Determinants of Elevated D-dimer Levels and Long-Term Outcome in Elderly Patients with End-Stage Heart Failure

Background: Previous studies have shown that heart failure is associated with hemostatic abnormalities and hypercoagulable state. Plasma D-dimer levels reect both brin formation and degradation, and elevated D-dimer levels have been associated with poor prognosis in patients with heart failure. However, little is known about their roles in elderly patients with end-stage HF. In present study, we aimed to explore the clinical signicance and determinants of plasma D-dimer in elderly patients with end-stage heart failure. Methods: A total of 177 patients with heart failure at Beijing Geriatric Hospital from November 1, 2015 to December 30, 2018 were enrolled. All hospitalized patients were obtained D-dimer levels within the rst 24 h following admission after obtaining informed consent. Primary endpoint was all-cause mortality. Results: A total of 60 patients had elevated D-dimer levels. Blood urea nitrogen (β=1.106, 95% CI: 1.029-1.190, p = 0.006), NYHA functional class (β=2.179, 95% CI: 1.170-4.056, p = 0.014) and white blood cell counts (β = 1.188, 95% CI: 1.040-1.358, p = 0.011) were independent risk factors for elevated D-dimer in elderly patients with end-stage heart failure. Albumin (β=0.803, 95% CI: 0.728-0.885, P (cid:0) 0.001) was negative risk factor for elevated D-dimer in elderly patients with end-stage heart failure. Elevated D-dimer level was independently associated with increased risk of long-term all-cause mortality (P = 0.048). Conclusions: For elderly patients with end-stage heart failure, D-dimer levels were associated with white blood cell counts, blood urea nitrogen, albumin and NYHA functional class and elevated D-dimer level was independently associated with poor long-term outcome. N-terminal pro-B-type natriuretic peptide; LVEF, Left ventricular ejection fraction (cid:0) LVEDD, Left ventricular end-diastolic diameter; NYHA, New York Heart Association; LAD, Left atrial diameter; PO2, Partrial pressure of blood oxygen; CA-125, Carbohydrate antigen-125; TCHO, Total cholesterol; LDL, Low density lipoprotein; HDL, High density lipoprotein; ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker.

Methods: A total of 177 patients with heart failure at Beijing Geriatric Hospital from November 1, 2015 to December 30, 2018 were enrolled. All hospitalized patients were obtained D-dimer levels within the rst 24 h following admission after obtaining informed consent. Primary endpoint was all-cause mortality.
Conclusions: For elderly patients with end-stage heart failure, D-dimer levels were associated with white blood cell counts, blood urea nitrogen, albumin and NYHA functional class and elevated D-dimer level was independently associated with poor long-term outcome.

Background
Epidemiologic and clinical data from the last 2 decades have led to the recognition that heart failure (HF) is the major contributor to cardiovascular morbidity and mortality, resulting in reduced longevity [1]. In particular, HF now has become one of the most common reasons for hospitalizations in people over 65 years [2] because of increased survival after improved treatment of coronary artery disease, valvular disease, and hypertension [3].
Patients with heart failure often die of pump failure late in life, nevertheless the causes of death are not studied in detail. Several studies have suggested that thrombosis may serve as signi cant contributors of poor prognosis and mortality in patients with chronic heart failure (CHF) [4]. As a biomarker of brin formation and degradation, D-dimer is related to the pathology of the cardiovascular system [5] and is a useful parameter for clinical evaluation of the degree of hypercoagulability and thrombotic disease [6]. Ddimer levels were found to be signi cantly higher in  year old HF patients than in healthy patients [7]. Furthermore, in a study with a mean follow up of 8 ± 5 months (up to 34 months), elevated D-dimer levels were found to predict cardiovascular mortality and the development of atrial brillation (AF) in hospitalized patients with end-stage heart failure [6]. However, there are scarce data concerning the D-dimer levels in elderly patients with end-stage heart failure with a long follow up (up to 2 years), which account for the majority form of HF. Therefore, in this study, we aimed to determine the clinical signi cance and determinants of plasma Ddimer in elderly patients with end-stage HF. Furthermore, we sought evidence to evaluate the value of the D-dimer baseline level to predict subsequent clinical outcome in patients with end-stage HF for an average of 21 months follow up.

