Inflammation-based assessment for the risk stratification of mortality in patients with heart failure

The Glasgow Prognostic Score (GPS) has been established as a useful resource to evaluate inflammation and malnutrition and predict prognosis in several cancers. However, its prognostic significance in patients with heart failure (HF) is not well established. To investigate the association between the GPS and mortality in patients with HF, we assessed 870 patients who were 20 years old and more and had been admitted for acute decompensated HF. The GPS ranged from 0 to 2 points as previously reported. Over the 18-month follow-up (follow-up rate, 83.9%), 143 patients died. Increasing GPS was associated with higher HF severity assessed by New York Heart Association functional class and B-type natriuretic peptide (BNP) levels. Kaplan–Meier analysis showed significant associations for mortality and increased GPS. In multivariate analysis, compared to the GPS 0 group, the GPS 2 group was associated with high mortality (hazard ratio 2.92, 95% confidence interval 1.77–4.81, p < 0.001) after adjustment for age, sex, blood pressure, HF history, HF severity, hemoglobin, renal function, sodium, BNP, left ventricular ejection fraction, and anti-HF medications. In conclusion, high GPS was significantly associated with worse prognosis in patients with HF. Inflammation-based assessment by the GPS may enable simple evaluation of HF severity and prognosis.


Discussion
This study demonstrated the prognostic significance of the GPS for predicting mortality in patients with HF. Specifically, it showed that: (1) patients with higher GPS had significantly higher severity of HF assessed by NYHA functional class and BNP levels than did patients with lower GPS, (2) age, BNP, and LVEF all correlated positively with GPS, (3) high GPS was significantly associated with a worse prognosis, (4) the GPS improve the predictive accuracy of all-cause mortality compared to other predictive model that did not require any form of laboratory testing, (5) both cardiac mortality and noncardiac mortality increased with increasing GPS, (6) better prognosis were observed in the GPS 0 group in all HF subtype groups and (7) significant associations between GPS and mortality were more evident in patients with HFrEF. The GPS has been identified as a reliable predictor of mortality in several cancers, including non-small-cell lung cancer, gastro-esophageal cancer, and hepatocellular carcinoma [23][24][25] . Chronic activation of the systemic inflammatory response induces cancer progression. Inflammation leads to the accumulation of regulatory T lymphocytes and chemokines as well as the stimulation of cytokines, which produce CRP, increase neutrophils, and induce an inappropriate immune response. As a result, tumor angiogenesis, invasion, and metastasis are promoted 25 . Moreover, systemic inflammation causes malnutrition leading to decreased body weight, fatigue, loss of lean mass, declines in activities of daily living, and poor survival 24 . It is also considered that the progression of sarcopenia and malnutrition due to systemic inflammation may contribute to poor outcomes in patients with HF. Indeed, our study showed that high GPS was significantly associated with a worse prognosis in patients with HF. These are consistent with a previous report by Namiuchi et al., which found that higher GPS was an www.nature.com/scientificreports/ independent risk factor for higher mortality among admitted patients with acute decompensated HF 28 . However, the present study analyzed the GPS for risk-stratification in patients with HF in an older and larger cohort than the above report 28 29,30 . In the present study, as GPS increased, mortality had tendency to increase in patients with HFrEF and HFpEF. In particular, the tendency was evident and significant in patients with HFrEF. Additionally, in all subtype of HF including HFmrEF, low GPS was suggested to associate with better prognosis. Further assessment of nutrition and inflammation is needed for a broader HF population age spectrum and more analyses according to HF subtype.
In patients with HF, hypoalbuminemia is caused by multiple factors, including anorexia, malabsorption, high energy demand, and cytokine-induced increases in metabolic rate 31 . Hypoalbuminemia may exacerbate HF severity by promoting peripheral and pulmonary congestion and cause increased oxidative stress and inflammation to create a vicious cycle of worsening inflammation, HF, and hypoalbuminemia 31 . HF also provokes inflammation by wall stress and the release of cardiac cytokines and other inflammatory mediators from the heart and other organs. Cytokines evoke myocardial remodeling and pump failure, resulting in the progression of HF 14 .
This study demonstrated that the GPS was useful for estimating the prognosis of patients with HF. Additionally, the GPS was associated with the severity of HF because the number of patients with NYHA functional class III or IV and levels of BNP increased with increasing GPS and BNP correlated positively with GPS, which indices are recognized to reflect severity of HF 5,32 . The GPS can predict mortality using only two routinely measured parameters, CRP and Alb, both of which can be easily obtained in clinical practice. Several nutritional indicators, www.nature.com/scientificreports/ such as the GNRI, PNI, and CONUT scores, as well as inflammatory markers including CRP, have been reported to individually associate with HF prognosis and severity 6,10,11,[17][18][19][20][21][22] . Prognosticators that consist of both nutritional and inflammatory markers would therefore appear most useful for patients with HF. Since the GPS reflects systemic inflammation (elevated CRP) and malnutrition (hypoalbuminemia), it may be considered appropriate to reflect the severity of HF. Moreover, nutritional assessment alone isn't enough to evaluate severity of inflammation and treat it, which is one of the causes of malnutrition. Similarly, inflammation assessment alone isn't enough to evaluate how much inflammation affects nutritional status. Assessment of combining nutritional status and inflammation allow not only nutritional intervention but also approach chronic inflammatory diseases in the background of malnutrition. Finally, the improvement of patient GPS by the early management of nutrition and inflammation may therefore improve also prognosis.
In this study, all-cause mortality was chosen as primary endpoint. Along with the aging society, the number of elderly HF patients is increasing 33,34 . Although the most common cause of death in patients with HF is cardiac death regardless of HF subtype 35,36 , it has been reported that the proportion of noncardiac death in elderly HF patients is higher than in younger HF patients 37 . Thus, it may be important to predict all-cause mortality as well as cardiac mortality, especially in elderly HF patients. Furthermore, because of increasing frailty and progression of HF in elderly patients, palliative and end-of-life care has been needed for them. However, it is often difficult to determine when to introduce palliative care for elderly HF patients, because of their clinical course before death which differ from cancer patients. Patients with HF have repeated acute exacerbations and diminished their physical function. Each exacerbation can lead to death, but usually symptoms at that time improve with treatment. Whereas, their conditions rapidly deteriorate before they die 38,39 . Hence, it is also difficult to recognize that they are at the end-of-life 40 . To predict not cardiac mortality but all-cause mortality by using the GPS may help introducing palliative care for elderly HF patients at the appropriate time.
This study had several limitations. First, the sample size was relatively limited; fewer significant differences in the subgroup analyses may have been caused by underpowered statistics. Second, since we evaluated a cohort of admitted patients with HF, our results might not be generalizable to patients with stable HF. Third, our analysis was based on a single data point taken at discharge, with no further testing during follow-up. Therefore, the Table 3. Univariate and multivariate cox proportional hazards analysis to identify prognostic ability of the GPS. Data presented are hazard ratios and 95% confidence intervals. ACEI angiotensin-converting enzyme inhibitor, ARB angiotensin-receptor blocker, BNP B-type natriuretic peptide, CI confidence interval, eGFR estimated glomerular filtration rate, GPS Glasgow Prognostic Score, HF heart failure, HR hazard ratio, LVEF left ventricular ejection fraction, NYHA New York Heart Association.  www.nature.com/scientificreports/ impact of GPS improvement on mortality is unknown and requires further study. Fourth, the GPS is associated with acute infection or chronic inflammation such as cancer and collagen disease. There were probably no patients with acute infection in this study, because their Alb and CRP were measured at discharge, and at that time they had no clinical symptoms and physical findings of infection. However, we could not exclude patients with cancer or collagen disease. Finally, because mortality data in this study was obtained from participating institutes instead of a governmental database, our data did not accurately assess mortality in our geographic region. Despite these limitations, however, inflammation-based assessment using the GPS appears useful for the risk stratification of mortality in patients with HF.
In conclusion, high GPS was associated with worse prognosis in patients with HF. Combined assessment of inflammation and nutrition may improve the evaluation of HF severity and prognosis and assist in patient treatment.

