Abstract
Obesity is related to better prognosis in heart failure with either reduced (HFrEF; left ventricular ejection fraction (LVEF) < 40%) or preserved LVEF (HFpEF; LVEF ≥50%). Whether the obesity paradox exists in patients with heart failure and mid-range LVEF (HFmrEF; LVEF 40–49%) and whether it is independent of heart failure aetiology is unknown. Therefore, we aimed to test the prognostic value of body mass index (BMI) in ischaemic and non-ischaemic heart failure patients across the whole spectrum of LVEF.
Consecutive ambulatory heart failure patients were enrolled in two tertiary centres in Italy and Spain and classified as HFrEF, HFmrEF or HFpEF, of either ischaemic or non-ischaemic aetiology. Patients were stratified into underweight (BMI < 18.5 kg/m2), normal-weight (BMI 18.5–24.9 kg/m2), overweight (BMI 25–29.9 kg/m2), mild-obese (BMI 30–34.9 kg/m2), moderate-obese (BMI 35–39.9 kg/m2) and severe-obese (BMI ≥40 kg/m2) and followed up for the end-point of five-year all-cause mortality.
We enrolled 5155 patients (age 70 years (60–77); 71% males; LVEF 35% (27–45); 63% HFrEF, 18% HFmrEF, 19% HFpEF). At multivariable analysis, mild obesity was independently associated with a lower risk of all-cause mortality in HFrEF (hazard ratio, 0.78 (95% confidence interval (CI) 0.64–0.95), p = 0.020), HFmrEF (hazard ratio 0.63 (95% CI 0.41–0.96), p = 0.029), and HFpEF (hazard ratio 0.60 (95% CI 0.42–0.88), p = 0.008). Both overweight and mild-to-moderate obesity were associated with better outcome in non-ischaemic heart failure, but not in ischaemic heart failure.
Mild obesity is independently associated with better survival in heart failure across the whole spectrum of LVEF. Prognostic benefit of obesity is maintained only in non-ischaemic heart failure.
Introduction
In the last decades, the prevalence of obesity has been growing, formerly limited to western countries and later spreading throughout the developing world.1 As obesity is a well-defined risk factor for several cardiovascular and non-cardiovascular diseases, such an epidemic has been related to increased morbidity and mortality in the general population.2 Nevertheless, when assessing the impact of obesity on outcome of patients with different chronic disorders (including cardiovascular, respiratory and neoplastic diseases) an unexpected protective effect often emerges, which has been termed the ‘obesity paradox’.3,4
The relation between body mass index (BMI) and chronic heart failure has been largely tested. While obesity is a strong risk factor for heart failure, even among categories otherwise at lower risk, such as younger women, overweight and mild-to-moderately obese patients survive longer than normal weight and underweight patients.5,6 A similar relation has been also reported recently for per cent body fat (PBF).7 Although many hypotheses have been proposed in the last years,8–12 the mechanisms leading to this prognostic benefit remain to be clarified. Most of the studies have enrolled only patients with heart failure and reduced left ventricular ejection fraction (LVEF) < 40% (HFrEF). Nevertheless, obesity paradox has been recently reported also in patients with heart failure with preserved ejection fraction (LVEF ≥50%; HFpEF), even though overweight and central obesity are strongly associated with higher incidence of heart failure in such a context.13,14 Whether obesity paradox is present also in patients with heart failure and mid-range ejection fraction (LVEF 40–49%; HFmrEF) has never been addressed in a dedicated analysis so far. Furthermore, the interaction between the prognostic value of BMI and heart failure aetiology remains unclear. Indeed, while a previous single study on patients with HFrEF reported no difference between ischaemic and non-ischaemic aetiology,15 obesity paradox was limited to patients with non-ischaemic heart failure in a preliminary study from the Badalona cohort.16 Finally, there is an inverse relation between BMI and circulating levels of both B-type natriuretic peptide (BNP) and the N-terminal fraction of pro-BNP (NT-proBNP) in the general population and in heart failure patients.17,18
Our aim was to test the prognostic value of BMI in ischaemic and non-ischaemic heart failure patients across the whole spectrum of LVEF.
Methods
Consecutive ambulatory patients with chronic heart failure, receiving guideline-driven therapy, were prospectively enrolled from May 1998 to August 2018 in two tertiary centres in Italy (Fondazione Toscana Gabriele Monasterio, Pisa) and Spain (Hospital Universitari Germans Trias i Pujol, Badalona). Diagnosis of chronic heart failure was made according to the contemporary European guidelines. 19 The clinical practice referrals to the heart failure clinics and the scheduled protocol management have been reported elsewhere.20–22 At time of recruitment, patients underwent resting 12-lead electrocardiogram and bio-humoral characterization, including NT-proBNP measurement.23,24 Standard 2D echocardiography was performed by expert cardiologists according to contemporary guidelines for the evaluation of wall thickness, chamber volumes, wall motion and indices of systolic and diastolic function. LVEF was calculated by the Simpson’s rule.25
Patients were classified according to LVEF as having HFrEF (LVEF < 40%), HFmrEF (LVEF 40–49%), HFpEF (LVEF ≥50%). Moreover, heart failure aetiology was considered as either ischaemic or non-ischaemic. Ischaemic aetiology was defined as the presence of at least one of the following criteria: 1) angiographic evidence of ≥75% lesion in at least one major epicardial vessel, associated with imaging findings consistent with myocardial necrosis; 2) history of myocardial infarction or revascularization. BMI was calculated as the ratio between weight in kilograms (kg) and the square of height in metres (m2). According to the World Health Organization (WHO) indication,26 patients were then stratified into: underweight (BMI < 18.5 kg/m2), normal weight (BMI 18.5–24.9 kg/m2), overweight (BMI 25–29.9 kg/m2), mild obesity (BMI 30–34.9 kg/m2), moderate obesity (BMI 35–39.9 kg/m2) and severe obesity (BMI ≥40 kg/m2). PBF was calculated by Jackson–Pollock and Gallagher equations.7,27,28 Patients were then stratified into the respective PBF tertiles: PBF < 25.1%, 25.1–33.1% and ≥33.1%, in first, second and third tertile, respectively, according to the Jackson–Pollock equation; PBF < 26.3%, 26.3–33.7% and ≥33.7%, in first, second and third tertile, respectively, according to the Gallagher equation. Body surface area (BSA) was calculated with the Mosteller formula (BSA < 1.77 m2, 1.77–1.96 m2 and >1.96 m2, in first, second and third tertile, respectively).29
All patients provided informed consent for the study, which was approved by the Institutional Ethics Committee and conducted in accordance with the Declaration of Helsinki of the World Medical Association.
