Body mass and survival in patients with chronic heart failure without cachexia: The importance of obesity☆

Methods 

We evaluated 589 unselected consecutive patients with chronic heart failure presenting at the Heart Failure Clinic of the Royal Brompton Hospital between March 1991 and January 2000 (Table 1). This is a mixed secondary and tertiary referral center for cardiology. The diagnosis of heart failure was made on the basis of symptoms of shortness of breath and symptomatic exercise intolerance with signs of pulmonary congestion or peripheral edema in the presence of objective evidence of left ventricular systolic dysfunction by echocardiography or radionuclide ventriculography.

Table 1. Clinical data for 525 patients with chronic heart failure*

		Noncachectic (n = 525)	Cachectic (n = 64)	P
	Age (y)	61.0 ± 12.4	65.4 ± 12.0	.006
	Male (%)	83	94
	NYHA	I–101	I–1
		II–230	II–16
		III–152	III–32
		IV–42	IV–15
	IHD (%)	57	59
	LVEF (%)	31.6 ± 14.8 (n = 369)	26.3 ± 15.1 (n = 57)	.014
	LVEDD (cm)	6.5 ± 1.1 (n = 368)	6.7 ± 1.2 (n = 46)	.21
	Height (cm)	171.4 ± 9.5	171.6 ± 9.8	.89
	Weight (kg)	80.8 ± 15.2	63.6 ± 10.3	<.0001
	BMI	27.5 ± 4.3	21.5 ± 2.5	<.0001
	Ideal weight (%)	119.7 ± 19.0	93.4 ± 11.1	<.0001
	Sodium (mmol/L)	138.2 ± 3.3	136.9 ± 4.0	.006
	Heart rate (/min)	88.2 ± 16.6	83.7 ± 17.7	.80
	Systolic BP (mm Hg)	124.5 ± 23.5	109.6 ± 25.7	<.0001
	Peak VO2 (mL·kg−1·min−1	18.2 ± 6.5 (n = 440)	15.0 ± 4.8 (n = 62)	.0002
	Peak VO2 (mL·min−1)	1488 ± 29	955 ± 42 (n = 62)	<.0001
	VE/VCO2 slope	36.1 ± 11.7	46.1 ± 15.2	<.0001
	*Sixty-four patients with cachexia shown for comparison.

NYHA, New York Heart Association; IHD, ischemic heart disease; LVEF, left ventricular ejection fraction; LVEDD, left ventricular end-diastolic dimension from two-dimensional echocardiography; BMI, body mass index; VO2, oxygen consumption; VE/VCO2 slope, the relation between ventilation and carbon dioxide production. Measurements from two-dimensional scans were not available in all patients.

After the diagnosis was made, the medical regimen of all the enrolled patients was optimized. Medications included digoxin, angiotensin-converting enzyme inhibitors, diuretics, nitrates, β-blockers, statins, and aspirin or warfarin in varying combinations. Details are given in Table 2. Cachectic patients were more likely to be taking digoxin, but there were no other significant differences between the cachectic and noncachectic patients.

Table 2. Drug treatment*

	Patients Taking (%)	Noncachectic (n = 525)	Cachectic (n = 64)	P
	Loop diuretic	84.6	82.0	>0.2
	ACE inhibitor	48.5	72.6	0.19
	ARB	35.5	14.5
	Calcium antagonists	16.2	9.4	0.17
	Digoxin	20.8	40.6	0.013
	Amiodarone	16.0	20.3	>0.2
	Warfarin	25.7	35.9	0.061
	Aspirin	49.1	39.1	0.16
	Lipid-lowering	18.7	9.4	>0.2
	Beta-blocker	7.0	1.6	0.07
	*The percentage of patients taking each particular agent is shown.

ACE, angiotensin-converting enzyme; ARB angiotensin receptor blocker. P values shown are for the comparison between noncachectic and cachectic patients. A single value is shown for ACE inhibitor + ARB.

