Predicting heart failure decompensation: focus on non-invasive monitoring

Abstract

Heart failure (HF) remains the foremost global health problem. Decompensation of chronic HF, char-acterized by exacerbation of symptoms and signs of congestion, reduces functional capacity and quality of life, and increases the risk for hospitalizations and mortality. To reduce the HF burden, patients at elevated decompensation risk must be quickly identified. However, reliable, validated risk scores for the prediction of worsening HF are lacking. Therefore, this clinically oriented review aims to outline the clinical factors predisposing for HF decompensation and discuss modern strategies for the non-invasive monitoring of patients with chronic HF, including telemonitoring and artificial intelligence-based tools.

Introduction

Heart failure (HF) is a heterogeneous syndrome comprising symptoms, such as breathlessness, ankle edema, and fatigue, as well as typical signs, including elevated jugular venous pressure, pulmonary crackle, and peripheral edema. Predominantly, HF results from diastolic and/or systolic myocardial dysfunction leading to increased intracardiac pressures and/or inadequate cardiac output [1]. The prevalence of HF is estimated at 1–3% of adults and is expected to increase worldwide given the aging of populations. Heart failure remains the primary cause of hospitalizations in individuals aged > 65 years. The overall number of HF hospitalizations increases, particularly among women with HF with preserved ejection fraction (HFpEF). The annual individual cost for an HF patient in developed countries is estimated to be as high as €25,500 [2].

Chronic HF refers to stable patients with an established HF diagnosis. However, chronic HF may deteriorate once neuro-hormonal compensatory mechanisms fail to maintain hemodynamic stability. Chronic HF decompensation, involving rapid or gradual onset of troubling symptoms, poses the need for acute medication adjustments in outpatient settings, emergency department visits, or unscheduled hospital admissions [3]. Among all patients with acute HF, those with decompensation of chronic HF have higher all-cause mortality and HF readmission rates compared to those with new-onset HF [4]. Furthermore, episodes of HF decompensation become more frequent over time, leading to recurrent hospitalizations or death [5]. Therefore, early identification of HF patients at elevated decompensation risk may reduce the devastating consequences of HF decompensation and the associated burden for the healthcare systems. This clinically oriented review aims to outline the clinical factors predisposing for HF decompensation and discuss modern strategies for the non-invasive monitoring of patients with chronic HF.

Association between comorbidities and the risk of heart failure decompensation

Heart failure decompensation cannot be accurately predicted by physical examination only, and no risk score has been widely validated for this purpose. According to the consensus statement of the Heart Failure Association of the European Society of Cardiology (ESC), the Kansas City Cardiomyopathy Questionnaire and exercise tests such as the 6-minute walking test or cardiopulmonary exercise test might be used for more precise and objective assessment of worsening HF symptoms [3]. The assessment of comorbidities and baseline features such as body composition, age, race, and sex, along with therapy adherence, might be beneficial to identify factors predisposing for HF decompensation (Fig. 1).

Figure 1.

Comorbidities predisposing to heart failure decompensation; COPD — chronic obstructive pulmonary disease

Coronary artery disease

Coronary artery disease (CAD) is the most frequent underlying cause of HF and thus should be considered in all patients with new-onset HF, especially to identify those who could benefit from revascularization [1]. Impaired coronary flow reserve in HFpEF patients with angina with non-obstructive coronary arteries is associated with a > 5-fold increased risk of HF hospitalization [6]. Conversely, coronary microvascular dysfunction (CMD) in HF patients with reduced ejection fraction (HFrEF) is an independent predictor of diminished left ventricular reverse remodeling. Therefore, in the absence of obstructive CAD, assessment of coronary microcirculation might allow for tailored HF treatment [7]. Multidisciplinary management programs are recommended for HF patients with chronic coronary syndrome to reduce the risk of HF hospitalization and to improve survival [8]. Moreover, timely, guideline-recommended CAD treatment is crucial to prevent acute coronary syndrome. Acute HF occurs in approximately 38% of acute coronary syndrome cases, resulting in 3- to 5-fold increased mortality and 10–15% incidence of future HF hospitalizations [9].