Survival analysis and endpoints
After baseline assessment, follow-up was performed at our out-patient clinic every 3–6 months, as clinically indicated. Independent interviewers obtained data from the patients, relatives or general practitioners. Information about the time and cause of death were retrieved from death certificates, post-mortem reports and family doctors. In Spain, data were verified using the databases of the Catalan and Spanish Health Systems. The primary endpoint was five-year all-cause mortality, while the secondary endpoint was five-year cardiac death (including sudden cardiac death and death due to heart failure progression or acute myocardial infarction). Patients were censored after the last medical contact.
Statistical analysis
The distribution of each variable was evaluated by Kolmogorov–Smirnov test. Normally distributed variables were reported as mean ± standard deviation, while not-normally distributed variables as median and interquartile range (IQR). Categorical data were reported as frequencies. Quantitative variables between two groups were compared using independent sample t-test or Mann–Whitney U-test, according to distribution. For multiple comparisons among groups, ANOVA or Kruskal–Wallis test were employed, according to variable distribution, with post-hoc Bonferroni correction for pairwise comparisons. For qualitative variables, χ2 or Fisher test were used.
For survival analysis, either normal weight (for BMI) or the first PBF tertile were considered as reference categories. Kaplan–Meier method and log-rank statistics (Mantel–Cox) were used to estimate survival according to BMI class in the whole population and then in subgroups of patients stratified for LVEF. Cox regression analysis was used to evaluate the independent prognostic contribution of each BMI and PBF class in the whole cohort and in HFrEF, HFmrEF and HFpEF. A multivariable model was built to account for possible confounders, selected a priori and including age, gender, therapy with β-blockers, angiotensin-converting enzyme inhibitors/angiotensin receptor-I blockers, mineralocorticoid-receptor antagonists, furosemide, New York Heart Association class and estimated glomerular filtration rate (eGFR, calculated with the CKD-EPI formula).
The prognostic value of BMI and PBF was also tested separately in patients with either ischaemic or non-ischaemic heart failure, adjusting for the aforementioned variables, and for LVEF class.
Finally, competing-risks regression analysis was used to test the prognostic value of either BMI categories or PBF for the secondary endpoint of five-year cardiac death, while non-cardiac death was considered as competing outcome.
Statistical analysis was performed using SPSS (IBM Statistics, version 25.0, 2017) or Stata Statistical Software (StataCorp, release 15, 2017) and a two-tailed p value ≤0.05 was considered significant.
Results
General features of the study population
We recruited 5155 patients (2763 in Pisa and 2392 in Badalona; 71% males, n = 3641; median age 70 years, IQR 60–77 years). Clinical, bio-humoral and echocardiographic characteristics of the study population are summarized in Table 1. HFrEF was diagnosed in 63% of patients (n = 3226), HFmrEF and HFpEF in 18% (n = 947) and 19% (n = 982), respectively. About half of the patients had heart failure of ischaemic aetiology. Median BMI was 26.8 kg/m2 (IQR 23.9–30.1 kg/m2) in the whole cohort.
Clinical, bio-humoral and echocardiographic data of the whole population and of subgroups according to body mass index categories.
| All patients N = 5155 | Normal weight BMI 18.5–24.9 kg/m2n = 1683 | Underweight BMI < 18.5 kg/m2n = 85 | Overweight BMI 25–29.9 kg/m2n = 2058 | Mild obesity BMI 30–34.9 kg/m2n = 937 | Moderate obesity BMI 35–39.9 kg/m2n = 270 | Severe obesity BMI >40 kg/m2n = 122 | |
|---|---|---|---|---|---|---|---|