At presentation, all patients were examined clinically. Weight and height were measured and the BMI (BMI = weight/height2) was calculated. The medical notes were reviewed and a history of weight change was carefully taken to document weight loss. Cardiac cachexia was defined as documented nonedematous and nonintentional weight loss of >7.5% of the previous nonedematous weight over at least 6 months (n = 64). No patient was suspected either clinically or after examination or investigation to have any other primary cachectic state (infection, malignant disorder, thyroid disease, severe liver disease, or acquired immunodeficiency syndrome). The noncachectic heart failure patients had no history of significant nonedematous weight loss in the 2 years before the study and were documented not to meet the definition of cachexia. We used the European Society of Cardiology Guidelines10 for heart failure treatment to define patients as overweight and obese.

All subjects underwent echocardiography to assess cardiac function. Resting left right ventricular ejection fraction (LVEF) and right ventricular ejection fraction were determined in 426 (72%) and 190 (32%) patients by radionuclide ventriculography.

Symptom-limited cardiopulmonary exercise testing with respiratory gas exchange was performed in 502 patients (85%) using the modified Bruce protocol, in which there was a stage zero of 1 mph at 5% gradient.11 Oxygen consumption, minute ventilation, and carbon dioxide production were measured online every 10 seconds by a standard inert gas dilution technique (Amis 2000, Odense, Denmark).12 Patients were encouraged to exercise to exhaustion. The peak oxygen consumption (peak VO2) was defined as the highest VO2 observed during exercise. The ventilatory response to exercise was measured as the slope relating ventilation to carbon dioxide production in 500 patients. All but 52 patients (10%) achieved a respiratory exchange ratio >1.00, indicating the achievement of anaerobic exercise conditions (mean 1.14 ± 0.14).

Results 

Clinical data for the 525 patients with no history of weight loss are shown in Table 1. For comparison, data on the group of 64 cachectic heart failure patients are also shown. The cachectic patients were, on average, 4 years older and tended to have worse heart failure as evidenced by worse exercise capacity, lower blood pressure, and a lower LVEF.

In the noncachectic group, 84.0% of patients were receiving an angiotensin-converting enzyme inhibitor or angiotensin II receptor antagonist. The proportion was 87.1% in the cachectic group; 9% and 11% were not receiving diuretic therapy in the two groups, respectively. Other drug therapy is shown in Table 2. Consistent with the time of recruitment into the study, there was a low level of β-adrenoceptor antagonist use.

In the noncachectic patients, peak VO2 was strongly related to body mass (r = .5; P < .0001). In further analysis, we thus used peak VO2 corrected for body weight.

Mean follow-up in the survivors was 53.0 ± 25.2 months. At 1 January 2001, 167 of the noncachectic and 41 of the cachectic patients had died. Average time from study entry to death was 24.9 ± 22.0 months and 19.2 ± 18.1 months in the two groups, respectively. The overall relative risk for cachectic vs. noncachectic patients using Cox proportional hazards model was 2.71 (95% confidence interval [CI] 1.94–3.79). Kaplan-Meier survival plots are shown in Fig. 1.

View Large Image Download to PowerPoint

Fig. 1. Kaplan-Meier plots for survival. The patients are shown as quintiles of body mass in ascending order from lightest (Q1) to heaviest (Q5). Survival in cachectic patients (C) is shown for comparison.

To assess the effect of body mass on survival in patients without cachexia, we divided the patients into quintiles of body mass index. Comparative data for the 5 groups are shown in Table 3.

Table 3. Noncachectic patients divided into quintiles on the basis of body mass index*

