Sex-specific temporal evolution of circulating biomarkers in patients with chronic heart failure with reduced ejection fraction

BACKGROUND
We aimed to assess differences in clinical characteristics, prognosis, and the temporal evolution of circulating biomarkers in male and female patients with HFrEF.


METHODS
We included 250 patients (66 women) with chronic heart failure (CHF) between 2011 and 2013 and performed trimonthly blood sampling during a median follow-up of 2.2 years [median (IQR) of 8 (5-10) urine and 9 (5-10) plasma samples per patient]. After completion of follow-up we measured 8 biomarkers. The primary endpoint (PE) was the composite of cardiac death, cardiac transplantation, left ventricular assist device implantation, and hospitalization due to acute or worsened CHF. Joint models were used to determine whether there were differences in the temporal patterns of the biomarkers between men and women as the PE approached.


RESULTS
A total of 66 patients reached the PE of which 52 (78.8%) were male and 14 (21.2%) were female. The temporal patterns of all studied biomarkers were associated with the PE, and overall showed disadvantageous changes as the PE approached. For NT-proBNP, HsTnT, and CRP, women showed higher levels over the entire follow-up duration and concomitant numerically higher hazard ratios [NT-proBNP: women: HR(95%CI) 7.57 (3.17-21.93), men: HR(95%CI) 3.14 (2.09-4.79), p for interaction = 0.104, HsTnT: women: HR(95%CI) 6.38 (2.18-22.46), men: HR(95%CI) 4.91 (2.58-9.39), p for interaction = 0.704, CRP: women: HR(95%CI) 7.48 (3.43-19.53), men: HR(95%CI) 3.29 [2.27-5.44], p for interaction = 0.106). In contrast, temporal patterns of glomerular and tubular renal markers showed similar associations with the PE in men and women.


CONCLUSION
Although interaction terms are not statistically significant, the associations of temporal patterns of NT-proBNP, HsTnT, and CRP appear more outspoken in women than in men with HFrEF, whereas associations seem similar for temporal patterns of creatinine, eGFR, Cystatin C, KIM-1 and NAG. Larger studies are needed to confirm these potential sex differences.

and thus carry potential to unveil and explain differences in HF pathophysiology between men and women; even those that may not be clinically apparent. Herewith they may also contribute to explaining differences in survival between men and women with HFrEF. Serial biomarker measurements are especially interesting in this context, as they may expose potential differences in the temporal evolution of HF mechanisms in men and women.
With the current investigation, we aimed to assess differences in clinical characteristics, prognosis, and the temporal evolution of circulating biomarkers (of wall stress, myocardial injury, inflammation, and glomerular and tubular renal function) in male and female patients with HFrEF.

Study design
The design of the Bio-SHiFT study has been published in detail elsewhere [14,15]. In brief, the Bio-SHiFT study is a prospective, observational study consisting of patients with stable CHF carried out at the Erasmus MC, University Medical Center, Rotterdam, Netherlands, and Northwest Clinics, Alkmaar, Netherlands. From October 2011 to June 2013, during the first inclusion round of Bio-SHiFT, 263 patients were enrolled. The main inclusion criteria were age ≥ 18 years, capability of understanding and signing informed consent and diagnosis of CHF ≥3 months ago according to European Society of Cardiology guidelines [16]. A total of 250 (95%) patients had HFrEF, and they were available for the current analysis. At baseline, blood-and urine sampling was performed and study follow-up visits with repeated sampling were planned every 3 months (±1 month was allowed), until a maximum of 10 study follow-up visits were completed. A medical evaluation was performed with a short questionnaire and samples were obtained during each study follow-up visit.
The study was approved by the medical ethics committee of the Erasmus MC, and conducted according to the Declaration of Helsinki. Written informed consent was obtained in all participants. The study is registered at ClinicalTrials.gov (NCT01851538).

Baseline assessment
We collected information on CHF-related symptoms and New York Heart Association (NYHA) class, CHF etiology, cardiovascular risk factors, medical history and medical treatment. Also, systolic blood pressure, heart rate, body mass index and left ventricular ejection fraction (LVEF) as measured by echocardiography were determined at baseline.