Methods
Study design and patient population. The present multi-center, prospective, observational study was conducted in Nagano Prefecture, Japan. The inclusion criteria were 20 years of age or more and admission for acute decompensated HF. Patients with acute coronary syndrome (ACS) were excluded. After providing informed consent, subjects were enrolled between July 2014 and December 2018 during a compensated HF state before discharge. We recorded medical history, HF etiology, comorbidities, socio-economic background, HFmrEF patients with GPS 0 had significantly higher survival probability than did those with GPS 1 or 2. (C) Although HFpEF patients with GPS 0 had significantly higher survival probability than did those with GPS 1, no other remarkable differences were seen. GPS Glasgow Prognostic Score, HFmrEF heart failure with mid-range ejection fraction, HFpEF heart failure with preserved ejection fraction, HFrEF heart failure with reduced ejection fraction. www.nature.com/scientificreports/ medications, and examination findings at discharge that included electrocardiogram (ECG), echocardiography, and blood test results when HF compensated and causes of HF such as arrythmia, anemia and infection were cured. HF and ACS were diagnosed by attending physicians based on symptoms, ECGs, echocardiography, laboratory data, chest X-rays, and coronary angiograms according to the Framingham criteria 41 43 . Details of the GPS have been described previously. Patients with both elevated CRP (> 1.0 mg/dL) and hypoalbuminemia (< 3.5 g/dL) were allocated a score of 2. Patients in whom only one of these biochemical abnormalities was present were allocated a score of 1. Patients in whom neither of these abnormalities existed were allocated a score of 0 ( Fig. 1)  Kaplan-Meier survival curves were calculated from baseline to the time of death for comparisons using the log-rank test. Cox proportional hazards regression analysis was conducted to identify the prognostic ability of the GPS. The multivariate analysis model was adjusted for age, sex, systolic blood pressure, NYHA functional class, prior HF hospitalization, hemoglobin, eGFR, sodium, BNP, LVEF, and the use of ACEi and/or ARB, beta-blockers, or aldosterone antagonists. These covariates were selected in advance which are recognized as prognostic factors in HF [45][46][47] . To evaluate whether the accuracy of predicting all-cause mortality would improve after the addition of the GPS into a baseline model with established risk factors, including age, sex, hypertension, diabetes mellitus, atrial fibrillation, NYHA functional class, C-statistics, NRI and IDI were calculated. The C-statistics is defined as the area under receiver-operating characteristic curves. A p-value of < 0.05 was considered statistically significant. All statistical analyses were performed with R (The R Foundation for Statistical Computing, Vienna, Austria) and EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan), which is a graphical user interface for R 48 .

IRB information.
This study was approved by the Ethics Committee of Shinshu University Hospital (approval number: 4237) and is registered with the University Hospital Medical Information Network (UMIN 000024470).

Data availability
The deidentified participant data will be shared on a request basis. Please contact the corresponding author to request data sharing. All data, including patients' clinical characteristics, outcome data, related documents on study protocol, and statistical analysis plan, will be available by digital media to medical doctors for all types of analyses for a period of 1 year following publication of the study. www.nature.com/scientificreports/