This hypothesis-generating post-hoc analysis on a small sample of young DMD boys, median 11 years of age, shows that tamoxifen over 48 weeks was well tolerated and suggests that tamoxifen might contribute to a better preservation of echocardiographically estimated cardiac dimensions and systolic function in DMD patients, so that confirmatory studies appear justified. Indeed, in front of an otherwise progressive natural disease course, a trend towards decreased left-ventricular dilation and a stabilised left-ventricular systolic function appeared to exist among children receiving tamoxifen for 48 weeks as compared to children receiving placebo. Given the small sample size of this post-hoc analysis, statistical comparisons were not meaningful and cannot be relied upon.

Nevertheless, these preliminary, exploratory results are novel and may entail important clinical implications. Indeed, cardiac disease has become the main life-limiting condition in DMD [8]. Because of the lack of DMD-specific drugs, present strategies for established DMD-related cardiomyopathy are based on treatment for adult heart failure [8], while available evidence is unfortunately still limited in paediatrics [16,17,18,19,20]. Improving cardiac management of DMD patients is therefore critical and urgently needed.

It is important to interpret the results of this preliminary experience and post-hoc analysis with an understanding of its inherent limitations, which are a direct consequence of its design. First, this post-hoc analysis grounded on obtainable data: the study was not designed to investigate the potential effect of 48 weeks of tamoxifen on echocardiography. Second, the available sample was small, translating into a very low power. Missing data hampered the analysis even further and mandated an exploratory, graphical approach. Within-group pre/post comparisons are reported for completeness but with full awareness that they are statistically impossible to interpret. Third, clinical symptoms and signs before and after treatment could not be compared, and NT-proBNP was not measured. Fourth, echocardiography was performed as per clinical routine in different local hospitals and collected measures were scanty (ejection fraction was not available as per the gold standard Simpson’s biplane method, diastolic function was not assessed), and the images could not be double-checked in a blind fashion by an independent paediatric cardiologist. Furthermore, z scores for LVEDd and LVEDs were not always documented and, weight and height not having been systematically collected at each echocardiography, they could not be reliably calculated at baseline and last available echocardiography, so that we had to rely on measurements of absolute dimensions. However, given the limited echocardiographic windows in DMD patients, fractional shortening (available in our database) has been proposed as a reliable LV systolic function measure in this population [21]. Fifth, it is meanwhile well-known that symptoms (because of reduced physical activity in DMD patients), NT-proBNP, and standard echocardiography are late markers and therefore suboptimal in investigating DMD-associated dilated cardiomyopathy. Advanced echocardiography (including regional LV motion analysis, speckle tracking, and strain analysis) and, especially, cardiac magnetic resonance imaging (assessing dimensions, function, water content, and fibrosis) are currently recommended for early detection of cardiac involvement in studies on DMD-associated cardiomyopathy [22, 23]. Sixth, tamoxifen dose was chosen based on the preliminary results of a previous open-label trial [12], but a solid dose rationale [24], based on current understanding of paediatric pharmacokinetics and comprehension of the purported cardiac effects, is still lacking. Finally, early recognition of cardiac muscle involvement, and standard institution of preventative angiotensin converting enzyme inhibitors from the age of ~ 10 years are the current mainstay of cardiac management in DMD children and adolescents [22]. Despite this, only one out of the 14 included patients was on treatment with lisinopril. Sadly, this is not surprising, and just represents one more example of the therapeutic nihilism that still characterises the current approach to paediatric heart failure [25]. This also implies that the present results might not be automatically generalised to a cohort treated according to current recommendations [22].

Despite all these limitations, the descriptive results we obtained are internally consistent, suggesting the possible existence of a real underlying biological process: LV dimensions (both end-diastolic and end-systolic) mildly worsened in the placebo, and slightly improved in the tamoxifen group, while fractional shortening (as simple but reliable systolic function measure), mildly worsened in the placebo group and minimally improved in the tamoxifen group. Furthermore, this observation is consistent with previous research, pre-clinical data, and pathophysiological considerations. In a mouse model of DMD, tamoxifen reduced cardiac fibrosis by approximately 50% [10], and arrested fibrosis processes are expected to translate into preserved relaxation, contractility, and left ventricular dimensions. Even more relevantly, on monolayers of human-induced pluripotent stem cell-derived cardiomyocytes, the active metabolite 4-hydroxytamoxifen reduced beating rate, raised beating velocity, improved calcium-handling, and prolonged viability [26].

Notably, this post-hoc analysis should be seen as hypothesis-generating [27]. In the drug development journey, we are still far from confirming any clinical effect. However, these preliminary observations suggest that efforts to define a dose rationale, followed by well-powered studies using appropriate endpoints, designed to answer the question of a possible benefit of tamoxifen on cardiac function in DMD patients, may be worth considering. The potential clinical impact on the lives of affected patients and their families might be substantial.