Harmonizing research outcomes for polycystic ovary syndrome (HARP), a marathon not a sprint: current challenges and future research need

,


Introduction
Caring for women with polycystic ovary syndrome (PCOS) has progressed over the last few decades thanks to an increasing, still shy, investment in clinical research and evidence-based medicine (Azziz et al., 2019). What was a vaguely reported syndrome by Stein and Leventhal in 1935 has transformed today into a common endocrine condition attracting the attention of women, health professionals and researchers worldwide (Azziz and Adashi, 2016). While certain aspects of the PCOS pathophysiology remain unknown (Dumesic et al., 2015), investing in research helped to substantially improve the care for women with PCOS . Examples include the ESHRE/ American Society for Reproductive Medicine standardized PCOS diagnostic criteria, which not only helped to improve clinical care but also enabled a more homogeneous amalgamation of research findings (Fauser et al., 2012). Public investment in large randomized trials and their meta-analyses helped to regulate the use of insulin sensitizers and ovulation stimulation agents to reduce the risk of adverse outcomes in women affected by PCOS (Wang et al., 2017;Sharpe et al., 2019). Epidemiological studies helped to crystallize the risk of endometrial cancer in this group and prompted the introduction of preventative treatments into routine care (Lauretta et al., 2016;MacKintosh and Crosbie, 2018). However, aspiring beyond the status quo, a major paradigm shift is needed to improve research quality and bridge the current knowledge gaps.
Several interventions helped the scientific community to improve the quality of research over the last 20 years such as prospective trial registration (DeAngelis et al., 2005), standardized reporting checklists (Begg et al., 1996) and mandatory disclosure of conflict of interest (Blum et al., 2009). Still, increasing research efficiency remains a major challenge yielding a recurrent wastage in health research outputs. A problem estimated to consume as much as 85% of the global research in biomedical and applied clinical sciences yearly budget (Macleod et al., 2014). Clearly, PCOS research is not immune to this issue.
The recent international guideline on the diagnosis and management of PCOS included 40 systematic and 20 narrative reviews to address 60 prioritized clinical questions . It generated 31 evidence-based recommendations where sufficient evidence was available to inform the guideline development group and 57 clinical consensus recommendations where, in the absence of sufficient evidence, a consensus was needed; highlighting the need for more evidence synthesis to inform practice.
To address this expressed need, as we start a new decade of research, we highlight some of the key challenges and suggest a roadmap to increase efficiency, reduce wastage and increase the impact of PCOS research on women's health.

Current challenges
Poor engagement of key stakeholders PCOS is unique in its multi-systemic effects and varied clinical presentation. As such, affected women were often cared for by health professionals within different specialized teams. Yet, for many years this segregated approach not only led to disharmonious and fragmented care but also produced isolated research efforts neglecting the lifelong impact of PCOS on women's wellbeing (Tay et al., 2018). To date, more than 14 evidence-based guidelines and consensus statements have been published on the diagnosis and management of PCOS, but only two of those guidelines adopted a lifelong view to PCOS in consultation with patient representatives (Teede et al., 2011. Such a segregated approach further hinders the translation and implementation of evidence into practice, leading to a persistent knowledge gap  and poor patient experience . Until recently, the dialogue between lay consumers and their caring professionals was largely absent from the research cycle, often pushing PCOS researchers astray from real patients' needs (Teede et al., 2014). For example, engaging patient representatives and lay consumers in our recently published core outcome set (Al Wattar et al., 2020) emphasized the importance of reporting on the psychological impact of PCOS on women's mental health, outcomes that were traditionally poorly reported in most published trials.
Going forward, the practice of silo research should be avoided. This can be achieved by investing in setting up joint research efforts within multidisciplinary teams to continuously refine research priorities (Legro et al., 2006), engage all key stakeholder groups including lay consumers (Al Wattar et al., 2020) and promote a lifelong approach in PCOS research (Weiss and Bulmer, 2011).

