. Thus, women in the estrogen alone trial could have had higher baseline estrogenic status. While the frequency of baseline UI was less in the estrogen alone trial (25.7% had <monthly in estrogen alone vs 31.7% in E + P), overall rates of baseline UI were similar (65% in estrogen alone and 63.5% in E + P), supporting the hypothesis that higher intrinsic estrogen levels mediate UI severity. Such inferences are necessarily limited because of other differences between the groups such as hysterectomy. Body mass index did not affect the incidence of new stress UI in both trials, but obesity tended to attenuate the impact of CEE + MPA and CEE alone on the amount and frequency of worsening UI (Table 6 in Hendrix et al⁴ ). However, there is still insufficient information as to how interindividual variability in estrogen levels are affected by supplementary estrogen in postmenopausal women, especially given the lack of estradiol assays that are applicable and reliable in women taking oral estrogen.⁸

An intriguing possibility in the biological plausibility arguments is that estrogen’s effect on UI is not mediated through the lower urinary tract. A considerable portion of incident UI in both WHI trials occurred in older women. For example, in the E + P trial, 21% of all incident UI occurred in women aged 70 to 79 years and approximately 70% in women older than 60 years. Older women are much more likely to have UI precipitated by comorbid conditions, medications, and functional impairment, many of which occurred at different rates in the treatment and placebo groups, as demonstrated in other published WHI reports.^{9 ,26 - 30} This is a key point that bears further elucidation in the WHI data and elsewhere.

What are the final conclusions from this analysis of UI in the WHI? First, clinicians should no longer prescribe long-term oral conjugated equine estrogens for treatment of urge, stress, or mixed UI in postmenopausal women aged 50 years or older. Hendrix et al have performed an important service by placing UI among the ranks of other significant women’s health problems that warrant formidable organizational, funding, and analysis efforts. Such trials carry enormous impact among both physicians and the public, which can lead to fruitful, if complicated, dialogues about the specific health problems investigated. It would be extremely positive if these trial results prompted women with UI—half of whom never discuss their condition with a physician—to ask their physicians about the many other available treatments for UI. Second, this trial is not the final word on using estrogens to treat UI. Whether topical estrogens might prove beneficial remains unknown, especially on a short-term basis and/or in combination with other therapies. Finally, both the scientific rigor of the WHI trial and the issues it raises should prompt continuing investigation of the basic science of estrogen in normal and aging detrusor function and pelvic floor composition, and lead to further treatment trials (particularly with topical estrogens) that address head on the methodological issues of unblinding, UI characterization (by patients and by physiological testing such as urodynamics), patient targeting, and comprehensive outcomes assessment. With such information, physicians and patients can finally have the answers as to whether it is truly never, now, or in the future for the use of estrogen in the treatment of UI.