Inflammation-Based Scores as a Common Tool for Prognostic Assessment in Heart Failure or Cancer

Background: Inflammation-based scores are widely tested in cancer and have been evaluated in cardiovascular diseases including heart failure. Objectives: We investigated the impact of established inflammation-based scores on disease severity and survival in patients with stable heart failure with reduced ejection fraction (HFrEF) paralleling results to an intra-institutional cohort of treatment naïve cancer patients. Methods: HFrEF and cancer patients were prospectively enrolled. The neutrophil-to-lymphocyte-ratio (NLR), the monocyte-to-lymphocyte-ratio (MLR), the platelet-to-lymphocyte-ratio (PLR), and the prognostic nutritional index (PNI) at index day were calculated. Association of scores with disease severity and impact on overall survival was determined. Interaction analysis was performed for the different populations. Results: Between 2011 and 2017, a total of 818 patients (443 HFrEF and 375 cancer patients) were enrolled. In HFrEF, there was a strong association between all scores and disease severity reflected by NT-proBNP and NYHA class (p ≤ 0.001 for all). In oncologic patients, association with tumor stage was significant for the PNI only (p = 0.035). In both disease entities, all scores were associated with all-cause mortality (p ≤ 0.014 for all scores). Kaplan–Meier analysis confirmed the discriminatory power of all scores in the HFrEF and the oncologic study population, respectively (log-rank p ≤ 0.026 for all scores). A significant interaction with disease (HFrEF vs. cancer) was observed for PNI (pinteraction = 0.013) or PLR (pinteraction = 0.005), respectively, with higher increase in risk per inflammatory score increment for HFrEF. Conclusion: In crude models, the inflammatory scores NLR, MLR, PLR, and PNI are associated with severity of disease in HFrEF and with survival in HFrEF similarly to cancer patients. For PNI and PLR, the association with increase in risk per increment was even stronger in HFrEF than in malignant disease.


INTRODUCTION
Heart failure with reduced ejection fraction (HFrEF) is a major cause of morbidity and mortality in developed countries (1). Comparable to cancer, previous research emerges that show that elevated inflammatory markers are characteristic in patients with HFrEF and correlate with disease severity and prognosis (2,3). This is reflected by a steady immune activation and low-grade inflammation in response to various stimuli including hypoxia, proinflammatory cytokines, and neurohormonal activation (4,5). Sustained expression and exposure to cytokines and proteolytic enzymes can lead to myocardial remodeling and altered cardiac metabolism resulting in ventricular remodeling and dysfunction, negative inotropic effects, and heart failure (HF) progression (6,7). Especially in more advanced stages of HF, the expression of inflammatory mediators leads to detrimental effects on other systems contributing to more systemic deterioration and end organ damage in the course of the chronic disease (8,9).
In recent years, the role of this low-grade systemic inflammatory activity in chronic diseases generally and especially in cancer development and progression has been investigated more thoroughly, proving an association between the level of chronic inflammation and worse outcome (7,10,11). Consequently, inflammation-based prognostic scores have been introduced as a simple index of systemic inflammation and suggested for improved risk stratification. The neutrophil-tolymphocyte ratio (NLR), the monocyte-to-lymphocyte ratio (MLR), and the platelet-to-lymphocyte ratio (PLR), as well as the prognostic nutritional index (PNI) share mutual quantitative relationships of blood cells and inflammatory activity. Inflammation-based scores are widely tested in cancer (12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28) and have been evaluated in primarily acute settings of cardiovascular diseases including heart failure (29)(30)(31)(32)(33)(34)(35)(36)(37). We aimed to broaden knowledge by investigating the impact of established inflammation-based scores on disease severity and survival in patients with stable chronic HFrEF and, more important, paralleling results to an intra-institutional cohort of treatment naïve cancer patients.

METHODS
The study protocol complies with the Declaration of Helsinki and was approved by the ethics committee of the Medical University of Vienna. All included patients had to be at least 18 years of age and provided written informed consent to study participation.

Study Population
Study participants for this observational, non-interventional study were recruited at the heart failure and the oncologic outpatient clinic at the Vienna General Hospital, both universityaffiliated tertiary centers, respectively. Patients with stable chronic HFrEF undergoing routine ambulatory care have been identified from a prospective registry. Heart failure with reduced ejection fraction was defined in line with the guidelines as a history of HF signs and symptoms as well as a documented left ventricular ejection fraction below 40% (1). Baseline examination included medical history, detailed assessment of current medication, and an electrocardiogram. Cardiovascular risk factors were recorded (1). Routine laboratory parameters for the respective visit were available.
Second, consecutive patients with a primary diagnosis of cancer but treatment naïve were prospectively enrolled. To avoid inflammatory bias, eligible patients were excluded if they showed clinical signs of infection. Patients were classified according to tumor entity and tumor stage. Comorbidities and current medical therapy were recorded. Further details on the patient cohort have been published previously (38)(39)(40)(41). The present analysis incorporates a subpopulation of patients for whom data for the calculation of respective risk scores were available.
All-cause mortality was defined as the primary endpoint and information on death was obtained from the Austrian Central Office of Civil Registration.