		Q1 (n = 105)	Q2 (n = 105)	Q3 (n = 105)	Q4 (n = 105)	Q5 (n = 105)	P value (trend)
	Age (y)	61.2 ± 15.7	61.5 ± 11.2	61.7 ± 12.7	61.2 ± 10.3	59.2 ± 11.1	0.61
	Ideal weight (%)	96.8 ± 7.0	109.0 ± 4.4	117.4 ± 5.0	128.0 ± 5.4	147.3 ± 14.0	—
	BMI	22.2 ± 1.5	24.9 ± 0.6	26.9 ± 0.6	29.2 ± 0.8	34.1 ± 2.8	—
	Diagnosis (IHD%)	49.5	61.9	56.2	59.0	58.1	0.45
	Peak VO2 (mL·kg−1·min−1)	18.7 ± 7.3	18.9 ± 6.4	18.2 ± 7.3	18.6 ± 6.2	16.6 ± 5.1	0.14
	Peak VO2 (mL·min−1)	1236	1441	1446	1622	1670	<0.001
		62	62	66	59	63
	VE/VCO2 slope	37.9 ± 13.9	35.9 ± 10.1	37.1 ± 14.6	34.3 ± 10.2	35.8 ± 9.2	0.31
	LVEF (%)	30.2 ± 14.6	29.6 ± 13.5	29.9 ± 14.9	33.2 ± 15.3	34.5 ± 15.3	0.12
	LVEF > 45% (%)	14	12	12	25	21	0.13*
	LVEF < 15% (%)	12	12	18	7	10	0.30
	LVEDD (cm)	6.6 ± 1.2	6.5 ± 1.0	6.4 ± 1.2	6.4 ± 1.1	6.4 ± 1.2	0.83
	Sodium (mmol/L)	137.9 ± 2.8	138.2 ± 3.5	138.2 ± 3.5	139.0 ± 3.1	137.6 ± 3.4	0.12
	Heart rate (/min)	81.0 ± 14.2	82.8 ± 19.3	83.7 ± 17.3	84.1 ± 16.2	84.1 ± 15.6	0.73
	Systolic BP (mm Hg)	120.0 ± 22.9	122.4 ± 23.6	124.1 ± 22.5	131.0 ± 26.4	124.6 ± 20.1	0.03
	Diastolic BP (mm Hg)	74.4 ± 11.9	74.3 ± 10.6	75.1 ± 11.5	79.7 ± 13.2	78.3 ± 10.4	0.004
	*Other than the differences between the groups in body mass index and % ideal weight, there was no significant difference between the groups in any of the variables reported apart from blood pressure. Diagnosis is presented as proportion of patients with an underlying diagnosis of ischemic heart disease.

LVEF > 45% is the proportion of each group with preserved LV systolic function. The P value for these row is from a X2 test. Other P values are for analysis of variance using all groups. Abbreviations as in Table 1.

Table 4. Survival rates for the 5 quintiles of body mass index

		1-y Survival (%)	3-y Survival (%)	Relative Risk	P
	Q1	75 (66–83)	63 (54–73)	2.3 (1.4–3.8)	.001
	Q2	91 (85–96)	71 (62–80)	1.7 (1.1–2.9)	.027
	Q3	86 (79–92)	70 (61–79)	1.8 (1.1–3.0)	.016
	Q4	91 (85–96)	81 (73–89)	1.0	—
	Q5	94 (90–99)	77 (68–85)	1.5 (0.9–2.5)	.14

95% confidence intervals for the survival rates are shown. The P values are for the comparison between each of the other quintiles and the quintile with the best survival (Q4).

The survival overall for patients in the lowest quintile was greater than for cachectic patients (hazard ratio 0.55; 95% CI 0.36–0.84; P = .005), but very similar over the first 2 years (2-year survival: 53% (46%–61%) quintile 5; 47% (35%–59%) cachectics; log rank X2 1.6: P = .21). Table 5 shows the survival for patients based on whether the patients might have been recommended to lose weight based on the European Society guidelines.10

Table 5. Survival rates by presence of overweight and obesity10

		n	1-y Survival (%)	3-y Survival (%)	Relative Risk	P
	Without cachectics
	BMI <25	167	80 (74–86)	47 (40–55)	1.5 (1.1–2.0)	.01
	Overweight (BMI >25)	358	91 (88–93)	60 (53–66)
	With cachectics
	BMI <25	227	77 (72–83)	43 (37–50)	1.8 (1.3–2.3)	<.0001
	Overweight (BMI >25)	362	90 (87–93)	57 (52–62)

95% confidence intervals for survival are shown. The relative risk is the risk of not being overweight relative to being overweight. BMI, body mass index.

Univariate predictors of survival are shown in Table 6.

