Diagnostic and prognostic value of the HFA-PEFF score for heart failure with preserved ejection fraction: a systematic review and meta-analysis

Aim To assess the diagnostic and prognostic performances of the Heart Failure Association Pre-test Assessment, Echocardiography & Natriuretic Peptide, Functional Testing, Final Etiology (HFA-PEFF) score for heart failure with preserved ejection fraction (HFpEF) in a comprehensive manner. Methods PubMed, Embase, Cochrane Library, and Web of Science were comprehensively searched from the inception to June 12, 2023. Studies using the “Rule-out” or “Rule-in” approach for diagnosis analysis or studies on cardiovascular events and all-cause death for prognosis analysis were included. The Quality Assessment of Diagnostic Accuracy Studies (QUADAS−2) tool was adopted to assess the quality of diagnostic accuracy studies. The sensitivity (SEN), specificity (SPE), positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnostic odds ratio (DOR), and area under the summary receiver operating characteristic (SROC) curve (AUC) were presented with 95% confidence intervals (CIs). For CVEs and all-cause death, the hazard ratio (HR) values were calculated. Results Fifteen studies involving 6420 subjects were included, with 9 for diagnosis analysis, and 7 for prognosis analysis. For the diagnostic performance of the HFA-PEFF score, with the “Rule-out” approach, the pooled SEN was 0.96 (95%CI: 0.94, 0.97), the pooled SPE was 0.39 (95%CI: 0.37, 0.42), and the pooled AUC was 0.85 (95%CI: 0.67, 1.00), and with the “Rule-in” approach, the pooled SEN was 0.59 (95%CI: 0.56, 0.61), the pooled SPE was 0.86 (95%CI: 0.84, 0.88), and the pooled AUC was 0.83 (95%CI: 0.79, 0.87). For the predictive performance of the HFA-PEFF score, regarding CVEs, the pooled SEN was 0.63 (95%CI: 0.58, 0.67), the pooled SPE was 0.53 (95%CI: 0.49, 0.58), and the pooled AUC was 0.65 (95%CI: 0.40, 0.90), and concerning All-cause death, the pooled SEN was 0.85 (95%CI: 0.81, 0.88), the pooled SPE was 0.48 (95%CI: 0.44, 0.52), and the pooled AUC was 0.65 (95%CI: 0.47, 0.83). A higher HFA-PEFF score was associated with a higher risk of all-cause death (HR 1.390, 95%CI 1.240, 1.558, P < 0.001). Conclusion The HFA-PEFF score might be applied in HFpEF diagnosis and all-cause death prediction. More studies are required for finding validation.


Introduction
The prognosis of heart failure patients is poor, stratified according to ejection fraction classification (1).Heart failure with preserved ejection fraction (HFpEF) is a common clinical syndrome, influencing half of all heart failure patients globally, with rising prevalence and significant morbidity and mortality (2,3).Individuals with HFpEF had a 5-year survival rate of 35%-40% after the first hospitalization (4).Although numerous attempts have been made to find an effective targeted therapy for HFpEF, the currently available evidence is inadequate to support specific drug regimens for patients who present with HFpEF (5)(6)(7)(8), probably because the fundamental pathophysiology of HFpEF is poorly understood, and a firm diagnosis of HFpEF remains a challenge in real-world practice.
In 2019, the Heart Failure Association (HFA) of the European Society of Cardiology (ESC) proposed the Heart Failure Association Pre-test Assessment, Echocardiography & Natriuretic Peptide, Functional Testing, Final Etiology (HFA-PEFF) algorithm to diagnose HFpEF (9), where the HFA-PEFF score incorporates three domains, functional, morphological, and biomarker, to estimate the likelihood (low, intermediate, or high) of suffering from HFpEF (10).Besides, the strategies for improving outcomes in patients with HFpEF are not welldefined.Better definitions of the population at higher clinical risk may be helpful in adjudicating the intensity of follow-up and optimizing therapies (11).Many clinical, biochemical, and echocardiographic derangements have been linked to worse outcomes, and the HFA-PEFF, which considers several easily available variables, has been shown by the existing studies to be helpful in both diagnosis and prognostic prediction of HFpEF (12-17).At present, Li et al. (4) has comprehensively investigated the diagnostic role of the HFA-PEFF score in HFpEF via a meta-analysis, whereas the prognostic value of the HFA-PEFF score for HFpEF is still unclear.
The objective of this systematic review and meta-analysis was to assess the diagnostic and prognostic performances of the HFA-PEFF score for HFpEF in a comprehensive manner, so as to provide a comprehensive understanding of the HFA-PEFF score and promote the clinical risk management of HFA-PEFF.