| Age, years | 70 (60–77) | 71 (61–78) | 73 (62–81) | 70 (60–77) | 68 (58–75)** | 66 (57–75) ** | 64 (53–73)** |
| Gender, males, n (%) | 3641 (71) | 1166 (69) | 44 (52)** | 1566 (76)** | 658 (70) | 147 (54)** | 63 (52)** |
| BMI, kg/m2 | 26.8 (23.9–30.1) | 23 (21.6–24.1) | 17.5 (16.6–18.0)** | 27.3 (26.1–28.4)** | 31.8 (30.8–33.1)** | 36.9 (35.8–38.4)** | 42.6 (41.1–45.3)** |
| LVEF, % | 35 (27–45) | 33 (25–42) | 34 (25–47) | 35 (27–45)** | 36 (29–48)** | 40 (30–53)** | 40 (30–55)** |
| HFrEF, % | 3226 (63) | 1153 (68) | 57 (67) | 1291 (63)** | 545 (58)** | 121 (45)** | 59 (48)** |
| HFmrEF, % | 947 (18) | 277 (17) | 8 (9) | 412 (20)** | 169 (18)** | 62 (23)** | 19 (16)** |
| HFpEF, % | 982 (19) | 253 (15) | 20 (24) | 355 (17)** | 223 (24)** | 87 (32)** | 44 (36)** |
| Ischaemic aetiology, n (%) | 2543 (49) | 836 (50) | 30 (35)** | 1070 (52)** | 458 (49) | 111 (41)** | 38 (31)** |
| NYHA Class III-IV, n (%) | 1610 (31) | 521 (31) | 38 (45) | 571 (28) | 267 (29) | 112 (42)** | 59 (48)** |
| Atrial fibrillation, n (%) | 1138 (22) | 312 (19) | 22 (26) | 463 (23) | 237 (25)** | 69 (26) | 35 (29) |
| Hypertension, n (%) | 3133 (61) | 879 (52) | 34 (40)** | 1242 (60)** | 671 (72)** | 208 (77)** | 99 (81)** |
| Diabetes mellitus, n (%) | 1834 (36) | 473 (28) | 14 (17)** | 737 (36)** | 403 (43)** | 135 (50)** | 72 (59)** |
| Dyslipidaemia, n (%) | 2507 (49) | 730 (43) | 20 (24)** | 1033 (50)** | 510 (54)** | 142 (53)** | 72 (59)** |
| LA volume, ml/m2 | 46 (42–51) | 45 (40–50) | 43 (37–50) | 47 (42–51)** | 47 (43–52)** | 48 (43–52)** | 49 (44–53)** |
| EDLVD, mm | 59 (52–65) | 58 (52–64) | 44 (36–51) | 59 (53–65) | 59 (52–65) | 59 (52–64) | 57 (51–66) |
| ESLVD, mm | 47 (39–54) | 47 (40–54) | 34 (25–47) | 47 (39–55) | 45 (37–53) | 45 (37–53) | 44 (37–53) |
| TAPSE, mm | 18 (15–22) | 18 (14–21) | 19 (14–21) | 18 (15–22) | 19 (16–23)** | 20 (17–25)** | 20 (15–25) |
| sPAP, mmHg | 44 (36–55) | 45 (37–55) | 44 (39–58) | 44 (36–55) | 42 (36–53) | 45 (37–54) | 48 (36–56) |
| Hb, g/dl | 13.1 (11.8–14.3) | 12.9 (11.6–14.1) | 12.4 (11.3–13.8) | 13.3 (12–14.4)** | 13.4 (12.2–14.6)** | 13.2 (11.7–14.5) | 13.1 (11.6–14.4) |
| Creatinine, mg/dl | 1.1 (0.9–1.4) | 1.1 (0.9–1.4) | 0.9 (0.7–1.3) | 1.1 (0.9–1.4) | 1.0 (0.8–1.4) | 1.0 (0.8–1.3) | 1.0 (0.8–1.4) |
| eGFR, ml/min per 1.73 m2 | 67 (45–86) | 65 (44–85) | 70 (43–91) | 66 (45–86) | 68 (48–87) | 68 (48–87) | 68 (47–87) |
| NT-proBNP, ng/l | 1422 (538–3635) | 2095 (788–5071) | 2864 (1176–7818) | 1355 (534–3348)** | 1022 (390–2135)** | 761 (304–1744)** | 916 (369–2140)** |
| β-blockers, n (%) | 4070 (79) | 1355 (81) | 159 (69) | 1620 (79) | 719 (77) | 214 (79) | 103 (84) |
| ACEi/ARBs, n (%) | 4078 (79) | 1309 (78) | 69 (81) | 1606 (78) | 772 (83) | 205 (76) | 94 (85) |
| MRAs, n (%) | 2448 (48) | 782 (47) | 39 (46) | 968 (47) | 451 (48) | 140 (52) | 68 (56) |
| Furosemide, n (%) | 3924 (76) | 1255 (75) | 66 (77) | 1591 (77) | 709 (76) | 208 (77) | 95 (78) |
| Digoxin, n (%) | 1067 (21) | 342 (20) | 13 (15) | 440 (21) | 184 (20) | 62 (23) | 26 (21) |
| OAC, n (%) | 1763 (34) | 564 (34) | 28 (33) | 712 (35) | 312 (33) | 100 (37) | 47 (39) |
| CRT/ICD, n (%) | 1150 (23) | 392 (24) | 7 (8) | 521 (26) | 181 (19) | 31 (11) | 18 (14) |
| All patients N = 5155 | Normal weight BMI 18.5–24.9 kg/m2n = 1683 | Underweight BMI < 18.5 kg/m2n = 85 | Overweight BMI 25–29.9 kg/m2n = 2058 | Mild obesity BMI 30–34.9 kg/m2n = 937 | Moderate obesity BMI 35–39.9 kg/m2n = 270 | Severe obesity BMI >40 kg/m2n = 122 | |
|---|---|---|---|---|---|---|---|
| Age, years | 70 (60–77) | 71 (61–78) | 73 (62–81) | 70 (60–77) | 68 (58–75)** | 66 (57–75) ** | 64 (53–73)** |
| Gender, males, n (%) | 3641 (71) | 1166 (69) | 44 (52)** | 1566 (76)** | 658 (70) | 147 (54)** | 63 (52)** |
| BMI, kg/m2 | 26.8 (23.9–30.1) | 23 (21.6–24.1) | 17.5 (16.6–18.0)** | 27.3 (26.1–28.4)** | 31.8 (30.8–33.1)** | 36.9 (35.8–38.4)** | 42.6 (41.1–45.3)** |
| LVEF, % | 35 (27–45) | 33 (25–42) | 34 (25–47) | 35 (27–45)** | 36 (29–48)** | 40 (30–53)** | 40 (30–55)** |