Study population
This was a retrospective observational follow-up study. The patients with chronic heart failure (CHF) were included in the cardiovascular medicine unit at Beijing Geriatric Hospital (Beijing, China) from November 1, 2015 to December 30, 2018. The patients were either admitted for the rst time or were repeat hospitalization for CHF. The diagnosis of heart failure was according to the Guideline for the diagnosis and treatment of heart failure in China 2018. We de ned end-stage HF as New York Heart Association class III-IV, despite of optimal medical therapy, such as beta blockers, angiotensin converting enzyme inhibitors (ACEIs). The causes of CHF included ischemic heart disease, valvular heart disease, hypertension, various cardiomyopathy et al. The patients were excluded if they had presented with HF due to acute coronary syndromes or accelerated hypertension or were in cardiogenic shock or suffered from any serious illness. Additionally, subjects who had a history of malignancy, pregnancy, aortic dissection, trauma or lie in bed were also excluded. Patients with deep vein thrombosis (DVT) and/or pulmonary thromboembolism (PTE) were also excluded from the study. Totally, 177 patients were included in the study. Data including the demographic, clinical, laboratory ndings and medications were collected through electronic medical record system. We conducted telephonic interviews with the participants with respect to their survival (see supplementary interview items).

Laboratory Evaluation
For measurement, blood samples were collected in fasting or nonfasting state within 24 hours of hospitalization and measured within 4 hours as routine sample analysis in our hospital laboratory. The concentration of each biomarker was assayed using commercially available assay kits according to manufacturer protocol and using standard curves and software. Plasma D-dimer was analyzed with immunoassays turbidimetry (Instrumantation Labroatory Company, Spain) as recommended by the manufacturer. The intra-assay coe cient of variation was 8.3% and the inter-assay coe cient of variation was 11.0%, respectively. A plasma D-dimer level would be reexamined if it was not consistent with clinical manifestations. The new results of D-dimer examination were used for analysis. The concetrations of biochemical biomarkers included blood urea nitrogen, serum creatinine, total protein, prealbumin, serum albumin and β2-microglobulin were measured by using a lactate dehydrogenase assay kit (Siemens Healthcare Diagnostics Inc, USA). The blood gas analysis was performed only once using the blood gas analyzer with the arterial blood immediately drawn within 2 hours of admission. PO2,

Baseline Data
Lac and SO2 were measured. Carbohydrate antigen-125 (CA-125) was obtained using CA125 assay kit (Siemens Healthcare Diagnostics Inc, USA). The FT3 and TT3 were quantitatively determined by free thyroxine assay kit with direct chemiluminescence (Siemens Healthcare Diagnostics Inc, USA).

Echocardiography
Transthoracic echocardiography was performed using an iE33 Color Doppler Ultrasound System (Philips Healthcare, EPIQ7, USA) in ultrasonic department in accordance with American Society of Echocardiography (ASE) guidelines, by dedicated analysts blinded to clinical information. LVEF were derived according to the modi ed biplane Simpson's rule.

Follow up
We followed up the patients by telephone interview at 6, 12, and 24 months after discharge and the median follow-up period was 21 months (interquartile range of 11 to 30 months). We collected data on all-cause death from the hospital medical records and telephone follow up. All-cause mortality was de ned as death from any cause during the follow up period.