Search strategy
The following four English databases were comprehensively searched by two independent investigators (XM Li and YY Liang) from the inception to June 12, 2023: PubMed, Embase, Cochrane Library, and Web of Science.The English search term was HFA PEFF.Primary screening was carried out by reading titles and abstracts of the retrieved studies with the help of Endnote X9 (Clarivate, Philadelphia, PA, USA).Subsequently, full texts were read to select eligible studies.Discussion was needed when differences arose regarding search results.This systematic review and meta-analysis was performed according to the Preferred Reporting Items for Systematic Reviews and Metaanalysis (PRISMA).

Eligibility criteria
Inclusion criteria included: (1) studies on individuals with suspected HFpEF (for diagnosis analysis) or diagnosed with HFpEF (for prognosis analysis); (2) studies reporting the HFA-PEFF score; (3) studies providing relevant data to calculate sensitivity (SEN), specificity (SPE), positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnostic odds ratio (DOR), area under the curve (AUC), and prognostic HR values; (4) studies using the "Rule-out" or "Rule-in" approach for diagnosis analysis or studies on cardiovascular events (CVEs, including cardiovascular death, hospitalization for HF decompensation, nonfatal myocardial infarction (MI), unstable angina pectoris, coronary revascularization for a new diagnosis of angina or in-stent restenosis after percutaneous coronary intervention, and nonfatal ischemic stroke) and all-cause death for prognosis analysis; (5) English studies.

Data extraction and quality assessment
Two investigators (XM Li and YY Liang) independently extracted data from the included studies, including the first author, year of publication, study period, study design, sample size, sex (male/female), age (years), body mass index (BMI, kg/m 2 ), comorbidities, medications, HFpEF diagnosis, HFA-PEFF assessment, follow-up time (months), and endpoints.A third author (XZ Lin) would settle relevant disagreements.The Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool was adopted to assess the quality of diagnostic accuracy studies, based on the risk of bias and clinical applicability (18).The risk of bias contained patient selection, index test, reference standard, and flow and timing.Clinical applicability consisted of patient selection, index test, and reference standard.Each item was graded as high (risk), low (risk), or unclear (risk).

Statistical analysis
Statistical analysis was performed using Meta-disc 1.4 (Clinical Biostatistics, Ramony Cajal Hospital, Madrid, Spain), Stata 15.1 (Stata Corporation, College Station, Texas, USA), and Revman 5.4 (The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark).Results were obtained via direct extraction or indirect calculation.Meta-disc 1.4 was applied to evaluate whether there was a threshold effect.When the Spearman correlation coefficient between the logarithm of sensitivity and the logarithm of 1-specificity showed a strong positive correlation, it indicated the existence of a threshold effect.To assess the diagnostic and prognostic value of the HFA-PEFF score, the SEN, SPE, PLR, NLR, and DOR as well as 95% confidence intervals (CIs) for clinical outcomes were reported.Summary receiver operating characteristic (SROC) curves were generated, and the AUC was calculated with 95%CIs.Besides, for CVEs and all-cause death, the hazard ratio (HR) values were calculated using Stata 15.1 with the HFA-PEFF as a continuous variable.Revman 5.4 was used to create a quality assessment chart for the included studies.Differences were significant when P values were less than 0.05.

Study characteristics
A total of 275 studies were retrieved from the four databases.After excluding duplicates, and based on the eligibility criteria, 15 studies (10, 12-17, 19-26) involving 6,420 subjects were included for this systematic review and meta-analysis in the end, with 5 studies from Japan, 2 from China, 1 from Germany, 1 from Italy, 1 from Poland, 1 from the Netherlands, 1 from Korea, 1 from USA, and 2 from multiple countries.The flow chart of study selection is demonstrated in Figure 1.The year of publication ranged from 2020 to 2023.Seven studies had prospective designs, and eight studies had retrospective designs.Nine studies were involved in diagnosis analysis, and seven were included for prognosis analysis.Table 1 exhibits the detailed characteristics of the included studies.With the QUADAS-2 for the quality assessment of the included studies, as regards the risk of bias and clinical applicability, most studies exhibited low risks, followed by unclear risks (Table 2).