Laboratory Assessment
Laboratory parameters were assessed from venous blood samples drawn from a peripheral vein at the respective outpatient clinic visit. Routinely available laboratory parameters were analyzed according to local laboratory standard procedures.

Assessment of Risk Scores
All scores were assessed as previously defined (12,14,18,25). The NLR, MLR, PLR, and PNI were defined as the ratios of the measured parameters (absolute numbers) as follows: NLR: the ratio of neutrophils (g/L) and lymphocytes (g/L), MLR: the ratio of monocytes (g/L) and lymphocytes (g/L), PLR: ratio of platelets (g/L) and lymphocytes (g/L), PNI: product of albumin and the total lymphocyte count [PNI = albumin (g/L) × total lymphocyte count (g/L)].

Statistical Analysis
Continuous variables are presented as median and interquartile range (IQR) and compared by using the Mann-Whitney Utest. Discrete data were presented as count and percentage and analyzed by using a χ 2 test. Central tendencies and correlation between assessed risk scores and disease severity reflected by New York Heart Association (NYHA) functional class and Nterminal pro-brain natriuretic peptide (NT-proBNP) (HFrEF population), or tumor-stage (oncologic population; solid tumors only), were compared by Kruskal-Wallis test and Spearman-Rho correlation coefficient, respectively. For outcome analysis, univariate and multivariate Cox proportional hazard regression analysis was applied to estimate hazard ratios (HRs) with 95% confidence interval (CI). Interaction analysis was performed to test differences in risk prediction between HFrEF and cancer of the respective scores. Kaplan-Meier analysis was applied to illustrate the prognostic ability of the respective scores for strata of tertiles whereas groups were compared using the log-rank test. Two-sided p < 0.05 were used to indicate statistical significance. The SPSS 24.0 software (IBM Corp, New York, NY) was used for all analyses. Illustrations were created with GraphPad Prism version 9.1.0.

Overall Study Population
Baseline characteristics of the study patients are detailed in

Association Between Risk Scores and Disease Severity
All scores differed significantly between both disease entities ( Table 2).

Predictive Power of Assessed Risk Scores and Outcome
All assessed risk scores were associated with all-cause mortality in crude Cox regression analysis in HFrEF or in cancer, respectively (p ≤ 0.014 for all, Figure 2). In the HFrEF study population, except for the NLR, all scores remained significantly associated with outcome after adjustment for NYHA classification, physical confounders reflected by age, sex, and BMI, or for laboratory parameters. The PNI remained significantly associated with outcome even after adjustment for NT-proBNP, the most prominent heart failure specific biomarker (Supplementary Table 2A). In the cancer study population, all scores remained significantly associated with outcome after adjustment for tumor stage (in the solid tumor study population only). Solely, the PLR lost significance when adjusting for the physical status confounder model and the laboratory confounder model (Supplementary Table 2B). Interaction analysis revealed significance between disease (HFrEF vs. cancer) and the PNI or the PLR, respectively, with higher increase in risk per inflammatory score increment for HFrEF (PNI: p interaction = 0.013; PLR: p interaction = 0.005). This was not observed for the NLR (p interaction = 0.258) or the MLR (p interaction = 0.192). The Kaplan-Meier analysis confirmed the discriminatory power of the scores for strata of tertiles for both cohorts (log-rank HFrEF study population: NLR p = 0.001, PLR p = 0.009, MLR p < 0.001, PNI p < 0.001; log-rank oncology study population: NLR p < 0.001, PLR p = 0.026, MLR p < 0.001. PNI p = 0.002; Figure 3).

DISCUSSION
The present long-term observational study showed inflammation-based risk scores NLR, MLR, PLR, and PNI, originally developed for prognostic assessment in malignancies, as a potent tool for refining risk in stable chronic HFrEF. The analyses directly contrasted the ability of the scores to predict outcome in patients with malignancies or with heart failure. We found that (i) in the HFrEF study population, the association with disease severity was significant for all scores; (ii) all scores were significantly associated with outcome at long-term followup in both disease entities, HFrEF or cancer, with ROC highest for NLR and PNI; and (iii) while PLR and PNI show a significant interaction with respect to the underlying disease, NLR and MLR do not.