| HFrEF, % | 3226 (63) | 1153 (68) | 57 (67) | 1291 (63)** | 545 (58)** | 121 (45)** | 59 (48)** |
| HFmrEF, % | 947 (18) | 277 (17) | 8 (9) | 412 (20)** | 169 (18)** | 62 (23)** | 19 (16)** |
| HFpEF, % | 982 (19) | 253 (15) | 20 (24) | 355 (17)** | 223 (24)** | 87 (32)** | 44 (36)** |
| Ischaemic aetiology, n (%) | 2543 (49) | 836 (50) | 30 (35)** | 1070 (52)** | 458 (49) | 111 (41)** | 38 (31)** |
| NYHA Class III-IV, n (%) | 1610 (31) | 521 (31) | 38 (45) | 571 (28) | 267 (29) | 112 (42)** | 59 (48)** |
| Atrial fibrillation, n (%) | 1138 (22) | 312 (19) | 22 (26) | 463 (23) | 237 (25)** | 69 (26) | 35 (29) |
| Hypertension, n (%) | 3133 (61) | 879 (52) | 34 (40)** | 1242 (60)** | 671 (72)** | 208 (77)** | 99 (81)** |
| Diabetes mellitus, n (%) | 1834 (36) | 473 (28) | 14 (17)** | 737 (36)** | 403 (43)** | 135 (50)** | 72 (59)** |
| Dyslipidaemia, n (%) | 2507 (49) | 730 (43) | 20 (24)** | 1033 (50)** | 510 (54)** | 142 (53)** | 72 (59)** |
| LA volume, ml/m2 | 46 (42–51) | 45 (40–50) | 43 (37–50) | 47 (42–51)** | 47 (43–52)** | 48 (43–52)** | 49 (44–53)** |
| EDLVD, mm | 59 (52–65) | 58 (52–64) | 44 (36–51) | 59 (53–65) | 59 (52–65) | 59 (52–64) | 57 (51–66) |
| ESLVD, mm | 47 (39–54) | 47 (40–54) | 34 (25–47) | 47 (39–55) | 45 (37–53) | 45 (37–53) | 44 (37–53) |
| TAPSE, mm | 18 (15–22) | 18 (14–21) | 19 (14–21) | 18 (15–22) | 19 (16–23)** | 20 (17–25)** | 20 (15–25) |
| sPAP, mmHg | 44 (36–55) | 45 (37–55) | 44 (39–58) | 44 (36–55) | 42 (36–53) | 45 (37–54) | 48 (36–56) |
| Hb, g/dl | 13.1 (11.8–14.3) | 12.9 (11.6–14.1) | 12.4 (11.3–13.8) | 13.3 (12–14.4)** | 13.4 (12.2–14.6)** | 13.2 (11.7–14.5) | 13.1 (11.6–14.4) |
| Creatinine, mg/dl | 1.1 (0.9–1.4) | 1.1 (0.9–1.4) | 0.9 (0.7–1.3) | 1.1 (0.9–1.4) | 1.0 (0.8–1.4) | 1.0 (0.8–1.3) | 1.0 (0.8–1.4) |
| eGFR, ml/min per 1.73 m2 | 67 (45–86) | 65 (44–85) | 70 (43–91) | 66 (45–86) | 68 (48–87) | 68 (48–87) | 68 (47–87) |
| NT-proBNP, ng/l | 1422 (538–3635) | 2095 (788–5071) | 2864 (1176–7818) | 1355 (534–3348)** | 1022 (390–2135)** | 761 (304–1744)** | 916 (369–2140)** |
| β-blockers, n (%) | 4070 (79) | 1355 (81) | 159 (69) | 1620 (79) | 719 (77) | 214 (79) | 103 (84) |
| ACEi/ARBs, n (%) | 4078 (79) | 1309 (78) | 69 (81) | 1606 (78) | 772 (83) | 205 (76) | 94 (85) |
| MRAs, n (%) | 2448 (48) | 782 (47) | 39 (46) | 968 (47) | 451 (48) | 140 (52) | 68 (56) |
| Furosemide, n (%) | 3924 (76) | 1255 (75) | 66 (77) | 1591 (77) | 709 (76) | 208 (77) | 95 (78) |
| Digoxin, n (%) | 1067 (21) | 342 (20) | 13 (15) | 440 (21) | 184 (20) | 62 (23) | 26 (21) |
| OAC, n (%) | 1763 (34) | 564 (34) | 28 (33) | 712 (35) | 312 (33) | 100 (37) | 47 (39) |
| CRT/ICD, n (%) | 1150 (23) | 392 (24) | 7 (8) | 521 (26) | 181 (19) | 31 (11) | 18 (14) |
Values are n (%) or median (interquartile range).
*p = 0.05 versus normal weight.
**p = 0.01 versus normal weight.
ACEi: angiotensin converter enzyme inhibitor; ARB: angiotensin II receptor blocker; BMI: body mass index; CRT: cardiac resynchronization therapy; EDLVD: end-diastolic left ventricular diameter; ESLVD: end-systolic left ventricular diameter; eGFR: estimated glomerular filtration rate (through CKD-EPI formula); Hb: haemoglobin; HFmrEF: heart failure with mid-range left ventricular ejection fraction; HFpEF: heart failure with preserved left ventricular ejection fraction; HFrEF: heart failure with reduced left ventricular ejection fraction; ICD: implantable cardioverter-defibrillator; LA: left atrium; LVEF: left ventricular ejection fraction; MRA: mineralocorticoid receptor antagonist; NT-proBNP: N-terminal pro-B-type natriuretic peptide; NYHA: New York Heart Association; OAC: oral anticoagulants; sPAP: systolic pulmonary arterial pressure; TAPSE: tricuspid annular plane systolic excursion.
Clinical, bio-humoral and echocardiographic data of the whole population and of subgroups according to body mass index categories.