Discussion
The current systematic review and meta-analysis comprehensively evaluated the diagnostic and prognostic value of the HFA-PEFF score for HFpEF, and illustrated that with both the "Rule-out" and the "Rule-in" approaches, the HFA-PEFF had a good diagnostic capability for HFpEF based on the pooled AUCs, and the pooled SEN of the Sensitivity (A), specificity (B) and SROC curve (C) of the HFA-PEFF score for hFpEF diagnosis using the "rule-out" approach.SROC, summary receiver operating characteristic; HFA-PEFF, heart failure association pre-test assessment, echocardiography & natriuretic peptide, functional testing, final etiology; HFpEF, heart failure with preserved ejection fraction; AUC, area under the curve; CI, confidence interval.Sensitivity (A), specificity (B) and SROC curve (C) of the HFA-PEFF score for hFpEF diagnosis using the "rule-in" approach.SROC, summary receiver operating characteristic; HFA-PEFF, heart failure association pre-test assessment, echocardiography & natriuretic peptide, functional testing, final etiology; HFpEF, heart failure with preserved ejection fraction; AUC, area under the curve; CI, confidence interval."Rule-out" approach was higher than that of the "Rule-in" approach, while the pooled SPE of the "Rule-in" approach was better than that of the "Rule-out" approach; for all-cause death, the HFA-PEFF exhibited a good predictive SEN, which indicated that the HFA-PEFF might be applied in HFpEF diagnosis and all-cause death prediction.
HFpEF diagnosis is usually performed based on three key components.These include symptoms and signs of heart failure (related to pulmonary and systemic congestion), evidence of "preserved ejection fraction", and the presence of diastolic dysfunction (27).The HFA-PEFF score algorithm was proposed based on a comprehensive diagnostic workup (9,10).In addition, this score also demonstrated its prognostic significance for individuals with HFA-PEFF in several studies (13,23,24).A prior meta-analysis pooled relevant studies to synthetically assess the diagnostic performance of the HFA-PEFF for HFpEF, and it was found that the HFA-PEFF algorithm showed acceptable SPE and SEN for the diagnosis and exclusion of HFpEF (4).This study further conducted an updated comprehensive analysis of HFA-PEFF prognostic value in HFpEF, and the HFA-PEFF was also found to have great performance in HFpEF diagnosis and all-cause death prognostication.
For the diagnosis of HFpEF, the pooled AUC, SEN and SPE of the "Rule-out" approach was 0.85, 0.96 and 0.39, respectively, and the pooled AUC, SEN and SPE of the "Rule-in" approach was 0.83, 0.59 and 0.86, respectively.These suggested that using either the "Rule-out" approach or the "Rule-in" approach, the HFA-PEFF score had a good diagnostic performance; the "Rule-out" approach showed higher SEN, and the "Rule-in" approach had better SPE.To be noted, for patients with intermediate likelihood of HFpEF, exercise induced echocardiography or invasive cardiac hemodynamic measurements would be recommended for the final diagnosis (25).A two-center study has discovered that exercise pulmonary ultrasound exhibits excellent diagnostic value for HFpEF, regardless of the exercise protocol or level of expertise (28).More precise diagnostic modalities are needed, especially in such a serious disease with an incidence close to the incident rate of tuberculosis and other plagues (29)(30)(31)(32), and the clinical indications would guide the diagnostic practice in the context of individualized medical management.Future studies are warranted to verify the diagnostic role of the HFA-PEFF score.
However, in a case-control study evaluating outpatient dyspnea of indeterminate cause and HFpEF, the H2FPEF score exhibited better diagnostic performance compared to the HFA-PEFF score (16).A study based on a Japanese patient cohort has found that the H2FPEF score significantly outperforms the HFA-PEFF score in terms of diagnostic accuracy for HFpEF (25).A research investigation into the diagnostic utility of the H2FPEF scoring system and the HFA-PEFF E-level scoring system within the context of HFpEF has revealed that both scores are efficacious in either excluding or establishing a definitive diagnosis of HFpEF (33).These results suggest the need for further research to compare the importance of different scoring systems in diagnosing HFpEF.