Shared Pathophysiologic Inflammatory Pathways Reflected by Ratios of Inflammatory Cell Lines in HFrEF and Cancer
Besides other joint risk factors (42), also the involvement of systemic inflammation is a recognized crucial pathophysiologic factor in disease development and progression, which is shared between HF and cancer. Several studies provided evidence of enhanced inflammation in either HF or cancer (3). Indeed, in heart failure, disease progression has been attributed to, among others, general but subclinical and non-specific inflammatory processes that have been associated with outcome (9,11,43). Nevertheless, the distinct immunologic mechanism is not entirely understood, but several players have been recognized. The specific dysregulated cell lines of interest include neutrophils, monocytes, platelets, and leucocytes (44,45). All of these cells are responsible for inflammatory event cascades. They induce and control specific programs involving complex series of paracrine and endocrine signals, attract additional cells, stimulate the production and secretion of pro-inflammatory factors, and induce the expression of receptors or factors that are receptive to additional inflammatory signals potentially contributing to disease progression (45). Comparable activated mechanisms of the immune system have been implied in the development and progression of malignancies (45). With the observation that systemic inflammation features circulating white blood cells' and platelets' alteration, the calculation of blood cell-based ratios/scores, as also examined in our study, was established. Thereby, it is sought to assess and further understand the burden and respective prognostic significance of subclinical inflammatory disarrangements. The presented data of the two distinct populations of heart failure and cancer patients may improve our understanding about shared pathways between the two diseases thereby adding to the field of cardio-oncology. Since the respective score calculation was primarily described for oncological patients, a wide range of analyses can be found in the literature. These include cancer entity-specific evaluations in, among others, lung cancer (12)(13)(14), head and neck malignancies (15,16), gynecologic (17) and gastrointestinal cancers entities (18)(19)(20), and hematologic malignancies (21,22), but also prediction analyses of scores independent of the tumor site of origin (23). The individual expressions of the ratios vary between the respective study cohorts, making it difficult to define uniform cutoff values. However, the respective trends of each score are consistent and comparable with those of our study population. Thus, evidence from our and previous analyses indicates that increased NLR, PLR, or MLR are each of significant prognostic value for outcome assessment in cancer. For the first time, the prognostic value was also shown for patients without prior anticancer therapy, demonstrating that the inflammatory activation can be attributed to the underlying malignant disease. To date, only few studies addressed the use of the aforementioned scores in cardiovascular outcome assessment. Data available show association between PLR and 30-day mortality in acute heart failure (33). In these patients, the NLR was shown to be associated with 30-day mortality, readmission rate, and long-term outcome (33,34). In addition, the score reveals association with mortality or heart transplantation in advanced heart failure (35). The risk assessment for heart failure development by MLR has been described for patients undergoing coronary angiography. In these patients, MLR was related to heart failure markers NT-proBNP, left ventricular ejection fraction, and HF hospitalizations during follow-up (36). We extended existing data with our analysis by showing prognostic validity of the scores in stable chronic HFrEF including a long-term follow-up time period. The data revealed a significant association of NLR, PLR, and MLR with disease severity, reflected by NT-proBNP and NYHA functional class, and outcome. More important, in direct comparison to cancer patients, an equipotent relevance of these markers is shown for both diseases.

The Prognostic Nutritional Index, a Rather Inflammatory Score
Even though serum albumin levels are widely used to determine the patient's nutritional status, in heart failure, hypoalbuminemia has been found present to a similar extent in lean, overweight, and obese patients (46). Furthermore, the positive overall correlation of hypoalbuminemia and inflammation with prognosis in patients with certain chronic diseases, such as cancer (47), heart failure (46), and end-stage renal disease (48), and both factors' mutual association (49) indicate inflammation to be among the underlying etiologies of hypoalbuminemia.
The PNI, calculated by lymphocytes and albumin, is often interpreted as a pure nutritional marker. However, the score, based on parameters representing driving forces in inflammatory processes, should be interpreted equivalently as a reflection of systemic inflammation in the course of diseases (25). In our study populations, PNI was negatively correlated with outcome and disease severity, respectively. This observation is supported by previous data in cardiovascular disease or cancer (50). Initially, the score has been designed for gastro-intestinal malignancies (19,24,25). Since then, its applicability has been investigated in the extended oncological field, including, e.g., ovarian (26), lung (27), or renal cancer (28). With regard to cardiology, there are data on myocardial infarction, chronic, and acute heart failure that support the applicability of PNI for prognostic assessment of these patients (30)(31)(32)35).

LIMITATIONS
The study reflects the experience of a single tertiary care center only. However, this ensures the inclusion of a homogenous patient population with adherence to a consistent clinical routine. Moreover, the study design allows for the so far unique, interdisciplinary direct comparison between two disease entities, which also enables interaction analyses between disease pattern and respective risk score.

CONCLUSION
In crude models, the inflammatory scores NLR, MLR, PLR, and PNI are associated with severity of disease in heart failure and with survival HFrEF similar to cancer patients. For PNI and PLR, the association with greater increase in risk per increment was even stronger in HFrEF than in malignant disease, whereas there was no disease-dependent interaction in NLR and MLR.

DATA AVAILABILITY STATEMENT
The dataset is not intended for forwarding to third parties. Requests for access to specific questions should be addressed to the corresponding author.

ETHICS STATEMENT
This study involving human participants was reviewed and approved by Ethics Committee of the Medical University of Vienna. The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
HA, NP, and MH designed the experiment and wrote the manuscript. HA, NP, AC, SP, GSp, JN, GG, PB, MR, HG, GK, and WK managed participant recruitment and were responsible for sample acquisition. HA, NP, MH, GG, PB, and GSt analyzed and interpreted the data. MH was the supervisor of the study. All authors contributed to manuscript revision, gave important intellectual content, read, and approved the submitted version.

FUNDING
The work was supported by a grant from the Austrian Science Fund (KLI 700-B30).