Sex differences in the association between inflammation and event-free survival in patients with heart failure

Abstract

Background:

Heart failure (HF) is associated with chronic inflammation, which is adversely associated with survival. Although sex-related differences in inflammation have been described in patients with HF, whether sex-related differences in inflammation are associated with event-free survival has not been examined.

Aim:

To determine whether the association between inflammation as indicated by TNF-α and event-free survival differs between men and women with HF after controlling for demographic and clinical variables.

Method:

This was a secondary analysis of data from 301 male (age 61.0 ± 11.4 years) and 137 female (age 60.3 ± 12.1 years) patients with HF. Serum levels of soluble tumor necrosis factor receptor1 (sTNFR1) were used to indicate inflammatory status. Patients were grouped according to median split of sTNFR1 level and sex into: male with low inflammation (≤ 1820 pg/ml) n=158 or high inflammation (> 1820 pg/ml) n=143, and female with low inflammation n=63 or high inflammation n=74. Cox regression model was run separately for men and women to determine whether inflammation contributed to differences in event-free survival between sexes with HF.

Results:

There were 84 male (27.9%) and 27 female (19.7%) patients who had an event. Event-free survival in women did not differ by the severity of inflammation in the Cox regression analysis. In contrast, men with high inflammation had 1.85 times higher risk for an event compared to men with low inflammation.

Conclusion:

These data provide evidence that inflammation contributed to differences in event-free survival in men but not women with HF. Clinicians should be aware that men patients who have higher inflammation may be at greater risk of HF or cardiac-related events than others with HF.

Chronic inflammation is considered a major pathophysiologic mechanism involved in cardiac remodeling, eliciting symptoms, and progression of heart failure (HF).^1–3 This has been demonstrated to have implications for determining the prognosis of HF and guiding therapeutic management.^1,3 Inflammation in HF is primarily driven by the overproduction of proinflammatory factors.¹ Tumor necrosis factor-α (TNF-α) is a proinflammatory cytokine that has been consistently reported to be increased in both HF with reduced (HFrEF) and preserved (HFpEF) left ventricular ejection fraction (LVEF).^1,2 TNF-α is released by immune cells and other cells in the myocardium and peripheral tissues in response to decreased cardiac output or in response to comorbidities such as diabetes mellitus.^2,4 Researchers have implicated systemic TNF-α release in several pathologic pathways underlying the manifestations and progression of the HF including activation of neurohormones, physical activity intolerance, cognitive impairment, sodium and water retention, and cardiac contractility dysfunction.^2,4–6 Higher levels of TNF-α have been associated with the severity of HF and worse prognosis.^1–3

Researchers have consistently reported that inflammation may have a greater adverse effect in women than men.^7,8 Women have a greater risk of HF around the age of menopause, in which TNF-α level is significantly increased and the effects of sex hormones on inflammation are decreased.^8,9 Women have been noted to have higher TNF-α levels after myocardial injury and greater non-cardiac comorbidities risk of HF than men.^10,11 Women with HF are more likely to report a greater symptom severity and a lower quality of life than men.^12,13

These results suggest that women are more likely to experience a greater inflammation-related HF burden than men. However, no study has analyzed sex-specific differences in health outcomes related to inflammation in HF. Therefore, the purpose of this study was to determine whether the association between inflammation and event-free survival differs between men and women with HF after controlling for demographic and clinical variables.

Methods

Design and Sample

This was a secondary analysis of longitudinal data obtained from the RICH Heart Program data registry.¹⁴ A subsample of 438 patients from the registry with either HFrEF or HFpEF confirmed by a cardiologist who had data on serum sTNFR1 levels and event-free survival was included in the analyses. Patients were recruited from 4 HF clinics located in the community hospitals or academic medical centers in Kentucky, Georgia, Indiana, and Ohio. All patients had the same following inclusion criteria: 1) no previous hospitalization within 3 months before enrollment, 2) New York Heart Association functional class II-IV, 3) on stable doses of HF medications for at least 3 months before enrollment, 4) able to understand and speak English, 5) not status 1 on a cardiac transplant list or an implanted left ventricular assist device, 6) not prescribed anti-inflammatory medications, 7) no history of stroke within 3 months prior to enrollment, 8) no diagnosed chronic condition that causes systemic inflammation, 9) no terminal illnesses such as end-stage cancer, lung, or liver disease, and 10) no apparent cognitive impairment. Data on inclusion criteria were obtained from a patient interview or medical record review.