Statistical analysis
We used a threshold of 0.5 mg/L to de ne a positive plasma D-dimer result because this value is widely used in clinic, an age-adjusted D-dimer cut-off model has recently decreased false-positive results without additional false-negative ndings in patients and therefore we applied this cut-off to our study. D-dimer values were compared to the current cut-off of 0.5 mg/L and to the suggested age-adjusted cut-off (age/100 mg/L) [8,9]. Before analysis, normal distribution was tested for continuous variables. Value of D-dimer was logarithm transformed to normalize its distribution. Continuous variables are expressed as mean ± SD, and categorical variables are expressed as percentages. Comparisons between two groups of continuous variables were performed using Student's t-test or Mann-Whitney U test. A chi-squared test was used to compare the difference between categorical variables. We used Spearman's correlation test or Pearson's correlation test to examine correlations between two continuous variables (as appropriate).
Stepwise multiple linear regression analysis was performed to identify independent variables that might determine plasma D-dimer levels (p value threshold to enter ≤ 0.05, ≥ 0.1 to remove). Forward LR method of multivariate logistic regression analysis was used to detect independent predictors of the elevated Ddimer levels. For multiple linear regression analysis and multivariate logistic regression analysis, we included variables based on the results of univariate analysis and professional knowledge. Event-free survival curves were constructed using the Kaplan-Meier methods and compared using log-rank test. A P value < 0.05 was considered statistically signi cant. All analyses were performed with the statistical package SPSS 21.0 (SPSS Inc, Chicago, Illinois). A 2-tailed p value < 0.05 was considered statistically signi cant.
The median plasma D-dimer levels was 0.95 mg/L (0.03-31.97). The frequency distribution of D-dimer levels was showed in Fig. 1A, and normal distribution of D-dimer was obtained from logarithmic transformation in Fig. 1B. Totally, one hundred seventy-seven patients were included in the present study.
The baseline demographic, clinical characteristics, laboratory ndings and drug treatments were summarized according to the cut off value of plasma D-dimer levels in Table 1A and Table 1B. The mean age of recruited patients was 73.7 ± 11.7 years (range 28 to 96), and 91 (51.4%) of the patients were older than 75 years, 83 (46.9%) were female. Almost all of them (99.4%) had dyspnea symptoms, the majority of them (91%) had severe dyspnea (New York Heart Association functional class NYHA III/IV). One The incidence elevated D-dimer levels in heart failure D-dimer values were compared to the current cut-off of 0.5 mg/L showed in Table 1A. A total of 60 (33.9%) patients had elevated D-dimer levels. The concentrations of white blood cell (p < 0.017), carbohydrate antigen-125 (p < 0.009), β2-microglobulin (p < 0.043) and blood urea nitrogen (p < 0.004) were signi cantly higher in patients with elevated D-dimer compared with those with normal D-dimer.
However, the inverse relationship of the D-dimer levels with the anemia indicators including red blood cell (p < 0.021), hemoglobin (p < 0.015) and hematocrit (p < 0.020) were found. Correspondingly, patients with elevated D-dimer had a tendency of hypoproteinemia, such as lower total protein (p = 0.009), albumin (p < 0.001) and prealbumin (p < 0.001). In addition, the elevated group had a lower serum sodium levels than that in normal group (p = 0.026). Importantly, patients with the higher levels of D-dimer were accompanied by high New York Heart Association functional class (III/IV) (p = 0.001) though the left ventricular ejection fraction (p = 0.830) and N-terminal probrain natriuretic peptide (NT-proBNP) (p = 0.437) did not differ between the two groups. Interestingly, there were less patients with spironoactone therapy in elevated D-dimer levels (p = 0.050). In addition, the percentage of patients with smoking in the elevated D-dimer group had a rising trend than that in normal group (p = 0.056). Interestingly, the results of the age-adjusted cut-off D-dimer is consistent with the current cut-off of 0.5 mg/L basically as shown in Table 1B. A total of 45 (25.4%) patients had elevated D-dimer levels. Additionally, elevated D-dimer group had a lower high density lipoprotein (p = 0.031).
Plasma D-dimer values according to clinical characteristics of the patients with heart failure were showed in Table 2. Higher D-dimer levels were found in patients more than seventy-ve years old compared those less than seventy-ve years old (p = 0.018). Moreover, plasma D-dimer levels were signi cantly higher with respect to severe heart failure symptoms, as judged by NYHA functional class (p < 0.001). Correspondingly, the presence of course of disease over ve years were associated with increasing plasma D-dimer values (p = 0.014). In addition, there were a Table 1 a. Baseline characteristics of 177 heart failure patients with and without elevated D-dimer (≥ 0.5 mg/L).

