Searching for an Elusive Phantom – Targeting Estrogen in Pulmonary Hypertension

The influence of sex hormones in pulmonary arterial hypertension (PAH) is an important and confusing topic. There is a well-described female predisposition to PAH—in particular, in idiopathic or heritable disease—but, paradoxically, once the disease is established, long-term outcomes are better (1). Sex hormones—particularly, estrogen—are clear potential etiological culprits, given the known vasoactive properties (2). Meaningful study in human populations is bedeviled by the correlation–causation debate and, in particular, the additional complications of the cyclical nature of estrogen release, the challenges in teasing out the different downstream metabolism pathways (3), and how they might affect mechanisms about the variation in tissue production versus systemic and the conflicting animal data (4, 5). We have not yet clarified that a nonhormonal mechanism is not in play. It is also entirely plausible that sex hormones are critical to the development of disease but protective of the effects of disease evolution, for example, on the right ventricle and, therefore, directly and paradoxically involved in improved survival. Much of this debate has, until now, been speculative, on the basis of observational data. In this issue of the Journal, Kawut and colleagues (pp. 1143–1151) bring some welcome randomized controlled trial data to the table with the results of the PHANTOM study (6). The trial is notable for a number of reasons. The authors are to be congratulated for conducting one of the largest investigator-driven studies in PAH to date, and especially so for executing it during the pandemic. In a rare disease area where much of the experimental action during the past 2 decades has been industry directed, this network of investigators has shown that well-conducted and practice-informing Phase-2 studies are not out of reach. Studies that are testing out inexpensive, universally available, repurposed, and repositioned therapies remain a rarity. The study is robustly powered, with a plethora of carefully organized secondary endpoints and, as a result, a consistent dataset that is in line with their null hypothesis across a range of efficacy measures. A large effect size for inhibition of aromatase with anastrozole would appear to be ruled out, although the study usefully did not suggest any signal of harm. Therefore, we have, really for the first time to our knowledge, randomized data to suggest that, in the population studied, inhibition of the conversion of androgens to estrogens is ineffective. There are caveats, however, and they are important.

First the “population” question where treatment is circumscribed to postmenopausal women and to men. There were differences between the first smaller study (n = 18) (7) and this one. The first study had slightly more men included (50%) but asymmetrically across placebo (67%) and anastrozole (42%). In the present study, the population had similar rates of underlying Group 1 causes and were more likely to be on triple therapy, with a notably higher mean baseline 6MWT, and both arms over 400 m (442/416 m vs. 378/336 m in the first). This raises the possibility of the ceiling effect for 6-minute walk test (6MWT), but perhaps, when compared with the earlier study, the divergence is most likely simply a Type 1 error. It is notable that the secondary and exploratory endpoints in the first study were not uniform; for example, although 6MWT improved, N terminal pro-brain natriuretic peptide did not significantly change.

Whether anastrozole is the optimal treatment paradigm here is a reasonable question to ask. There are known differences in pharmacological and target engagement profiles to the different triazoles and selective estrogen receptor modulators (8). Despite this, in the area where there is most evidence, early breast cancer, the existing data do not support a clear difference with, for example, data from randomized trials in 3,697 postmenopausal women with hormone receptor–positive early breast cancer showing no differences with regard to anastrozole, letrozole, and tamoxifen in six different treatment schedules (9). The target engagement data are excellent, with a clear and consistent effect on estrogens over the whole study, so we can be convinced that within the confines of a cohort that starts with lower levels to begin with, the drug is doing its job well. It therefore seems unlikely that the lack of a treatment response relates to the specific drug intervention.

An additional and strong feature of the study is the robust 12-month timeframe they chose for study. Most PAH studies leave the longer term follow-up to open-label extension, a valid approach, but having longer term data gives us more time points to observe time courses and speaks better to the sustainability of effects, especially as it has been challenging to link treatment responses at 6 months directly to longer term outcomes (10).

So where does this leave the “estrogens: good, bad … or good and bad debate”? If we stick to the trial data, we can now say that, in postmenopausal women and predominantly older men on dual and triple therapy who have a good baseline function, there appears to be no additional value in blocking estrogens but also no strong signal to suggest that it conversely negatively impacts vascular remodeling or cardiac function. In this population, therefore, endogenous estrogens don’t appear to be playing a very prominent role either way. The jury remains out in the population in which we are arguably most likely to see the biggest effect for either side of the debate: premenopausal women. We have taken a big step forward, but perhaps not to the end of the road.