| All patients N = 5155 | Normal weight BMI 18.5–24.9 kg/m2n = 1683 | Underweight BMI < 18.5 kg/m2n = 85 | Overweight BMI 25–29.9 kg/m2n = 2058 | Mild obesity BMI 30–34.9 kg/m2n = 937 | Moderate obesity BMI 35–39.9 kg/m2n = 270 | Severe obesity BMI >40 kg/m2n = 122 | |
|---|---|---|---|---|---|---|---|
| Age, years | 70 (60–77) | 71 (61–78) | 73 (62–81) | 70 (60–77) | 68 (58–75)** | 66 (57–75) ** | 64 (53–73)** |
| Gender, males, n (%) | 3641 (71) | 1166 (69) | 44 (52)** | 1566 (76)** | 658 (70) | 147 (54)** | 63 (52)** |
| BMI, kg/m2 | 26.8 (23.9–30.1) | 23 (21.6–24.1) | 17.5 (16.6–18.0)** | 27.3 (26.1–28.4)** | 31.8 (30.8–33.1)** | 36.9 (35.8–38.4)** | 42.6 (41.1–45.3)** |
| LVEF, % | 35 (27–45) | 33 (25–42) | 34 (25–47) | 35 (27–45)** | 36 (29–48)** | 40 (30–53)** | 40 (30–55)** |
| HFrEF, % | 3226 (63) | 1153 (68) | 57 (67) | 1291 (63)** | 545 (58)** | 121 (45)** | 59 (48)** |
| HFmrEF, % | 947 (18) | 277 (17) | 8 (9) | 412 (20)** | 169 (18)** | 62 (23)** | 19 (16)** |
| HFpEF, % | 982 (19) | 253 (15) | 20 (24) | 355 (17)** | 223 (24)** | 87 (32)** | 44 (36)** |
| Ischaemic aetiology, n (%) | 2543 (49) | 836 (50) | 30 (35)** | 1070 (52)** | 458 (49) | 111 (41)** | 38 (31)** |
| NYHA Class III-IV, n (%) | 1610 (31) | 521 (31) | 38 (45) | 571 (28) | 267 (29) | 112 (42)** | 59 (48)** |
| Atrial fibrillation, n (%) | 1138 (22) | 312 (19) | 22 (26) | 463 (23) | 237 (25)** | 69 (26) | 35 (29) |
| Hypertension, n (%) | 3133 (61) | 879 (52) | 34 (40)** | 1242 (60)** | 671 (72)** | 208 (77)** | 99 (81)** |
| Diabetes mellitus, n (%) | 1834 (36) | 473 (28) | 14 (17)** | 737 (36)** | 403 (43)** | 135 (50)** | 72 (59)** |
| Dyslipidaemia, n (%) | 2507 (49) | 730 (43) | 20 (24)** | 1033 (50)** | 510 (54)** | 142 (53)** | 72 (59)** |
| LA volume, ml/m2 | 46 (42–51) | 45 (40–50) | 43 (37–50) | 47 (42–51)** | 47 (43–52)** | 48 (43–52)** | 49 (44–53)** |
| EDLVD, mm | 59 (52–65) | 58 (52–64) | 44 (36–51) | 59 (53–65) | 59 (52–65) | 59 (52–64) | 57 (51–66) |
| ESLVD, mm | 47 (39–54) | 47 (40–54) | 34 (25–47) | 47 (39–55) | 45 (37–53) | 45 (37–53) | 44 (37–53) |
| TAPSE, mm | 18 (15–22) | 18 (14–21) | 19 (14–21) | 18 (15–22) | 19 (16–23)** | 20 (17–25)** | 20 (15–25) |
| sPAP, mmHg | 44 (36–55) | 45 (37–55) | 44 (39–58) | 44 (36–55) | 42 (36–53) | 45 (37–54) | 48 (36–56) |
| Hb, g/dl | 13.1 (11.8–14.3) | 12.9 (11.6–14.1) | 12.4 (11.3–13.8) | 13.3 (12–14.4)** | 13.4 (12.2–14.6)** | 13.2 (11.7–14.5) | 13.1 (11.6–14.4) |
| Creatinine, mg/dl | 1.1 (0.9–1.4) | 1.1 (0.9–1.4) | 0.9 (0.7–1.3) | 1.1 (0.9–1.4) | 1.0 (0.8–1.4) | 1.0 (0.8–1.3) | 1.0 (0.8–1.4) |
| eGFR, ml/min per 1.73 m2 | 67 (45–86) | 65 (44–85) | 70 (43–91) | 66 (45–86) | 68 (48–87) | 68 (48–87) | 68 (47–87) |
| NT-proBNP, ng/l | 1422 (538–3635) | 2095 (788–5071) | 2864 (1176–7818) | 1355 (534–3348)** | 1022 (390–2135)** | 761 (304–1744)** | 916 (369–2140)** |
| β-blockers, n (%) | 4070 (79) | 1355 (81) | 159 (69) | 1620 (79) | 719 (77) | 214 (79) | 103 (84) |
| ACEi/ARBs, n (%) | 4078 (79) | 1309 (78) | 69 (81) | 1606 (78) | 772 (83) | 205 (76) | 94 (85) |
| MRAs, n (%) | 2448 (48) | 782 (47) | 39 (46) | 968 (47) | 451 (48) | 140 (52) | 68 (56) |
| Furosemide, n (%) | 3924 (76) | 1255 (75) | 66 (77) | 1591 (77) | 709 (76) | 208 (77) | 95 (78) |
| Digoxin, n (%) | 1067 (21) | 342 (20) | 13 (15) | 440 (21) | 184 (20) | 62 (23) | 26 (21) |
| OAC, n (%) | 1763 (34) | 564 (34) | 28 (33) | 712 (35) | 312 (33) | 100 (37) | 47 (39) |
| CRT/ICD, n (%) | 1150 (23) | 392 (24) | 7 (8) | 521 (26) | 181 (19) | 31 (11) | 18 (14) |
| All patients N = 5155 | Normal weight BMI 18.5–24.9 kg/m2n = 1683 | Underweight BMI < 18.5 kg/m2n = 85 | Overweight BMI 25–29.9 kg/m2n = 2058 | Mild obesity BMI 30–34.9 kg/m2n = 937 | Moderate obesity BMI 35–39.9 kg/m2n = 270 | Severe obesity BMI >40 kg/m2n = 122 | |
|---|---|---|---|---|---|---|---|
| Age, years | 70 (60–77) | 71 (61–78) | 73 (62–81) | 70 (60–77) | 68 (58–75)** | 66 (57–75) ** | 64 (53–73)** |
| Gender, males, n (%) | 3641 (71) | 1166 (69) | 44 (52)** | 1566 (76)** | 658 (70) | 147 (54)** | 63 (52)** |
| BMI, kg/m2 | 26.8 (23.9–30.1) | 23 (21.6–24.1) | 17.5 (16.6–18.0)** | 27.3 (26.1–28.4)** | 31.8 (30.8–33.1)** | 36.9 (35.8–38.4)** | 42.6 (41.1–45.3)** |
| LVEF, % | 35 (27–45) | 33 (25–42) | 34 (25–47) | 35 (27–45)** | 36 (29–48)** | 40 (30–53)** | 40 (30–55)** |
| HFrEF, % | 3226 (63) | 1153 (68) | 57 (67) | 1291 (63)** | 545 (58)** | 121 (45)** | 59 (48)** |
| HFmrEF, % | 947 (18) | 277 (17) | 8 (9) | 412 (20)** | 169 (18)** | 62 (23)** | 19 (16)** |
| HFpEF, % | 982 (19) | 253 (15) | 20 (24) | 355 (17)** | 223 (24)** | 87 (32)** | 44 (36)** |