Variable
Total Cohort (n = 177) Data are expressed as mean ± SD, number (percentage), or median (interquartile range). Data are expressed as mean ± SD, number (percentage), or median (interquartile range). Abbreviations as in Table 1.
Data are expressed as median (interquartile range).
trend toward higher plasma D-dimer levels in smokers (p = 0.085).

Clinical Outcomes
The patients of the present study was followed up for a median of 21 months (ranged from 11 to 30 months), 12 patients were lost to follow up. During the follow up, all-cause death occurred in 78 patients (44.1%). Figure 3 showed the Kaplan-Meier curves for cumulative survival strati ed by admission D-dimer level. It was found that patients with elevated D-dimer levels had signi cantly higher cumulative mortality than patients with normal D-dimer levels (log rank P = 0.048).

Discussion
In recent years, advanced treatment have reduced the mortality and improved the quality of life for patients with heart failure. However, altered hemostasis and abnormalities in endothelial function are thought to be other possible mechanisms that contribute to disease progression, ischemic events, and intracardiac thrombosis in heart failure [10]. HF is associated with an increased risk of venous thromboembolism (VTE) as a result of vascular abnormalities, increased coagulability, and impaired blood ow [11]. VTE includes two main clinical manifestations, deep vein thrombosis (DVT) and pulmonary embolism (PE), leading to a higher mortality in heart failure [12]. Patients with heart failure have a higher risk of VTE than those without heart failure, and the risk increases with the severity of heart failure. D-dimer is a product of the degradation of cross-linked brin and its plasma levels re ect both brin formation and degradation, and hence, D-dimer is elevated when there is thrombosis in the whole circulatory system. D-dimer is considered a useful biomarker worthy of continued attention because of its potential to identify patients with hypercoagulability [13]. Alexandre and Theodore E. et al. who evaluated 2593 hospitalized HF patients, observed an positive association between HF severity and both VTE risk and D-dimer concentration [11]. Marcucci et al. found that D-dimer levels increasd signi cantly in acute decompensated HF patients [14]. In addition, In the study of Ali Zorlu, it was shown that D-dimer was predictive of mortality in patients with systolic HF [4]. In consistent with the previous studies, our data suggested that plasma D-dimer levels were elevated in patients with HF. Importantly, patients with the higher levels of D-dimer were accompanied by high New York Heart Association (NYHA) founctional calss. Furthermore, our study showed that a elevated D-dimer level associated with an increased risk of all-cause death in elderly patients with CHF during 2 years follow up (Fig. 3).
Generally, the patients with HF were often accompanied by a disorder of protein metabolism. Albumin as the most represented plasma protein plays a pivatal role in the maintenance of homeostasis and physiological functions [15]. Serum albumin as a nutritional marker is related to cardiovascular mortality as well as important survival determinants and indicators for hemodialysis patients [16]. Hypoalbuminemia is common in elderly HF patients with a prevalence of 25-42% [17]. Previous studies also showed that hypoalbuminemia was associated with cardiovascular disease in the general population and adverse outcome in patients with HF independent of natriuretic peptides [18]. Arnaud et al studied 546 patients with HF, and found that 64.8% patients with a serum albumin lower 38.7 g/L and hypoalbuminemia (< 34 g/L) yielded the best sensitivity (78.8%) and speci city (75%) for predicting hospital death [19]. A retrospective studies in a real-world cohort of patients with HF performed by Israel et al demonstrated that decreasing albumin levels were directly associated with reduced survival and serum albumin is a signi cant predictor of a worse outcome in these patients [20]. Another study done in 177 patients with acute decompensated heart failure (ADHF) performed by Megan M et al showed that baseline serum albumin concentration of 30 g/L or less had a value for predicting worsening renal function in patients with ADHF [21]. The nding is consistent with our belief, our data suggested that patients with HF had a less serum albumin level compared to the threshold, 36.44 ± 4.60 g/L and 40 g/L, respectively. And about one hundred thirty-three patients (75%) had a serum albumin value less than 40 g/L. Additionally, our results showed that the serum albumin level were negatively signi cant associated with both elevated D-dimer and plasma D-dimer levels on univariate analysis. Furthermore, multivariate analysis showed that the serum albumin level was independently related to elevated D-dimer levels.
Several mechanisms have been proposed to explain the association between hypoalbuminemia and plasma D-dimer, including hypoalbuminemia-induced hypercoagulability and thrombosis [22,23]. Data from Bang et al and Iioka Y et al showed negative correlations between serum albumin level and plasma D-dimer level in patients with nephrotic syndrome [24,25]. A similar result was obtained by Remuzziet al [26]. Combined with the negative correlations between plasma D-dimer and serum albumin found in this study, which have also been indicated by Soon Bae Kim et al [27], we speculated that elevated plasma Ddimer in HF may be partially related to hypoalbuminemia-induced hypercoagulability.