Procedure

Approval of the studies was obtained from Institutional Review Boards at each study site. Patients with HF attending HF clinics were screened for enrollment eligibility by trained research nurses. Eligible patients were contacted in the clinic and invited to participate. Signed informed consent was obtained from those who agreed to participate prior to data collection. Visits were scheduled at the clinical research center at each site for baseline data collection and to draw blood samples for measurements of biomarkers. All patients were followed for over 3 years or until an event of cardiac- or HF-related hospitalization or all-cause death.

Measurements of variables

Demographic.

Age, sex, marital status, and ethnicity were collected using a self-report questionnaire.

Biomarkers.

Soluble tumor necrosis factor receptor1 (sTNFR1) and N-terminal pro-B-type natriuretic peptide (NT-proBNP) levels were obtained from blood samples. All samples were frozen at −80°C and shipped to the primary study site on dry ice for analysis. Serum levels of sTNFR1 were used as the indicator of the systemic inflammatory activity of TNF-α.^2,4,15 TNFR1 is a protein found in the plasma membrane that becomes a soluble (sTNFR1) by shedding from the cell membrane in response to binding with TNF-α. Higher serum sTNFR1 levels are the result of greater cell binding of TNF- α indicating greater TNF- α activity. Serum sTNFR1 level is considered a good long-term prognostic marker in HF compared with TNFα, sTNFR2 soluble receptor, or other inflammatory biomarkers such as interleukin-6 or interleukin-1 receptor antagonist.¹⁶ Serum levels of sTNFR1 were measured in pg/ml using ELISA kits (R&D Systems, Minneapolis, MN). There is no established clinical diagnostic cutoff value of sTNFR1 in patients with HF. In previous clinical and experimental studies, the median value was the common method used for comparisons of those with higher to those with lower TNFR1 levels within the sample.¹⁷ Consequently, high inflammation status was defined as a sTNFR1 level greater than the sample median of 1820 pg/ml.

NT-proBNP was used to characterize our sample. Higher NT-proBNP levels indicate greater HF severity and worse HF prognosis.¹⁸ Serum levels of NT-proBNP were measured in fmol/ml using enzyme immunoassay kits (ALPCO Diagnostics).

The accuracy and reproducibility of the sTNFR1 and NT-proBNP measurements in our sample were demonstrated previously.^18,19

Event-free survival.

Event-free survival was defined as days elapsed between baseline data and the date of the first event defined as HF- or cardiac-related hospitalization or all-cause mortality. All-cause mortality was used because the cause of death is not consistently reliable.²⁰ Information for events was obtained by monthly telephone contact with patients or their family members. All events were verified by medical record review over a 3-year follow-up.

Body mass index (BMI).

BMI was included as a covariate because it has been directly associated with better outcomes.²¹ BMI was calculated as weight in kilograms per height in meters squared. Weight and height were assessed by a research nurse who was trained according to standardized measurement protocols.²²

New York Heart Association (NYHA).

NYHA was used to indicate HF severity as a covariate. NYHA classifies patients with HF in one of four functional classes based on the degree of physical activity limitation due to symptoms. The higher functional class indicates greater physical activity limitation due to greater symptom severity.¹⁸ NYHA functional class was determined by a trained research nurse through a standardized face-to-face interview. High interrater and intrarater reliabilities were established for classifying the patients in our sample.²³ For analysis, functional classes were categorized into III/IV versus I/II.

Comorbidity.

Comorbidity burden measured by the Charlson Comorbidity Index was included as a covariate. The index assesses the coexistence of 19 health conditions. Each condition is weighed a score range from 1 to 6 based on the risk of mortality associated with each one. Higher total scores indicate an increasing burden of comorbidities, with the maximum score being 34. Charlson Comorbidity Index has well-established reliability and validity.^24,25

Quality of life.