Biomarker measurements
Plasma and urine samples were collected at baseline and at each follow-up visit, and were processed and stored at a temperature of −80 within 2 h after blood collection. When applicable, samples were transported to the central laboratory (Erasmus MC, Rotterdam, the Netherlands) under controlled conditions (at a temperature of −80°C) and stored until batch analysis was performed. Laboratory personnel was blinded for clinical data.
NT-proBNP, HsTNT, and CRP were measured in stored serum samples. Plasma NT-proBNP was analyzed using an electrochemiluminescence immunoassay (Elecsys 2010; Roche Diagnostics, Indianapolis, IN), which measures concentrations ranging from 5 to 35,000 ng/L. Cardiac troponin T was also measured using an electrochemiluminescence immunoassay (Elecsys 2010 immunoassay analyzer; Roche Diagnostics, Indianapolis, IN) which measures concentrations ranging from 3 to 10,000 ng/L. CRP was measured using an immunoturbidimetric assay (Roche Hitachi 912 chemistry analyzer; Roche, Basel, Switzerland). Measurement of renal biomarkers in plasma and urine samples was performed at HaemoScan BV (Groningen, Netherlands).
GFR was determined by the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation that has been validated in HF patients [17]. Patients were categorized using National Kidney Foundation-Kidney Disease Outcome Quality Initiative (K/DOQI) clinical practice guidelines [18].

Follow-up and adverse events
Because of the batch laboratory analysis after completion of followup, results of the biomarker assays were not available to treating physicians at the time of the outpatient visits. As such, this was an observational study, all patients were treated according to prevailing ESC guidelines and the biomarker measurements performed for this study did not alter usual patient care.
Patients were followed at the outpatient clinic every 3 months with a maximum of 10 follow-up visits per patient. The follow-up visit consisted of a short medical evaluation, during which all medication changes and adverse cardiovascular events since the previous visit were reported and blood and urine samples were collected. The primary endpoint (PE) was the composite of cardiac death, cardiac transplantation, left ventricular assist device implantation, and hospitalization due to acute or worsened CHF, whichever occurred first during follow-up.

Statistical analysis
Normality of continuous variables, including biomarker concentrations, was assessed by the Kolmogorov-Smirnov test. Normallydistributed continuous variables were reported as means and standard deviations, non-normally distributed continuous variables as medians and interquartile ranges (IQRs). In case of skewed distributions, continuous variables were logarithmically transformed (log base 2) for further analysis. All biomarker values were transformed in this way. Categorical variables were reported as numbers and percentages.
Univariate cox regression was performed to investigate whether there was an association between baseline characteristics and biomarker levels and the PE. Unadjusted hazard ratios (HRs) were reported with 95% confidence intervals (CIs) per two fold increase in biomarker level. We stratified for sex. Interaction terms between sex and the biomarkers of interest were added to the models to determine potential differences between men and women.
We used linear mixed effect models to describe the average temporal pattern of each biomarker for men and women with and without a PE during study follow-up. To estimate the associations between repeated biomarker measurements and the hazard of PE, we applied joint modeling (JM) analyses, which combine linear mixed effect models for temporal evolution of the repeated measurements with relative risk models for the time-to-event data. By using the JM technique, analyses inherently accounted for different follow-up durations between patients [19]. For the joint models, unadjusted HRs and corresponding CIs, per doubling of the biomarker level at a given follow-up time, were reported.
Analyses were performed in SPSS and R package JM Bayes [20]. All statistical tests were two-tailed. P-values <0.05 were considered statistically significant.
At baseline, HsTnT level was higher in men than in women [median (IQR): 19.    were associated with the PE, while in women these associations did not reach statistical significance. However, interaction terms with sex were only significant for PCI (p for interaction = 0.018) and valvular heart disease (p for interaction = 0.031). (Appendix Table 1).

Associations between NT-proBNP, HsTnT, CRP, and the PE
Baseline NT-proBNP was positively associated with the PE in both sexes. Differences in temporal patterns between patients with and without the PE were similar for men and women, with rising levels as the PE approached in both sexes (Fig. 1A). However, NT-proBNP levels appeared higher over the full time-course in women experiencing the PE than in men experiencing the PE. This was reflected by the HRs entailed by the serially measured NTproBNP levels, which were numerically but not significantly higher in women [HR (95%CI): women; 7.57 (3.17-21.93), men; 3.14 (2.09-4.79), p for interaction = 0.104] ( Table 2).
For HsTnT, the association between baseline levels and the PE was stronger for women than for men [HR (95%CI) 1.26 (1.11-1.42), and [HR (95%CI) 1.09 (1.03-1.15), respectively; p for interaction = 0.038)]. Likewise, although levels rose in both sexes as the PE grew near, the difference in level at any moment in time between those with and those without the PE appeared larger in women than in men, although this difference was not statistically significant (Fig. 1B) Fig. 1C). Women showed a greater incline in CRP than men as they approached the PE. The temporal evolution of CRP level was associated with the PE in both sexes, but relative risk was numerically higher in women [HR (95%CI): women; 7.48 (3.43-19.53), men; 3.29 (2.27-5.44); p for interaction = 0.106] ( Table 2).  Fig. 1D).