With respect to prognosis in HFpEF, the HFA-PEFF presented a pooled SEN of 0.85 in the prediction of all-cause death, although the pooled AUC was 0.65, while for CVEs, the HFA-PEFF did not have a favorable predictive ability.As reported by Seoudy et al. (34), the HFA-PEFF score is associated with all-cause mortality and heart failure rehospitalization in patients with preserved ejection fraction after transcatheter aortic valve implantation.The HFA-PEFF score has three components that each contribute equally to the overall score (9).One component relies upon natriuretic peptide levels, which have been shown to be strongly related to adverse clinical outcomes in HFpEF as well as HFrEF (35, 36).The second component of the score relies on morphological criteria such as left atrial volume index and left ventricular mass.Left atrial volume index in particular has been demonstrated to be among the most powerful echocardiography predictors of future HF events and reflects the risk of mortality as well (37-39).Finally, the third component of the HFA-PEFF score is represented by functional parameters that reflect left ventricular diastolic dysfunction and/or elevated cardiac filling pressures.One of the parameters incorporated is tricuspid valve regurgitation velocity, with high values indicating pulmonary hypertension, which is strongly associated with mortality in HFpEF (40).For every one point increase in the HFA-PEFF score, the risk of all-cause mortality significantly increased by 39%, which showed a quantitative information to facilitate understanding of the association between the HFA-PEFF and the death risk.Corresponding optimized treatments could be provided to high-risk patients with HFpEF to improve their prognosis.
As demonstrated by this study, the HFA-PEFF score may be taken into consideration by clinicians in the diagnosis and allcause death prognostication of HFpEF, which may help in planning therapeutic methods and improving HFpEF management.Several limitations should be mentioned when interpreting the results.First, there were threshold effects on the diagnostic performance of the HFA-PEFF score for HFpEF using the "Rule-out" and "Rule-in" approaches, which may affect the stability of the results.Second, the availability of a limited number of studies for outcomes such as CVEs may have influenced the reliability of our findings.The small sample size within these studies could have reduced the statistical power to detect significant associations, and the heterogeneity across studies in terms of design, population characteristics, and followup duration further complicates the interpretation of the results.Third, an important limitation to consider in the validation of the HFA-PEFF score is the heterogeneity of criteria used by each study to define HFpEF patients.The included studies span a period from 2000 to 2021, during which time diagnostic and classification criteria for HFpEF have evolved.This variability in the definition of HFpEF could introduce bias and affect the comparability of results across studies.Our findings highlight the urgent need for additional studies to evaluate the prognostic value of the HFA-PEFF score in HFpEF, with the aim of providing a more robust evidence base to support its use in clinical practice.

Conclusion
The HFA-PEFF had a good diagnostic capability for HFpEF using both the "Rule-out" and the "Rule-in" approaches based on the pooled AUCs, and it exhibited a good predictive SEN for all-cause death in patients with HFpEF, suggesting that the HFA-PEFF may be considered in HFpEF diagnosis and all-cause death prediction.More studies are needed for finding validation.

FIGURE 1 Flow
FIGURE 1Flow diagram of study screening.

FIGURE 5
FIGURE 5Sensitivity (A), specificity (B) and SROC curve (C) of the HFA-PEFF score for all-cause death prediction.SROC, summary receiver operating characteristic; HFA-PEFF, heart failure association pre-test assessment, echocardiography & natriuretic peptide, functional testing, final etiology; HFpEF, heart failure with preserved ejection fraction; AUC, area under the curve; CI, confidence interval.

TABLE 1
Characteristics of the included studies.

TABLE 2
Quality assessment of the included studies by the QUADAS-2.
QUADAS-2, quality assessment of diagnostic accuracy studies; L, low risk; H, high risk; U, unclear risk.

TABLE 3
Diagnostic and prognostic performances of the HFA-PEFF score for hFpEF.