Quality of life was included as a covariate because it was associated with event-free survival.²⁶ Quality of life was measured by the Minnesota Living with Heart Failure Questionnaire. This instrument consists of 21 items measuring the impact of a different aspect of HF condition on various aspects of daily life on a 6-point scale ranging from 0=no impact to 5=very much impact. Total scores could range from 0 to 105, with higher scores indicating poorer quality of life. Reliability and validity have previously been demonstrated.²⁶

Other variables of interest.

Data on prescribed medication therapy were obtained during patients interviews and were validated by medical record review. Etiology of HF was classified as non-ischemic or ischemic HF and was retrieved from the medical record. Non-ischemic HF was defined as ventricular dysfunction secondary to causes other than coronary artery disease.²⁷ Left ventricular ejection fraction was retrieved from the medical record. Patients were classified as having either HFrEF defined as LVEF ≤ 40% or HFpEF.⁹

Data Analyses

For the purpose of this study, TNFR1 level and sex were used to categorize patients into 4 groups: males with low inflammation (sTNFR1 ≤ 1820 pg/ml), n=158 or high inflammation (>1820 pg/ml), n=143, and females with low inflammation, n=63 or high inflammation, n=74. This approach allowed us to examine gender-based differences in event-free survival related to inflammation.

Statistical analyses were performed using SPSS software for Windows (version 25.0; IBM Corporation). Patients’ demographic and clinical characteristics were summarized as means and standard deviations, medians and interquartile ranges, or frequencies and percentages. Comparisons for continuous variables between men and women were performed using independent-sample t tests. One-way analysis of variance with the least significant difference post hoc tests was conducted to determine differences among the 4 sex/inflammatory status groups. All comparisons for categorical variables were accomplished using ϰ² test of association. The NT-proBNP and sTNFR1 levels were transformed for analyses to approximate normality. All other variables were normally distributed.

Cox regression analysis was conducted separately for each sex to determine whether the association between inflammation as a categorical variable (low versus high) and event-free survival differed between men and women with HF. Cox regressions included age, ethnicity, BMI, LVEF, HF etiology (ischemic versus none ischemic), NYHA (I/II versus III/IV), comorbidity burden, and quality of life as covariates. The outcome was defined as a composite event-free survival endpoint of HF- or cardiac-related hospitalization or all-cause mortality. A 95% confidence interval was considered for hazard ratios and statistical significance was defined as a P value ≤ .05.

Results

Patient characteristics

Baseline demographic and clinical characteristics and comparison among men and women in the low and high inflammation status groups are presented in Table 1. The mean age of patients was 60.8 (± 11.6) and was similar for men and women patients. The majority of patients were Caucasian, married or cohabitating. Approximately half of the patients were obese, in NYHA functional class III or IV, and had ischemic etiology of HF. On average, patients had an NT-proBNP level of 710 (± 550 fmol/ml) and a comorbidity burden of 2.9 (range: 1–8). A greater proportion of patients had HF with reduced ejection fraction (LVEF ≤ 40%) and was on angiotensin-converting enzyme inhibitors, β-blocker, diuretics, aspirin, and statin medications.

Table 1.

Sample demographic and clinical characteristics and comparisons among male and female patients with low or high inflammation status