Prealbumin (PAB) is another indicator of nutritional status, similar to albumin, also known as transthyretin. Compared with albumin, PAB has the advantages of shorter half-life and smaller serum pool. Its main functions are to bind and transport endogenous proteins and small molecules. Thus, PAB is more sensitive to changes in acute protein status [28]. Previous studies have shown that prealbumin is most commonly used as an indicator of malnutrition, and hypoalbuminemia has an adverse effect on disease outcomes because of increase in incidence of infection, length of hospital stay, and risk of death [29]. Andreia et al found that hypoprealbuminemia are more susceptible to early posttransplant thrombotic complications, particularly hepatic artery thrombosis which correlates with high morbidity, in patients with hereditary transthyretin amyloidosis (ATTR) [30]. Patrícia and his colleagues showed that prealbumin is an independent predictor of in-hospital mortality in HF patients [31]. Furthermore, they suggested that prealbumin levels below a cut-off < 15 mg/dL were found to be associated with reduced survival and increased rates of all-cause and HF readmission at 6-month follow up [32]. In consistent with the previous studies, our nding using multivariate linear regression analysis showed that the serum prealbumin level was inversely related to elevated plasma D-dimer levels after adjustment for multiple variables including age, gender, HF risk factors, and medicine treatment. More interestingly, our data showed that elevated D-dimer level was with a lower serum prealbumin level about 15.3 mg/dL. Blood urea nitrogen (BUN) as a nitrogenous end product of protein metabolism has been observed to be associated with mortality in various diseases [33]. A large body of evidence have demonstrated that elevated BUN levels are associated with adverse outcomes in patients with HF, especially in acute decompensation state [34,35]. The ACTIV study showed highest BUN quartile had the highest 60-day mortality, more importantly, BUN was a predictor of the composite end point of mortality and 60-day postdischarge death or hospitalization for chronic heart failure [36]. Fonarow et al identi ed BUN level of 43 mg/dL on admission as the best single indicator between hospital survivors and nonsurvivors by evaluating the Acute Decompensated Heart Failure National Registry (ADHERE) database. Of note, they suggested that BUN may be a stronger predictor of outcomes [37]. In a total of 541 patients with decompensated HF, Aronson et al showed BUN was also a predictor of adverse outcomes after discharge, comparing with the impact of serum creatinine. Moreover, BUN had at least equal role in determining prognosis and predicting mortality compared with N-terminal brain precursor natriuretic peptide (NT-proBNP) [38]. In addition, It has been shown that in patients with stable CHF, elevated BUN levels are still strongly associated with short and long-term mortality [39]. The precise mechanisms underlying the relationship between BUN and HF are that in patients with HF, low cardiac output decreases renal blood ow and GFR which leading to an increase of urea and thus BUN. Meanwhile, BUN may also be affected by intestinal function, nutritional status such as protein intake or systemic catabolism, and neurohumoral factors [40]. As expected, our data suggested that patients with heart failure had a higher BUN levels than that in the general population, of note, the positive relationship between BUN and D-dimer remained signi cant even after adjustment for potential confounding factors, in multivariate linear regression analysis. Although we were unable to elucidate the de nite causes for this correlation, in ammation might play a role in this regard. It was proposed that BUN increased levels of in ammatory markers and mediators which leading to hypercoagulability and embolism in alcoholic hepatitis patients [41].
As we known, D-dimers are brin degradation products released into the bloodstream after blood clot brinolysis that have classically been used for the evaluation of venous thromboembolism. However, Ddimers are also serum acute-phase proteins (APP) that show upregulated expression after stress, infection, or worsening disease states [42]. Increasing evidences have showed that in ammatory response played an important role in the development of DVT. Acute DVT causes a systemic in ammatory response characterized by elevated in ammatory factors and temporal accumulation of in ammatory cells including white blood cells [43,44]. Dawei Liu et al studied 1179 patients of tibial plateau fractures, and found that platelet and neutrophil count were independently associated with elevated D-dimer levels [45]. Moreover, data from 84 patients with lung cancer showed the accumulation of massive neutrophils in the in ammation sites leads to embolism and thrombosis which resulting poor prognosis [46]. One reason D-dimers may be predictive of the outcome is because they serve as a biomarker of in ammation. In consistent with this studies, our data suggested that plasma D-dimer levels were signi cantly positively associated with white blood cell counts, furthermore, white blood cell counts was an independent predictors for a high D-dimer level.
Our study has several limitations, worthwhile mentioning. Firstly, it was a single centre and observational study with a limited number of patients, and our results may not be extended to patients in other regions due to the heterogeneity in individual. Secondly,we only measured D-dimer levels at admission, and it was therefore not possible to assess changes in D-dimer levels over time and to evaluate the implications of these changes on CHF outcomes. Thirdly, the follow up time was only 2 years, and the more long-term effect of D-dimer on prognosis still needs to be further studied. We're just looking at all causes of death and the exact cause of death was not analyzed. Last but not least, due to the observational study design, the effect of unmeasured confounders may have impacted the results, although we performed statistical adjustments. Despite these limitations, our results still have a potential clinical signi cance and merit further researches, particularly for elderly patients with end-stage heart failure.