Associations between glomerular and tubular renal markers and the PE
The association between baseline levels of eGFR was similar in men and women. (Table 2, Fig. 1E) The temporal evolution of eGFR was associated with the PE in both sexes, and did not show relevant sex differences [HR (95%CI): women; 1.19 (1.00-1.48), men; 1.14 (1.02-1.26); p for interaction = 0.679]. Levels of Cystatin C at baseline were related to the PE in both sexes (  Table 2).
With regard to NAG, baseline levels were associated with the PE without differences in the associations between men and women (

Discussion
In this cohort of 250 patients with HFrEF with a median follow-up time of 2.2 (IQR:1.4-2.5) years, the cumulative incidence of the PE was 28.3% in men and 21.2% in women although age at baseline was similar, confirming better prognosis in women with HF. Baseline levels of HsTnT and CRP were significantly stronger associated with the PE in women than in men. The temporal patterns of all studied biomarkers were associated with the PE, and overall showed disadvantageous changes as the PE approached. For NT-proBNP, HsTnT, and CRP, changes in temporal patterns appeared more outspoken in women than in men as the PE grew near, with higher levels over the entire follow-up duration and concomitant numerically higher hazard ratios. However, these differences did not reach statistical significance. Temporal patterns of glomerular and tubular renal markers showed similar associations with the PE in men and women.
Differences between men and women in the epidemiology of HF are most apparent when the type of HF is considered [1]. In general, women are more often affected by HFpEF, and men by HFrEF. Nevertheless, over 20% of outpatients with HFrEF also consists of women [10]. To date, only one study has examined sex difference in clinical characteristics and in survival in HFrEF patients. They observed that women had higher prevalence of risk factors including obesity, higher systolic blood pressure, and higher heart rate, but were less likely than men to have comorbidities, except for hypertension. Women had lower mortality than men [HR(95%CI): women; 0.68 (0.62-0.74), p < 0.001], but more symptoms and worse quality of life [11]. Moreover, few studies are available on sex differences in circulating biomarkers in heart failure patients, and in particular in HFrEF patients. Meyer et al. examined HF patients with the full continuum of LVEF included in the Coordinating study evaluating Outcomes of Advising and Counseling in Heart failure (COACH). They determined biomarkers in 567 patients (38% women) shortly before discharge following a heart failure hospitalization [21]. Several biomarkers reflecting inflammation, remodeling and renal function were significantly lower in women, suggesting differences in biological disease expression, etiology and influence of comorbidities. Also, while NTpro-BNP did not differ between men and women at baseline, it showed different predictive value in women and men for 3-year all-cause mortality. This was also observed for several other biomarkers, albeit less explicitly.
Our study extends these findings by examining differences in temporal evolution of circulating biomarkers between men and women with HF, and moreover it is the first to focus specifically on HFrEF. While median baseline NT-proBNP level was similar in men and women, rise in level was more outspoken in women as the PE approached. Differences in sex hormones may contribute to these findings. Previous research has shown that women with hormone replacement therapy (HRT) have higher NT-proBNP levels than women without HRT, suggesting a stimulating effect of estrogen on natriuretic peptides that may in part also explain sex differences [22]. It should be noted however that given the relatively high age of our study population and the limited use of hormone therapy in the Netherlands, other mechanisms are probably also at play here.
The association of Hs-TnT with clinical outcome was also stronger in women than in men in our study, although men had higher baseline Hs-TnT levels than women. The latter has recently also been shown by Lew et al. [23]. Sex-related differences in body composition [24,25], differences in anatomy leading to variation of the cardiac mass and coronary artery size [26] and cardiac remodeling related to sex hormones have been suggested to underlie these differences in troponin levels [27].
How above-described mechanisms could also lead to more explicit temporal biomarker changes in women remains to be elucidated. In this context, previous research has demonstrated that estrogen receptor activation leads to reduction in necrotic and/or apoptotic cell death in animal models. Thus, apoptosis rate may be higher in men, which could potentially result in more unfavorable cardiac remodeling directly after ischemic injury [28,29]. Such differences in the remodeling process may in part explain why biomarkers associated with cardiac remodeling including NT-proBNP and hsTnT show different patterns in men and women experiencing adverse clinical events.