Characteristics	Total sample (n = 438)	Male with low inflammation (n = 158)	Male with high inflammation (n = 143)	Female with low inflammation (n = 63)	Female with high inflammation (n = 74)
Age	60.8 ± 11.6	58.3 ± 11.4	63.5 ± 10.6a	56.1 ± 12.0	63.6 ± 10.7a
Race  White  Black or other minority	346 (79.0) 92 (21.0)	128 (81.0) 30 (19.0)	127 (88.8) 16 (11.2)	38 (60.3) 25 (39.7)	53 (71.6) 21 (28.4)
Married or cohabitate	250 (57.1)	103 (65.2)	94 (65.7)	28 (44.4)	25 (33.8)
BMI (kg/m2)  Obese ≥ 30 kg/m2	31.6 ± 7.3 241 (55.0)	30.1 ± 6.4 77 (48.7)	31.9 ± 7.3 79 (55.3)	31.9 ± 7.4 39 (61.9)	32.2 ± 9.1 46 (62.2)
LVEF (%)  HFrEF (≤ 40%)	35.0 ± 14.2 321 (72.1)	32.9 ± 12.9 124 (78.5)c	32.2 ± 13.5 112 (76.7)c	40.2 ± 14.6 41 (63.1)	40.6 ± 17.3 44 (57.9)
Etiology of heart failure  Ischemic  Non ischemic	194 (44.3) 244 (55.7)	78 (49.4)c 80 (50.6)	79 (55.3)c 64 (44.8)	13 (20.6) 50 (79.4)	24 (32.4) 50 (67.6)
NYHA (III/IV)	203 (46.3)	51 (32.3)	74 (51.8)a	32 (50.8)	46 (62.2)a
ACEI ARB Digoxin Diuretics β-blocker Nitrates Aldosterone antagonist Calcium channel blocker Anti-dyslipidemia Aspirin Antidepressant	300 (68.5) 75 (17.1) 106 (24.2) 317 (72.4) 379 (86.5) 77 (17.6) 99 (22.6) 59 (13.5) 293 (66.9) 249 (56.9) 106 (24.2)	125 (79.1) 23 (14.6) 39 (24.7) 101 (63.9) 139 (88.0) 29 (18.4) 30 (19.0) 21 (13.3) 121 (76.6) 94 (59.5) 25 (15.8)	99 (69.2) 21 (14.7) 30 (21.0) 111 (77.6) 124 (86.7) 33 (23.1) 43 (30.1) 19 (13.3) 101 (70.6) 91 (63.6) 30 (21.0)	34 (54.0) 15 (23.8) 12 (18.8) 45 (70.3) 54 (84.4) 6 (9.5) 11 (17.5) 14 (22.2) 34 (54.0) 27 (42.2) 20 (31.3)	42 (56.8) 16 (21.6) 25 (33.8) 60 (81.1) 62 (83.8) 9 (12.2) 15 (20.3) 5 (6.7) 37 (50.0) 37 (50.0) 31 (41.9)b
Comorbidity score	2.9 ± 1.9	2.6 ± 1.7	3.6 ± 2.1a	2.2 ± 1.3	3.1 ± 2.0a
NT-proBNP (fmol/ml)	710.4 ± 550.2	639.1 ± 446.4	849.1 ± 554.0a	521.4 ± 357.6	732.2 ± 594.2a
sTNFR1 (pg/ml)	2052.1 ± 1025.2	2002.6 ± 947.4	2160.8 ± 1174.6
Quality of life	39.3 ± 24.8	35.8 ± 23.0	42.2 ± 25.3a	34.8 ± 26.2	42.8 ± 23.7a
Event-free survival  All-cause mortality  HF-related hospitalization  Cardiac-related hospitalization  Total Events  Median time to first event (days)	10 (2.3) 39 (8.9) 62 (14.2) 111 (25.3) 194 (80, 396)	2 (1.3) 10 (6.3) 26 (12.7) 38 (24.1) 182 (91, 336)	6 (4.2) 20 (14.0) 20 (18.2) 46 (32.2) 200 (80, 378)	1 (1.6) 3 (4.8) 10 (15.9) 14 (22.2) 391 (187, 802)b	1 (1.4) 6 (8.1) 6 (8.1) 13 (17.6) 35 (11, 346)

Values are mean ± SD, frequency (percent), or median (25th, 75th quartile).

sTNFR1: soluble tumor necrosis factor receptor1; NT-proBNP: N-terminal pro-B-type natriuretic peptide; NYHA: New York Heart Association functional class; LVEF: left ventricular ejection fraction; BMI: body mass index; ACEI: angiotensin converting enzyme inhibitor; ARB: angiotensin II receptor blocker; HFrEF: heart failure with reduced ejection fraction.

Log transformation was conducted before the group comparison of sTNFR1 and NT-proBNP levels.

P < .05

^aversus the low inflammation groups

^bversus all the other groups

^cversus women

More men were married or cohabiting (ϰ² = 27.5, P < .001). Men had higher comorbidity scores (F = 1.6, P = .04) and serum NT-proBNP (F = 4.5, P = .04) compared with women. A greater proportion of women were obese (ϰ² = 4.0, P = .05), in NYHA class III or IV (ϰ² = 9.0, P = .003), and prescribed antidepressant medication (ϰ² = 18.0, P < 0.001).