| Ischaemic aetiology, n (%) | 2543 (49) | 836 (50) | 30 (35)** | 1070 (52)** | 458 (49) | 111 (41)** | 38 (31)** |
| NYHA Class III-IV, n (%) | 1610 (31) | 521 (31) | 38 (45) | 571 (28) | 267 (29) | 112 (42)** | 59 (48)** |
| Atrial fibrillation, n (%) | 1138 (22) | 312 (19) | 22 (26) | 463 (23) | 237 (25)** | 69 (26) | 35 (29) |
| Hypertension, n (%) | 3133 (61) | 879 (52) | 34 (40)** | 1242 (60)** | 671 (72)** | 208 (77)** | 99 (81)** |
| Diabetes mellitus, n (%) | 1834 (36) | 473 (28) | 14 (17)** | 737 (36)** | 403 (43)** | 135 (50)** | 72 (59)** |
| Dyslipidaemia, n (%) | 2507 (49) | 730 (43) | 20 (24)** | 1033 (50)** | 510 (54)** | 142 (53)** | 72 (59)** |
| LA volume, ml/m2 | 46 (42–51) | 45 (40–50) | 43 (37–50) | 47 (42–51)** | 47 (43–52)** | 48 (43–52)** | 49 (44–53)** |
| EDLVD, mm | 59 (52–65) | 58 (52–64) | 44 (36–51) | 59 (53–65) | 59 (52–65) | 59 (52–64) | 57 (51–66) |
| ESLVD, mm | 47 (39–54) | 47 (40–54) | 34 (25–47) | 47 (39–55) | 45 (37–53) | 45 (37–53) | 44 (37–53) |
| TAPSE, mm | 18 (15–22) | 18 (14–21) | 19 (14–21) | 18 (15–22) | 19 (16–23)** | 20 (17–25)** | 20 (15–25) |
| sPAP, mmHg | 44 (36–55) | 45 (37–55) | 44 (39–58) | 44 (36–55) | 42 (36–53) | 45 (37–54) | 48 (36–56) |
| Hb, g/dl | 13.1 (11.8–14.3) | 12.9 (11.6–14.1) | 12.4 (11.3–13.8) | 13.3 (12–14.4)** | 13.4 (12.2–14.6)** | 13.2 (11.7–14.5) | 13.1 (11.6–14.4) |
| Creatinine, mg/dl | 1.1 (0.9–1.4) | 1.1 (0.9–1.4) | 0.9 (0.7–1.3) | 1.1 (0.9–1.4) | 1.0 (0.8–1.4) | 1.0 (0.8–1.3) | 1.0 (0.8–1.4) |
| eGFR, ml/min per 1.73 m2 | 67 (45–86) | 65 (44–85) | 70 (43–91) | 66 (45–86) | 68 (48–87) | 68 (48–87) | 68 (47–87) |
| NT-proBNP, ng/l | 1422 (538–3635) | 2095 (788–5071) | 2864 (1176–7818) | 1355 (534–3348)** | 1022 (390–2135)** | 761 (304–1744)** | 916 (369–2140)** |
| β-blockers, n (%) | 4070 (79) | 1355 (81) | 159 (69) | 1620 (79) | 719 (77) | 214 (79) | 103 (84) |
| ACEi/ARBs, n (%) | 4078 (79) | 1309 (78) | 69 (81) | 1606 (78) | 772 (83) | 205 (76) | 94 (85) |
| MRAs, n (%) | 2448 (48) | 782 (47) | 39 (46) | 968 (47) | 451 (48) | 140 (52) | 68 (56) |
| Furosemide, n (%) | 3924 (76) | 1255 (75) | 66 (77) | 1591 (77) | 709 (76) | 208 (77) | 95 (78) |
| Digoxin, n (%) | 1067 (21) | 342 (20) | 13 (15) | 440 (21) | 184 (20) | 62 (23) | 26 (21) |
| OAC, n (%) | 1763 (34) | 564 (34) | 28 (33) | 712 (35) | 312 (33) | 100 (37) | 47 (39) |
| CRT/ICD, n (%) | 1150 (23) | 392 (24) | 7 (8) | 521 (26) | 181 (19) | 31 (11) | 18 (14) |
Values are n (%) or median (interquartile range).
*p = 0.05 versus normal weight.
**p = 0.01 versus normal weight.
ACEi: angiotensin converter enzyme inhibitor; ARB: angiotensin II receptor blocker; BMI: body mass index; CRT: cardiac resynchronization therapy; EDLVD: end-diastolic left ventricular diameter; ESLVD: end-systolic left ventricular diameter; eGFR: estimated glomerular filtration rate (through CKD-EPI formula); Hb: haemoglobin; HFmrEF: heart failure with mid-range left ventricular ejection fraction; HFpEF: heart failure with preserved left ventricular ejection fraction; HFrEF: heart failure with reduced left ventricular ejection fraction; ICD: implantable cardioverter-defibrillator; LA: left atrium; LVEF: left ventricular ejection fraction; MRA: mineralocorticoid receptor antagonist; NT-proBNP: N-terminal pro-B-type natriuretic peptide; NYHA: New York Heart Association; OAC: oral anticoagulants; sPAP: systolic pulmonary arterial pressure; TAPSE: tricuspid annular plane systolic excursion.
Distribution and clinical correlates of BMI categories in heart failure
As shown in Table 1, 33% (n = 1683) of patients had normal weight, 2% (n = 85) underweight, 40% (n = 2058) overweight, 18% (n = 937) mild obesity, 5% (n = 270) moderate obesity and 2% (n = 122) severe obesity. HFrEF was the single most common diagnosis in each BMI class; still, HFmrEF and HFpEF were more prevalent in moderate-to-severe obesity. The relative prevalence of HFmrEF, albeit lower in underweight patients, was similar across the other BMI categories (Figure 1). Non-ischaemic aetiology was more common in the extreme classes of BMI. Circulating NT-proBNP levels for each BMI category are shown in Table 1 and Supplemental Material Figure 1 online. At multivariable linear regression analysis, BMI predicted NT-proBNP levels independently from LVEF class, heart failure aetiology and eGFR (r = –0.155, p = 0.001).
Prevalence of HFrEF, HFmrEF and HFpEF, in each body mass index category. Heart failure with reduced ejection fraction (HFrEF) was the single most common diagnosis in each body mass index (BMI) class; taken together, heart failure with mid-range and with preserved ejection fraction (HFmrEF and HFpEF respectively) were more prevalent in moderate-to-severe obesity.