In our study, the association of CRP with clinical outcome was stronger in women than in men as well. Studies in the general population point towards higher CRP levels in women compared to men [30]. As there is a strong correlation between CRP levels and subcutaneous fat, these higher levels could in part be explained by women having more subcutaneous, but not visceral, fat compared to men [31,32]. Further to these differences, an unfavorable temporal evolution of inflammatory status as signified by CRP, might have stronger consequences for clinical outcome in women than in men, in line with the known stronger impact on HF evolution carried by factors such as diabetes mellitus in women [33].
For clinical settings, the above findings imply that increasing levels of NT-proBNP, hsTnT and CRP over time may carry stronger predictive value for major adverse cardiovascular events in women compared to men with HFrEF. Therefore, levels of these biomarkers might need to be evaluated according to sex during clinical follow-up, and any consequences for estimation of prognosis based on changes in these biomarkers should also be evaluated according to sex. However, before any specific clinical recommendations can be made about subsequent timing or adaptation of therapy based on changes in biomarkers, trials on multiple-biomarker guided therapy which focus on both men and women are warranted.
In contrast with the sex differences we found for aforementioned markers, we found no differences between men and women in levels of glomerular and tubular markers, nor in the associations of renal markers with clinical outcome. An exception was baseline eGFR, which was lower in women than in men. It has previously been demonstrated that in general eGFR is lower in women than men with HF [21]. The hypothesis has been raised that men have a higher renal functional reserve to compensate for the loss of glomeruli during aging [34,35]. This suggests that eGFR is a less sensitive marker for HF in men than women. Recently, a study was published showing that Cystatin C is a gender-neutral glomerular rate biomarker that is preferred to eGFR [36]. In order to prevent underestimation of renal dysfunction in men with HF, use of Cystatin C as a glomerular rate biomarker should be considered as well [36]. Previous studies on sex differences in tubular renal markers in HF patients are scarce. Some aspects of our study warrant consideration. Firstly, our study was not originally designed to examine sex differences, and consequently statistical power for this investigation was limited. Studies with larger sample size are needed to confirm our results. Furthermore, only limited molecular biological research is currently available on the effect of sex differences on proteins involved in wall stress, myocyte injury, inflammation and renal function. Such knowledge however is necessary to fully understand the biological processes underlying clinical epidemiological sex differences observed in these biomarkers.
In conclusion, this study demonstrates for the first time that the associations of temporal patterns of NT-proBNP, HsTnT, and CRP appear more outspoken in women than in men with HFrEF, whereas associations are similar for temporal patterns of eGFR, Cystatin C, NAG and KIM-1. Additional studies with larger sample sizes are warranted in order to confirm these findings, while molecular biology studies are needed to further unravel underlying mechanisms.

Author agreement form
This statement is to certify that all authors have seen and approved the manuscript.
being submitted, have contributed significantly to the work, attest to the validity and legitimacy of the data and its interpretation, and agree to its submission to the International Journal of Cardiology.
We attest that the article is the Authors' original work, has not received prior publication and is not under consideration for publication elsewhere. We adhere to the statement of ethical publishing as appears in the International of Cardiology (citable as: Shewan  On behalf of all Co-Authors, the corresponding Author shall bear full responsibility for the submission. Any changes to the list of authors, including changes in order, additions or removals will require the submission of a new author agreement form approved and signed by all the original and added submitting authors.
All authors are requested to disclose any actual or potential conflict of interest including any financial, personal or other relationships with other people or organizations within three years of beginning the submitted work that could inappropriately influence, or be perceived to influence, their work. If there are no conflicts of interest, the COI should read: "The authors report no relationships that could be construed as a conflict of interest".

Declaration of competing interest
All authors declare no conflict of interest. ⁎ Composite of cardiac death, cardiac transplantation, left ventricular assist device implantation, and hospitalization due to acute or worsened chronic heart failure ⁎⁎ Some could not be tested because of paucity of events