Table 6. Single predictors of outcome

	Predictor	Coefficient	95% Cl	P
	Age	1.03	1.01–1.04	<.0001
	BMI	0.88	0.83–0.94	.0002
	Diagnosis	0.60	0.44–0.81	.001
	NYHA			<.0001
	Peak VO2	0.90	0.87–0.92	<.0001
	Serum Na+	0.91	0.87–0.95	<.0001
	Systolic blood pressure	0.99	0.98–0.99	<.0001

The coefficient describes the change in relative risk per unitary increase in a continuous variable. For diagnosis, the relative risk for dilated vs. ischemic etiology is shown. For NYHA, the risks relative to class IV were: I, 0.116; II, 0.235; III, 0.473. Abbreviations as in Table 1.

There was no significant difference in the duration of heart failure between the 5 quintiles and between the quintiles and the patients with cachexia, but a relation between duration of heart failure (per month) and survival (1.005 [1.001–1.008]; P = .01). Age (1.01 [0.98–1.05]) and diagnosis (1.56 [0.78–3.11]) were not predictors of survival.

Discussion 

Obesity is an epidemic of Westernized societies13 and is strongly related to the development of cardiovascular disease.14, 15, 16 In apparently healthy subjects, the risk or coronary disease starts rising above a BMI of approximately 23.5 to 24.9 in men17 and approximately 22 in women.18, 19 By contrast, weight loss lowers the risk factors associated with coronary disease in the obese,20 although there is some evidence to suggest that both voluntary and nonvoluntary weight loss are associated with increases in all-cause mortality in otherwise apparently healthy people.21, 22

Cachexia is associated with a poor outcome in a variety of disease states—for example, in chronic renal failure,23 after cardiac surgery,24, 25 in patients with dementia26 and chronic lung disease,27 as well as malignant disease (for review see references 28 and 29). We have previously demonstrated that cachexia is associated with a poor prognosis in patients with chronic heart failure2 in whom cachexia was defined as a process of weight loss.

The present study confirms these results in a larger group of patients. We were mainly concerned about addressing the issue of whether body mass affected survival in the absence of active weight loss. Previous investigators have suggested that there is a relation between increasing BMI and improved survival,7 but did not exclude patients with cachexia. The patients in this study were documented not to have lost weight before the study. We have found that the best survival was seen in moderately obese subjects, with a relative risk of 2.3 in the quintile of patients with an “ideal” BMI of 22 (in this study, these patients are in the lightest quartile with a BMI of 22.2 ± 1.5).16 Increasing BMI was a predictor of 1-year survival independent of exercise capacity and left ventricular function.

Chronic heart failure is a catabolic state.3 The origins of catabolism remain uncertain, but are related neurohormonal systems activated as part of the response to the heart failure syndrome. Patients with moderate obesity may represent a group of patients with a higher metabolic reserve, who are better (and longer) able to tolerate the metabolic stress of heart failure than those with more “ideal” body weights. The heavier patients had higher blood pressure than the controls. This may represent an effect of obesity per se, rather than reflecting better cardiac function. Other variables suggest that the severity of heart failure was the same in all groups. The role of nutritional support for heart failure patients needs to be investigated further.30

The survival rate of the lowest quintile patients was similar to that of cachectic patients over the first 2 years of follow up (although not overall). This raises the possibility of misclassification of patients as non-cachectic, although we were careful only to include as patients with documented absence of weight loss as noncachectic. We have not been able to correct peak exercise capacity for lean body mass because few of these patients have had dual energy x-ray absorptiometry (DEXA) scanning, and it we cannot say whether one body compartment or another (fat or lean body mass) is most strongly related to survival. We tried to get an impression of the effect of correcting peak VO2 for lean body mass by using peak VO2/kg normalized for BMI. We found that this approach increased the statistical significance attributable to the effect of increasing BMI on survival.

A further possibility is that there is a U-shaped relation between BMI and survival in heart failure, with worse survival for the heaviest patients. There was no difference between the 2 heaviest quartiles in our analysis, but there were too few very heavy patients to answer this question with these data. The lightest quartile might have been underweight through excessive diuresis, which itself might carry an adverse prognosis. However, the diuretic dose was not significantly different between the 5 groups, and, at most, this factor could only have contributed 1–2 kg.

In summary, this study suggests that from the standpoint of survival, the “ideal” body weight is higher for heart failure patients than in the general population; this seems to be true regardless of whether the patient has an ischemic etiology.