BMI and outcome in HFrEF, HFmrEF, and HFpEF
Median follow-up was 40 months (IQR 17–83 months). During this period, 2152 patients died, of whom 1148 with cardiac death (938 for heart failure progression, 222 for sudden cardiac death and 102 for fatal myocardial infarction).
Among normal weight patients, five-year crude mortality rate was 39%. Only underweight patients survived less (62% five-year mortality rate), while overweight and obese patients had better prognosis (34%, 26%, 24% and 32% five-year mortality rate in overweight, mild obesity, moderate obesity, and severe obesity, respectively; log-rank = 87.26, p = 0.001) (Supplemental Figure 2). Similar findings were observed when considering separately HFrEF (log-rank = 49.59, p = 0.001), HFmrEF (log-rank = 31.03, p = 0.001) and HFpEF (log-rank = 16.45, p = 0.006) (Supplemental Figure 3), and for the secondary endpoint of five-year cardiac death (log-rank = 29.85, p = 0.001).
In the whole population, underweight was an independent predictor of worse outcome, while BMI categories ranging 25–39.9 kg/m2 predicted better outcome (Supplemental Table 1). Underweight was independently associated with poor prognosis in both HFrEF (hazard ratio 2.43 (95% CI 1.69–3.47), p = 0.001) and HFmrEF (hazard ratio 3.24 (95% CI 1.13–9.27), p = 0.031); mild obesity was independently associated with better outcome in HFrEF (hazard ratio 0.78 (95% CI 0.64–0.95), p = 0.020), HFmrEF (hazard ratio 0.63 (95% CI 0.41–0.96), p = 0.029) and HFpEF (hazard ratio 0.60 (95% CI 0.42–0.88), p = 0.008). Furthermore, while underweight predicted five-year all-cause death in patients with both ischaemic and non-ischaemic heart failure, increased BMI ranging 25–39.9 kg/m2 predicted better prognosis only in the subset of patients with non-ischaemic heart failure (Figure 2). Forcing atrial fibrillation or LVEF as a continuous variable in the multivariable model yielded similar results. Association of BMI classes with five-year all-cause death in HFrEF, HFmrEF and HFpEF of ischaemic and non-ischaemic aetiology is reported in Supplemental Table 2.
Prognostic significance of body mass index categories in either ischaemic or non-ischaemic heart failure. Forrest plot with adjusted hazard ratios for five-year all-cause mortality for each body mass index (BMI) category among patients with either ischaemic or non-ischaemic heart failure. Underweight was associated with worse outcome independently of heart failure aetiology. Conversely, increased BMI ranging 25–39.9 kg/m2 predicted better prognosis only in non-ischaemic heart failure. CI: confidence interval; HF: heart failure; HR: hazard ratio; ref: reference.
Finally, at competing-risk regression analysis, mild obesity and underweight were independently associated with a lower (subdistribution hazard ratio (SHR) 0.78 (95% CI 0.62–0.97), p = 0.026) and a higher risk of five-year cardiac death (SHR 1.63 (95% CI 1.04–2.55), p = 0.032), respectively. Underweight patients had a higher risk of non-cardiac death (SHR 1.93 (95% CI 1.19–3.13), p = 0.008); conversely, the risk of non-cardiac death was lower in overweight (SHR 0.82 (95% CI 0.68–0.98), p = 0.029) and mildly (SHR 0.72 (95% CI 0.57–0.92), p = 0.008) and moderately obese patients (SHR 0.54 (95% CI 0.34–0.87), p = 0.012), compared with normal weight.
PBF and outcome in HFrEF, HFmrEF and HFpEF
Using the Jackson–Pollock equation, the third PBF tertile was independently associated with better prognosis in HFrEF (hazard ratio 0.78 (95% CI 0.63–0.95), p = 0.013), in HFmrEF (hazard ratio 0.54 (95% CI 0.33–0.88), p = 0.013) and in HFpEF (hazard ratio 0.49 (95% CI 0.33–0.72), p = 0.001). Similar results were observed using the Gallagher equation, as PBF ≥33.7% was associated with better prognosis in HFrEF (hazard ratio 0.77 (95% CI 0.63–0.95), p = 0.014), in HFmrEF (hazard ratio 0.56 (95% CI 0.34–0.92), p = 0.021) and in HFpEF (hazard ratio 0.51 (95% CI 0.34–0.75), p = 0.001).
When classifying patients according to heart failure aetiology, both the second and the third PBF tertiles were independently associated with better outcome only in non-ischaemic heart failure, both using Jackson–Pollock (hazard ratio 0.76 (95% CI 0.62–0.93), p = 0.009 for the second, and hazard ratio 0.47 (95% CI 0.37–0.61), p = 0.001 for the third PBF tertile) and Gallagher equation (hazard ratio 0.78 (95% CI 0.63–0.96), p = 0.017 for the second hazard ratio 0.49 (95% CI 0.38–0.63), p = 001 for the third PBF tertile). Conversely, no prognostic effect was observed in patients with ischaemic heart failure. Association of PBF classes with five-year all-cause death in HFrEF, HFmrEF and HFpEF of ischaemic and non-ischaemic aetiology is reported in Supplemental Tables 3 and 4.
At competing-risks regression analysis, the third PBF tertile was independently associated with a lower risk of five-year cardiac death both when using the Jackson–Pollock equation (SHR 0.79 (95% CI 0.64–0.99), p = 0.041) and the Gallagher equation (SHR 0.78 (95% CI 0.62–0.97), p = 0.030).
Finally, we also tested the prognostic value of BSA, observing similar results to the PBF (Supplemental Table 5).
Discussion
The results of our large double-centre prospective study show, for the first time in a dedicated analysis, that obesity paradox also exists in patients with HFmrEF, and that obesity-related prognostic benefit is restricted to patients with non-ischaemic heart failure. Finally, circulating NT-proBNP levels are inversely correlated to BMI, independently of LVEF class, heart failure aetiology and renal function.