Conclusions
In the present study, we found numbers of elderly patients with end-stage CHF had a signi cant increase in D-dimer levels. Blood urea nitrogen, NYHA functional class and white blood cell counts were independent risk factors for elevated D-dimer, whereas, albumin and pre albumin were negative risk factor for elevated D-dimer. Moreover, elevated D-dimer level was independently associated with poor long-term outcome in elderly patients with end-stage heart failure. However, more long-term effect of D-dimer on prognosis needed to be further studied.

Ethics approval and consent to participate
The study was approved by the Institutional Review Board (IRB) of Beijing Geriatric Hospital. Written informed consent was obtained from the patients for their anonymized information to be published in this article.

Consent for publication
Not applicable.

Availability of data and materials
Datasets used during the current study are included in supplementary information les.

Competing interests
The authors declare no con icts of interest.

Fundings
This study was supported by Beijing Hospitals Authority Clinical Medicine Development of Special Funding Support (ZYLX202131) and Geriatric research project of Beijing geriatric hospital (2020bjlnyyfocus-2).

Authors' contributions
Aiju Tian conceived, designed and drafted the project; Shengfeng Weng provided administrative, technical, or material support; Xiaoli Chen and Hong Liu collected the data; Chengzhi Yang conducted the data analysis and suggested revisions of the paper. Zhi Luo conceived and designed the project. All the authors have read and approved the nal manuscript.
(A) Frequency distribution of plasma D-dimer levels in patients with heart failure. (B) Frequency distribution of plasma D-dimer levels in patients with heart failure after logarithmic transformation.