Moving in the right direction but not there yet: the utility, measurement, and analysis of health status in heart failure trials

This article refers to ‘Quality of life in men and women with heart failure: association with outcome, and comparison between the Kansas City Cardiomyopathy Questionnaire and the EuroQol 5 dimensions questionnaire’ by A. Ravera et al. and ‘Pharmacotherapy for heart failure with reduced ejection fraction and health-related quality of life: systematic review and meta-analysis’ by R.D. Turgeon et al., published in this issue on pages 567–577 and 578–589. Having disease-specific measurements that are valid, reproducible, and sensitive to clinical change, are among the most important considerations when developing novel therapies. For example, the benefits of long-term clopidogrel in patients with coronary disease were largely based on a decrease in non-fatal myocardial infarctions, as measured by elevations in troponin levels.1 These beneficial results would have been difficult to demonstrate using for example lactate dehydrogenase or myoglobin levels as markers to define myocardial infarction. While it may be ambiguous to compare the measurement of disease-specific health status questionnaire to cardiac biomarkers, their developments nevertheless have undergone similar journeys. Given that one of the primary goals in treating patients with heart failure (HF) is to improve patient's health status, including symptoms, function, and quality of life, creating tools to effectively measure health status as an outcome is intuitive. The importance of health status has been recognized by the regulatory authorities who have endorsed them as potential approvable outcome.2 As such, the collection of health status data in clinical trials has become routine. While measuring health status may now be routine, the methodology to do so is evolving. This journey started with generic quality of life measures and now have refined and shortened to disease-specific measures. However, there still remains a variety of instruments being used and a myriad of methods to statistically analyse them. In this issue of the Journal, Ravera et al.,3 compared both a disease-specific [Kansas City Cardiomyopathy Questionnaire (KCCQ)] and a generic [EuroQol 5 dimensions (EQ-5D)] quality of life questionnaire. As has been shown in other HF populations, the KCCQ had better discrimination in predicting future HF events.4 It is noteworthy that similar to other contemporary HF trials ∼3/4 of the patients involved in the study did not have an ‘HF event’ (hospitalization or death).5 Given that the majority of patients had severely symptomatic HF with a median KCCQ score of 44 for women and 53 for men, one can argue health status is a more applicable outcome for the overall population. It has also been demonstrated that nearly half of patients with symptomatic HF are willing to trade their longevity for improved quality of life.6 Ravera et al.7 also demonstrated that women have lower KCCQ scores than men, which has been shown in prior analyses as well. This difference persists after adjustment for comorbidities and demographic factors. This consistent observation across registries and trials suggests that more research is needed to better understand and intervene on these sex-related differences. In another paper published in this Journal, the heterogeneity in approaches to measure health status in HF trials was highlighted by Turgeon et al.,8 who compared the effects of various HF therapies on health status. Given that the treatment of HF has undergone a revolution with new therapeutic options over the past decade, including angiotensin receptor–neprilysin inhibitors, sodium–glucose co-transporter inhibitors 2, omecamtiv mecarbil, vericiguat, intravenous ferric carboxymaltose, and multiple novel devices (e.g. intra-atrial shunts, cardiac contractility stimulation), a meta-analysis comparing health status across interventions will be helpful. However, there are limitations in comparing health status across trials. The older trials used the Minnesota Living with Heart Failure Questionnaire (MLHFQ), while the newer trials used the KCCQ. Within the KCCQ, trials have variably used overall summary score (OSS), clinical summary score (CSS), and total symptom score (TSS) as endpoint. There is evidence suggesting that the KCCQ has significantly better discrimination for predicting HF hospitalizations and death.9 When evaluating short-term changes in health status, it intuitively makes sense to evaluate the physical limitation and symptom domains. The OSS includes social limitations and quality of life domains, which have questions involving hobbies, activities, relationships with loved ones, that are unlikely to change in a short-term clinical trial in less than a year. It is becoming more common to use the CSS in clinical trials that only includes symptoms and physical limitations, but one can imagine tailoring the domain to the proposed mechanism of benefit of the drug. Such that a trial of diuretics or vasodilators would measure TSS, a drug such mavacamten aiming to relieve obstruction in hypertrophic cardiomyopathy would measure physical limitations or CSS, and a drug aimed at long-term weight loss such as semaglutide would use OSS. Moreover, while there are well-established thresholds in terms of what is a meaningful change in KCCQ, with a change of 5 points considered to be a small but clinically important, changes of 10 and 20 points are considered moderate-to-large and large-to-very large clinical improvement. Such categorizations do not exist with the MLWHF. The sparse data trying to equate the KCCQ and the MLHFQ suggest the thresholds of minimally important are larger with the MLHFQ.10 There is a wide variety in duration of follow-up in trials, and optimal timing to measure health status change is unknown, and likely varies based on the intervention. Some trials included in the meta-analysis have short-term (∼3 months) while others had longer (∼24 months) follow-up. Having standard follow-up times to measure health status with different therapies is problematic as its likely some therapies result in more rapid improvement than others, e.g. those targeting left ventricular remodelling may take longer to show health status benefits. One could envision that with beta-blockers, there may be actually short-term worsening in health status but long-term improvement. Additionally, methods of analysing health status data are not consistent. Randomized controlled trials have typically compared mean changes; recently trials have evaluated changes in health status as a function of baseline quality of life. The principle is that most HF trials have significant heterogeneity in baseline health status, and a major determinant of improvement in health status is patient's baseline status. Thus, a plan to evaluate outcomes and treatment effects as a function of baseline health status is rational. Also, the trajectory of health status change in patients with acute, worsening, or chronic HF may differ. Studies such as the Biology Study to Tailored Treatment in Chronic Heart Failure (BIOSTAT-CHF) included both inpatients and outpatients representing significant heterogeneity in health status at baseline as well as short-term treatment alterations.11 Failure to account for this heterogeneity may dilute a health status benefit in the sickest patients by including patients with better health status at baseline with little opportunity to improve. In the future, comparing health status benefit across trials should become easier with KCCQ being routinely collected, and assessing proportion of patients with a clinically important difference is becoming standard.12 There is also a welcome evolution that health status is considered as the primary outcome of trials as opposed to secondary or exploratory endpoints.13 While recognition of the importance of health status is welcome, correctly quantifying the effects of HF therapies on health status is in its infancy. But then, Rome was not built in a day. Conflict of interest: M.N. declares that he serves as a consultant for Roche, Vifor, and Amgen, and has received speaking honoraria from Abbott. J.B. declares that he serves as a consultant for Abbott, Amgen, Array, AstraZeneca, Bayer, Boehringer Ingelheim, Bristol Myers Squibb, CVRx, G3 Pharmaceutical, Impulse Dynamics, Innolife, Janssen, LivaNova, Luitpold, Medtronic, Merck, Novartis, NovoNordisk, Relypsa, Roche, V-Wave Limited, and Vifor.