Obesity paradox in HFmrEF
The prognostic role of BMI was firstly explored in patients with HFrEF, in which obesity was observed to be associated with better outcome.3 In the last decade, several studies have confirmed this prognostic advantage, which has also been described in patients with HFpEF.13 Nevertheless, to the best of our knowledge, no study has specifically addressed the presence of obesity paradox in HFmrEF.
Patients diagnosed with HFmrEF are known to have an intermediate phenotype between HFrEF and HFpEF, for both clinical and bio-humoral features.30–32 In our population, we report that overweight was consistently the most represented BMI category in HFrEF, HFmrEF and HFpEF (40%, 44% and 36%, respectively). The risk of mortality attributable to differences in BMI category was similar among LVEF classes, with mild obesity emerging as protective and underweight as detrimental. Thereby, obesity paradox is similarly observed in patients with HFrEF, HFmrEF and HFpEF, in spite of different all-cause and cardiac mortality rate previously reported in each LVEF class.22
Notably, we reported similar results also using either PBF or BSA instead of BMI. Indeed, the highest PBF/BSA tertiles were independently associated with better outcome across the whole spectrum of LVEF.
Obesity paradox and heart failure aetiology
The link between obesity, atherosclerosis and coronary artery disease may explain our observation that both higher BMI and PBF are associated with improved outcome only in non-ischaemic heart failure. Indeed, both coronary artery disease and obesity are frequently associated with other conditions, such as systemic hypertension, diabetes mellitus, dyslipidaemia and obstructive sleep apnoeas, holding a negative impact on outcome.31,33 The prognostic benefit of obesity may thus be nulled in heart failure of ischaemic aetiology.
In a previous smaller study by Arena and co-workers,15 obesity paradox was observed in both ischaemic and non-ischaemic heart failure. Beyond sample size, several differences in the clinical features of the study populations could be responsible for the discrepancy with our observations. Indeed, in the study by Arena, the population was younger and with a less severe disease profile, and included only HFrEF patients. Conversely, the present study is expanding in a larger population the evidence previously reported in 500 patients by Zamora et al. about the lack of prognostic benefit by obesity in patients with heart failure of ischaemic origin.16 Furthermore, in the present study, we stratified patients according the latest WHO indication for BMI categories,26 while other studies used outdated classifications, neither distinguishing among underweight and normal weight patients, nor sub-stratifying obesity severity.15
Notably, despite underweight patients accounting for only a minority of the whole population, we consistently report a worse outcome in such subgroup, independently of heart failure aetiology and of left ventricular systolic function. Cardiac cachexia, that is, progressive weight loss and body-functions deteriorating, as frequently observed in end-stage heart failure, may account for this finding, in line with previous observations.34,35
NT-proBNP paradox in HFrEF, HFmrEF and HFpEF
Independently of disease severity and other possible confounders, NT-proBNP levels are usually higher in patients with HFrEF than in HFmrEF and HFpEF, while they are consistently associated with outcome in all LVEF categories.16 Moreover, circulating levels of NT-proBNP have been reported to be influenced by BMI.17,32,36 Still, while circulating levels of NT-proBNP were shown to inversely correlate with BMI both in HFrEF and in HFpEF, we report for the first time a similar relation also in HFmrEF.36,37 Indeed, in our study we observed the highest NT-proBNP levels in underweight and normal weight patients, while the nadir was observed in mild-to-moderately obese patients in either HFrEF, HFmrEF and HFpEF.
The causes underlying the relationship between BMI and NT-proBNP are poorly understood. Although a decreased production of natriuretic peptides in obese patients with heart failure, attributed to lower cardiac wall stress, has been proposed, a pathophysiologic explanation is still lacking.38 Increased clearance of natriuretic peptide due to either fat-cell-dependent enzymatic activity or increased glomerular filtration were envisaged as other possible causes.17 Nevertheless, in our population, such inverse relation persisted even after the model had been adjusted for eGFR.
Furthermore, a potential role for steroid hormones and cytokine/adipokine interactions was hypothesized as well, but preclinical and clinical studies are still missing or inconclusive.17,39
Study limitations
The first limit of the present study was that while BMI does not allow to distinguish lean from fat mass, PBF was estimated using equations and not directly measured. Nevertheless, in the last decades, the obesity paradox has been confirmed through many other measures, more reliable but often more difficult to be applied in the clinical routine, especially in large population studies. In particular, evaluation of skinfolds, waist circumference, bioelectrical impedance and advanced imaging were tested in smaller populations and proved to correlate with BMI and to be associated with outcome in heart failure patients (as recently reviewed34). Further, BMI may be an indirect index of nutritional status, which has a major prognostic impact in heart failure.40 Secondly, the relatively low number of underweight and extreme-obese patients may limit the robustness of our findings in these populations when multiple sub-classifications (based on both LVEF and heart failure aetiology) were performed. Third, patients were only evaluated at recruitment, thus we could not account for possible changes in BMI, LVEF or other parameters. Fourth, eGFR was as the result of the CKD-EPI formula, which is considered the most reliable estimation method; nevertheless, there could have been some under- or overestimation among the extreme BMI classes. Finally, although all echocardiographic examinations were performed by experienced physicians, inter- and intra-observer reproducibility of LVEF measurements could not be assessed.
Conclusions
Obesity paradox is present across the whole spectrum of LVEF, including patients with HFmrEF. Nevertheless, such prognostic advantage for obese patients is maintained only when heart failure aetiology is non-ischaemic, but not when ischaemic, also in HFmrEF patients. Finally, NT-proBNP levels inversely correlate with BMI in patients with heart failure, independently of LVEF class, heart failure aetiology and renal function.
Supplementary material
Supplementary material is available at European Journal of Preventive Cardiology.
Author contribution
Conception of the study: GV, FG, ME, JL, AB. Data acquisition, analysis and interpretation: GV, FG, AA, CP, PS, ME, JL, EZ, MdA, ES, MD, PC, AG. Drafting the work or revising it critically for important intellectual content: GV, FG, AA, ME, JL, AB.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
References
National Heart, Lung and Blood Institute. Classification of overweight and obesity by BMI, waist circumference, and associated disease risks, http://www.nhlbi.nih.gov/health/public/heart/obesity/lose_wt/bmi_dis.htm (accessed 26 January
Comments