Men and women with high inflammation were older (F = 9.5, P < .001), more were in NYHA class III/IV (ϰ² = 22.2, P < .001), had higher comorbidity scores (F = 12.3, P < .001), NT-proBNP levels (F = 6.1, P < .001), and lower quality of life (F = 3.6, P = .01) than those with low inflammation. A greater proportion of men had HFrEF (ϰ² = 15.3, P = .002) and ischemic etiology of HF (ϰ² = 27.1, P < .001) than women regardless of inflammation status. There were 84 males (27.9%) and 27 females (19.7%) who had an event, including 10 deaths (9%) and 101 hospitalizations (81%) during a median follow-up of 194 days (25th, 75th quartile; 80 – 396 days). Women with low inflammation had a longer mean time to the first event compared to all other groups (F = 7.4, P < .001).

The differences in the association between inflammatory status and event-free survival between sexes

There was no difference in the risk for an event in women with high vs. low inflammation status controlling for age, ethnicity, BMI, LVEF, etiology of HF, NYHA class, comorbidity burden, and quality of life (Wald = .12, P = .73) (Table 2). In contrast, men with high inflammation had 1.85 times higher risk of an event compared with men with low inflammation (Wald = 5.12, P = .02) (Fig 1 and 2).

Table 2.

Cox proportional hazards regression model for time to composite HF or cardiac-related hospitalization or all-cause mortality events in men and in women

	Men			Women
	Adjusted hazard ratio	95% CI	p	Adjusted hazard ratio	95% CI	p
Age	1.05	.99 – 1.04	.21	.96	.92 – 1.00	.06
Ethnicity (minorities)	1.78	.93 – 3.42	.08	2.02	.85 – 4.80	.11
Body mass index	.97	.94 – 1.01	.16	1.00	.94 – 1.06	.95
Left ventricular ejection fraction	.99	.97 – 1.01	.38	.98	.96 – 1.01	.26
Etiology of heart failure (non-ischemic)	1.05	.60 – 1.83	.87	.42	.16 – 1.11	.08
New York Heart Association (III/IV)	1.14	.67 – 1.93	.63	.76	.30 – 1.91	.56
Comorbidity burden	.99	.86 – 1.15	.92	1.08	.83 – 1.41	.56
Quality of life	1.01	1.00 – 1.02	.03	1.00	.99 – 1.02	.73
High inflammation status	1.85	1.09 – 3.15	.02	1.189	.45 – 3.13	.73

Reference group: low inflammation status, Caucasian ethnicity, ischemic etiology of heart failure, New York Heart Association functional class I or II

Figure 1.

Cox regression for time to HF or cardiac-related hospitalization or all-cause mortality among men (ϰ² = 22.31, P < 0.01)

Figure 2.

Cox regression for time to HF or cardiac-related hospitalization or all-cause mortality among women (ϰ² = 23.48, P < 0.01)

Discussion

This study demonstrated sex differences in the role of inflammation in event-free survival among patients with HF. We found that higher inflammation was associated with shorter event-free survival among men but not women. Prior studies on sex differences in survival did not consider inflammatory status, though it has been recognized as an important pathologic process underlying HF symptoms and progression.^12,13,28,29 Our findings demonstrate that inflammation may be important to consider with examining to the differences between sexes in event-free survival.

Several sex-related differences in the etiology of and response to inflammation are proposed to explain our findings. First, myocardial contractile dysfunction has a major role in the development and progression of HF.³⁰ Various inflammation pathways have been demonstrated to affect cardiac function and may contribute to differences in outcomes between sexes.³¹ Inflammation in men was contributed to the development of HFrEF phenotype compared with HFpEF women.³¹ Several clinical studies on the survival differences between HFrEF versus HFpEF reported worse survival among the patients with HFrEF.^32,33