P-hacking, heterogeneity and varied outcomes reporting
Selective reporting on surrogate outcomes with a 'significant' P-value is a common malpractice in health research permeating to the pages of leading medical journals (Chuard et al., 2019). This is particularly prevalent in PCOS outputs mostly in small single-centre studies reporting on outcomes of limited significance . Several research initiatives and journals are now promoting the practice of transparent and meaningful reporting on key outcomes of importance in clinical studies (Williamson et al., 2011;Khan and O'Donovan, 2014). While this could help to improve the quality of evidence synthesis, systematic reviewing and meta-analyses, effective implementation in practice remains an important obstacle. For example, in a current systematic review (CRD42020186571), we identified 59 randomized clinical trials that evaluated lifestyle interventions (dietary, physical, behavioural, pharmacological and their combinations) in women with PCOS. Yet only 12 of these trials reported on both BMI and weight changes, 16 reported exclusively on BMI, 10 exclusively reported on weight and 18 (1084 women) reported on neither. Such selective reporting is yielding a net data loss towards evidence synthesis and increasing wastage (Fig. 1).
The harmonizing research outcomes for PCOS (HARP) core outcome set was published as an international effort to address this issue, promoting higher research homogeneity (Al Wattar et al., 2020). While a positive step, its adoption and impact in practice are yet to be determined. Searching clinicaltrials.gov, we identified 29 ongoing trials registered since the HARP core set was published: we call upon their investigators to adopt this core outcome set when reporting their findings (Fig. 2). We do acknowledge that some trials might not be able to collect all of the core outcomes. However, the HARP set should provide a strong incentive to argue for adequate funding to adopt a standardized reporting of those core outcomes. Of course, other outcomes outside the core set should also be further explored and  sample size (power) calculations should not be based on a core outcome exclusively. Finally, we acknowledge that standardized definitions and measurement tools for some of those core outcomes are lacking. More consensus work is needed among stakeholders to address this research need.
Limited scope for data sharing and joint research efforts As we enter into the third age of computing innovation (Miailhe and Hodes, 2017), a remarkable shift towards digitalized healthcare and healthinformatics is taking place worldwide (Herland et al., 2014). Harvesting this emerging computing power to enable large scale 'big data' health research is all but inevitable (Yang et al., 2017). Yet, this practice remains scarce in PCOS research. The benefits of investing in a platform to enable big data research are numerous and specific to PCOS. Put simply, current means are unable to address several key questions such as the differential effect of different treatments across the PCOS phenotypes and the epigenetic expression of PCOS manifestation across generations and families (Legro et al., 2006). Furthermore, as treatment algorithms get more complex with many competing treatment options and higher patient expectations for increased safety, it would be naïve to expect randomized controlled trials (RCTs) to address such large questions efficiently. For example, direct evaluation of the effectiveness and safety of all available ovulation stimulation agents to reduce the risk of ovarian hyperstimulation syndrome in women with PCOS would require a six arms multicentre RCT, not counting treatments' combinations with huge sample size, a near-impossible task at least in the near future. Capturing long-term safety outcomes (e.g. stroke and thrombotic events) is particularly difficult in RCTs but could be recorded efficiently using digitalized health records. Therefore, serious investment in collaborative efforts to enable large scale cohort studies (Azziz et al., 2019) and quality big data research is a priority to address those impenetrable, yet important, questions.

Research need
A proposed roadmap to address the above challenges and remedy future research need could include the approaches outlined below.

Promote patient and public involvement in future research
A key priority is to establish a true partnership between the academic and lay community to deliver impactful research that directly addresses patients' health needs. This entails involving lay consumers in the evidence synthesis and translation ecosystem. Women's needs and preferences should advise future research design, resources allocation and evidence dissemination strategies via multi-faceted, multi-modal communication channels. Therefore, there is a need to invest in training both health researchers and lay partners on patient and public involvement (PPI) methodology (Dudley et al., 2015). While several generic resources exist (Bagley et al., 2016), a PCOS specific PPI toolkit is needed.
Adopt a multidisciplinary lifelong approach to new research All new treatments and preventions for PCOS will have implications on the wellbeing of affected women encompassing their endocrine, reproductive, and mental health. Therefore, new trials should engage representatives from all stakeholder groups to plan and report their findings keeping in mind the lifelong impact of PCOS on women's wellbeing. International professional societies and women health regulators should invest in disseminating new research across their membership to boost collaboration across disciplines and facilitate the participation of all relevant stakeholders beyond geographical boundaries. Establishing such an international cross-disciplines collaboration platform could boost the participation of traditionally absent stakeholders in PCOS research.

Continuously refine research needs and priorities
With ever finite research resources, there is a need to continuously refine the current research need, assess the knowledge gap, guide future research efforts and reallocate resources judiciously. As the findings of existing initiatives are likely to become obsolete (Legro et al., 2006), investing in an efficient platform to produce regular updates is needed. This is particularly relevant to translational research where experimental treatments could be evaluated and introduced into practice, safely bridging the gap between experimental and applied clinical research.

Standardize outcome definitions and measurement tools
To date, the definitions of several key clinical outcomes remain elusive such as insulin resistance and eating disorders. Similarly, key outcomes, such as depression and anxiety, are commonly measured and reported using several measurement tools, which hinders evidence synthesis. Consensus work is needed among all stakeholders (academic, clinical, industry and lay groups) to harmonize these elements and promote excellence in outcomes reporting.

Invest in data sharing infrastructure
Innovation is key to reap the benefits of big data research (Rothstein, 2015;Sharon, 2016). Investing in a common platform is needed to facilitate and standardize data collection and harmonization. Initial consensus steps could include standardized recommendations for enhanced routine data collection, unified data time points and a shared data dictionary for PCOS core outcomes' definitions and characteristics. This should be complemented with guidance on joint international efforts to govern ethical reporting, safeguard co-investigators and empower lay consumers.

Conclusion
Collaborative efforts are needed to improve the quality of PCOS research conduct and increase its impact. The HARP initiative aims to address the identified challenges and deliver the proposed solutions to