Diagnosing Heart Failure with Preserved Ejection Fraction – what’s the Score?

This article refers to ‘Validation of the HFA-PEFF score for the diagnosis of heart failure with preserved ejection fraction’ by A. Barandiarán Aizpurua et al., published in this issue on pages 413–421. Definitive diagnosis of heart failure with preserved ejection fraction (HFpEF) remains a challenge in the absence of invasive haemodynamic measurements.1, 2 In part, this is because indirect measures of raised intracardiac filling pressures are imprecise and, in the case of natriuretic peptides may be influenced by comorbidities such as obesity (associated with lower concentrations) and chronic kidney disease and atrial fibrillation (associated with elevated concentrations). Moreover, left ventricular filling pressure may be normal at rest in patients with HFpEF, especially in individuals taking a diuretic, and only rise to abnormally high levels during exercise.1, 2 Recently, two scores have been developed to help diagnose HFpEF. Reddy et al.3 derived and validated a predictive model in two cohorts of patients referred for investigation of possible HFpEF, all of which underwent invasive haemodynamic exercise testing. Six variables were independent predictors of HFpEF: age > 60 years, body mass index >30 kg/m2, hypertension treated with ≥2 antihypertensive medications, atrial fibrillation, Doppler echocardiographic E/e′ > 9, and pulmonary artery systolic pressure > 35 mmHg (Table 1). The resultant total H2FPEF score ranges from 0 to 9 with the scores <2 and scores ≥6, respectively, reflecting a low or high likelihood of HFpEF. Individuals with an intermediate score (between 2 and 5) require further investigation, e.g. invasive haemodynamic assessment. Of note, natriuretic peptides were not found to be predictive and are not included in this score. Unfortunately, as is often the case with scores of this type, an external validation raised doubts about its usefulness in less selected cohorts.4 Age >60 years (1) Body mass index >30 kg/m2 (2) Hypertension treated with ≥2 antihypertensive medications (1) AF (3) LAVI >34 mL/m2 (2) or LVMI >149/122 g/m2 (m/w) and RWT >0.42 (2) LAVI 29–34 mL/m2 (1) or LVMI >115/95 g/m2 (m/w) (1) or RWT >0.42 (1) or LV wall thickness ≥ 12 mm Cardiac function/ non-invasive haemodynamics Average E/e′ > 9 (1) TR velocity > 2.8 m/s (PASP >35 mmHg) (1) Average E/e′ ≥ 15 (2) or septal e′ < 7 cm/s (2) or lateral e′ < 10 cm/s (2) or TR velocity > 2.8 m/s (PASP >35 mmHg) (2) Average E/e′ 9–14 (1) or GLS <16% (1) NT-proBNP >220 pg/mL (2) or BNP >80 pg/mL (2) [AF: NT-proBNP >660 pg/mL (2) or BNP >240 pg/mL (2)] NT-proBNP 125–220 pg/mL (1) or BNP 35–80 pg/mL (1) [AF: NT-proBNP 365–660 pg/mL (1) or BNP 105–240 pg/mL (1)] <2 low (HFpEF unlikely) 2–5 intermediate (further investigation) ≥6 high (confirmed) <2 low (HFpEF unlikely) 2–4 intermediate (further investigation) ≥5 high (confirmed) Recently, the Heart Failure Association of the European Society of Cardiology published a 4-step algorithm for the diagnosis HFpEF, based upon expert consensus.5 The suggested approach also uses a score, the HFA-PEFF score. Step 1 is assessment of the pre-test likelihood of HFpEF. Step 2 involves scoring based on major (2 point) and minor (1 point) criteria in three categories: function, morphology and biomarker. The resultant score includes some of the same variables as in the H2FPEF score, such as average E/e' and tricuspid regurgitation velocity, but is considerably more complex (Table 1). The pre-test probability step in the HFA-PEFF score includes consideration of some of the demographic/comorbidity variables included in the H2FPEF score but does not allocate points for these. In the HFA-PEFF score, echocardiographic measurements are used to provide the supporting functional and morphological evidence and a natriuretic peptide (B-type or N-terminal pro-B-type natriuretic peptide) measurement the biomarker evidence in support of the diagnosis. Combined, the major and minor criteria yield a HFA-PEFF score of between 0 and 6 points (the maximum score allowed in each category is 2 points). An overall score of <2 points means HFpEF is unlikely. A score of 2–4 points indicates that a diagnosis of HFpEF is possible, but further work-up is needed (Step 3), and a score of ≥5 points that HFpEF is confirmed. The fourth and final step is an optional one where the goal is to define the underlying aetiology. In this issue of the Journal, Barandiarán Aizpurua and colleagues have examined the performance of the second step in the HFA-PEFF algorithm in an important validation study in two single-centre patient cohorts.6 The Maastricht cohort included 270 patients referred with suspected HFpEF, the diagnosis of which was confirmed in 228. The Chicago cohort comprised only patients with confirmed diagnosis of HFpEF, all of which had been hospitalized for heart failure previously. It is important to note that the final clinical diagnosis against which the score was compared included haemodynamic measurements during rest and exercise in only around half of the Chicago cohort and none of the Maastricht cohort, yet this is arguably the gold standard against which any new test should be compared. As a result, the present study compared the HFA-PEFF score against a diagnosis of HFpEF made by experts mainly using the same variables as those included in the score. Moreover, the ratio of HFpEF ‘cases’ to non-HFpEF ‘controls’ was very high. These two factors may have led to overestimation of the performance of the HFA-PEFF score. These caveats need to be taken into account when considering the apparently excellent results of the ‘rule-in approach’ with a specificity of 93%. Only 13 (3%) individuals from the Chicago cohort and 3 (1%) from the Maastricht cohort had a diagnosis of HFpEF but a HFA-PEFF score of <2 point. Similarly, the performance of the score used as a ‘rule-out’ test, where a score of <2 had a sensitivity of 99% in the Maastricht cohort, may be overoptimistic, especially with the small number of non-HFpEF controls. Barandiarán Aizpurua and colleagues also suggest simplifications of the algorithm and, interestingly, found that natriuretic peptides had a similar performance to the total score. Although potentially very valuable, we cannot draw definitive conclusions about these observations in the present cohorts, because of the limitations discussed above. However, should natriuretic peptides alone be confirmed as useful as the HFA-PEFF score, clinical practice could be simplified greatly. The present report by Barandiarán Aizpurua and colleagues is a very welcome first step in evaluating the HFA-PEFF score and hopefully more investigators will evaluate this and the H2FPEF score in further cohorts of less selected patients, including a larger proportion of patients without a final diagnosis of HFpEF. An evaluation of both scores simultaneously in the same cohort will be especially important given the simplicity of the H2FPEF score, should it perform as well as the HFA-PEFF score. Although difficult, it will be particularly important to test both scores against the gold standard of invasive measurements of cardiac filling pressures, including pressures during exercise. Which if any of these two scores will prove useful in the longer term remains to be seen and further iterations of this type of approach may emerge as a result of future investigation. Ultimately, any validated aid to the diagnosis of HFpEF will be a big step forward. Conflict of interest: none declared.