Second, various etiologies of HF have been demonstrated to trigger different inflammatory responses that may contribute to differences in outcomes between sexes.³¹ HF in men is mostly a consequence of cardiomyocyte damage (e.g., ischemic heart disease) compared with HF in women that is more commonly preceded by non-cardiac comorbidities such as hypertension, obesity, and diabetes mellitus.³¹ Cardiomyocyte damage causes inflammation that has been demonstrated to induce ventricular dilatation with reduced ejection fraction.^31,34 On the other hand, non-cardiac comorbidities are considered to induce low grade inflammation responsible for long-term cardiac remodeling, including cardiac hypertrophy and stiffness and preserved ejection fraction.^31,34 These etiologies-related differences in inflammatory response were used to suggest that inflammation in HF post cardiomyocyte damage might be an indicator of worsening cardiac function.³⁴ In clinical studies, acute and greater remodeling post cardiomyocytes injury was directly associated with inflammation, myocardial contractile dysfunction, and the risk of 1-year mortality.³⁴ HF post cardiomyocyte ischemic injury was associated with shorter event-free survival compared with HF related to non-cardiac etiology.^28,29,35

Third, observations from the animal models of HF and clinical studies have demonstrated the greater adverse response of male cardiomyocytes to TNF-α than those in females. Male animal cardiomyocytes were observed to have greater apoptotic and fibrotic rates, cellular remodeling response, and contractile dysfunction in response to TNF-α signaling compared with females.^36–40 These differences between sexes were suggested to provide women with a protective mechanism against the development of HF with reduced ejection fraction, even after myocardial infarction.⁴⁰ Consistent with previous observations, several clinical studies showed that men with HF were more likely to have faster ventricular remodeling in response to hemodynamic overload, significant remodeling changes including greater ventricular dilatation and hypertrophy, and lower preserved ventricular function post remodeling than women.^40,41

Consistent with previous studies, our data showed that TNF-α inflammatory status was highest among both sexes with the highest NT-proBNP levels and NYHA functional class, and lower quality of life.⁴² However, our data indicated that the severity of HF had no impact on event-free survival in women with high inflammatory status. Several observations may support this finding. First, male/higher inflammation group had a greater proportion of HFrEF with the higher prevalence of HFpEF in women/higher inflammation group. Second, high NT-proBNP level is considered to have a strong prognostic determinant for cardiac-related outcomes in HFrEF and non-cardiac-related outcomes in HFpEF.^43,44 This suggests that inflammation in men might be related to worsening cardiac function while in women might be related to the systemic inflammation associated with exacerbation of non-cardiac comorbidities and worse HF symptoms. Researchers consistently reported that cardiac causes of event-free survival had a higher incidence among men and non-cardiac causes of event-free survival had a higher incidence among women.⁴³ Thus, previous studies may provide an explanation for differences in inflammation-related event-free survival between men and women with HF.

Our findings should be interpreted with consideration of several limitations. We only had measures of sTNFR1 levels at baseline. Consequently, we were not able to determine whether levels changed over time. However, baseline sTNFR1 levels have been shown to be a strong long-term prognostic predictor of HF outcomes suggesting baseline sTNFR1 levels were a valid predictor of even-free survival.¹⁶ Increased sTNFR1 levels can also be a reflection of comorbidities which could affect event-free survival.^2,4 We controlled for comorbidities in our analyses to account for their effect, which indicates that it was inflammation related to HF that predicted event-free survival. Despite that HF is equally affects men and women; the representation of men in our study was greater than women. Regardless, the sample sizes were large enough to be able to draw meaningful clinical differences among groups. Although there was a difference in the representation of race between men and women, race did not predict event-free survival. This is likely due to a relatively small number of minorities in our sample. Therefore, caution should be used in extending these findings to other sub-populations of patients with HF.

Conclusion and Clinical Implications

Prior evidence suggested that inflammation is an important pathologic process underlying HF symptoms and progression. Although researchers consistently reported differences in event-free survival between men and women with HF, such differences related to inflammation were not considered. Our data provide evidence that inflammation may play a role in event-free survival differences between men and women with HF. Additional research is needed to confirm these results. In the meantime, clinicians should be aware that inflammation may have a greater prognostic impact in men compared with women. To date, no clinical trials have been published demonstrating that the usage of an anti-inflammatory therapy in the management of HF is beneficial.⁴⁵ However, our results suggest research on interventions to lower inflammation may be beneficial.