Arterial hypertension

Arterial hypertension is a well-established risk factor for HF, affecting almost two-thirds of HF patients [1]. Data from the ESC Heart Failure Long-Term Registry revealed that arterial hypertension is the primary etiology of HF in 18.1% of HFpEF patients and only 4.5% of HFrEF patients [10]. Blood pressure (BP) lowering therapy exerts a protective effect against HF, while poorly controlled hypertension may result in HF decompensation [1, 11]. The recommended treatment target for patients with baseline BP > 130/80 mmHg is 120–129/70–79 mmHg if well tolerated [12]. The analysis of the OPTIMIZE-HF registry suggested that systolic BP < 120 mmHg is associated with an increased risk of HF readmission and all-cause mortality in hospitalized HFpEF patients as compared to systolic BP > 120 mmHg at 30 days [hazard ratio (HR) 1.71, p < 0.001], 1 year (HR 1.21, p = 0.004), and overall (HR 1.12, p = 0.03) [13]. Accordingly, the initial systolic BP < 120 mmHg and diastolic BP < 80 mmHg was associated with a higher risk of long-term mortality among 1594 hospitalized HFrEF patients from the Korean KorHF registry at the median follow-up of 362.5 days [odds ratio (OR) 1.81 and 2.24, respectively, p < 0.001 for all] [14]. Nevertheless, BP < 120/70 mmHg is not infrequent in patients on maximal HFrEF therapy, and therapy de-escalation is recommended only in case of symptomatic hypotension [12].

Atrial fibrillation

The prevalence of atrial fibrillation (AF) in HF patients reaches nearly 50%. Real-world data from HF patients showed that AF increases the risk of all-cause mortality and HF hospitalization (HR 1.65 and 1.32, respectively) [15]. Among the PARADIGM-HF and ATMOSPHERE trial participants with HFrEF, new-onset AF was associated with elevated rates of all-cause mortality, HF hospitalization, and stroke (HR 2.26, 2.11, and 2.20, respectively; p < 0.001 for all) [16]. Hence, identifying AF, particularly of paroxysmal phenotype, in HF patients, along with optimal management of both conditions per current guidelines, must be endeavored [1, 17]. Of note, the Thai Heart Failure Registry revealed that an average daily heart rate of 60 to 80 beats per minute is associated with the best outcomes in HFrEF patients regardless of AF status [18]. In addition, a recent analysis of 19,510 HFpEF patients demonstrated that excessive heart rate dropping during hospitalization is associated with increased rates of cardiovascular death or HF hospitalization at 12 months as compared to no and moderate dropping (17.1 vs. 14.5 vs. 14.0%, respectively; p < 0.001 for all) [19].

Valvular heart disease

Valvular heart disease is a relevant comorbidity in HF patients [1]. Moderate aortic stenosis increases the risk of HF hospitalization and mortality (HR 1.24, p = 0.01) in HFrEF patients, but aortic valve replacement may improve survival (HR 0.60, p < 0.05) [20]. Furthermore, in a post-hoc analysis of the BIOSTAT-CHF study, moderate-to-severe mitral regurgitation was associated with higher rates of a composite of all-cause death or HF hospitalization (HR 1.28, p < 0.0001) in patients with worsening chronic HF or new-onset acute HF [21]. In patients with HFrEF under optimal medical therapy, severe functional mitral regurgitation predicted all-cause mortality (HR 1.76, p < 0.001) independently of clinical and echocardiographic characteristics [22]. Therefore, timely diagnosis, evidence-based HF treatment, and valve intervention in eligible patients with severe aortic stenosis or secondary mitral regurgitation are essential to improve overall outcomes, but also to prevent HF decompensation [1].

Diabetes mellitus

Type 2 diabetes mellitus (DM) or impaired glucose tolerance occurs in 30–40% of HF patients [23]. The ESC Heart Failure Long-Term Registry demonstrated a positive correlation between DM and elevated risk of 1-year all-cause death, cardiovascular death, and HF hospitalization (adjusted HR 1.28, 1.28, and 1.37, respectively) [24]. Accordingly, the REACH registry revealed that patients with HF and DM have higher cumulative rates of cardiovascular death, myocardial infarction, and stroke compared to non-diabetic individuals at 4 years (16.5 vs. 13.1%, p < 0.001). Moreover, DM increased the odds of HF hospitalization by 33% [25]. Adequate choice of antidiabetic drugs is crucial to provide optimal management in this population. Independently of diabetic status, dapagliflozin and empagliflozin are recommended in patients with HFrEF, HFpEF, or HF with mildly reduced ejection fraction (HFmrEF) to diminish the risk of HF hospitalization and cardiovascular death [1, 26].

Chronic kidney disease

Chronic kidney disease (CKD) affects approximately 50% of all HF patients and 30% of those hospitalized for acute decompensated HF [27]. Among 3791 CRIC study participants, rate ratios for HF hospitalizations reached 1.7 and 2.2 in patients with estimated glomerular filtration rate (eGFR) of 30–44 or < 30 mL/min/1.73 m², respectively, and 1.9 and 2.6 in those with an albumin-to-creatine ratio of 30–299 or ≥ 300 mg/g, respectively. In addition, the 30-day HF readmission rate was as high as 20.6% [28]. In another CKD cohort, patients with HFpEF were hospitalized more often than those with HFrEF, but the correlation between CKD and mortality was more pronounced in HFrEF patients (HR 2.73 vs. 1.99, p = 0.0002) [29]. Accordingly, an association between admission eGFR and mortality was reported across all left ventricular ejection fraction (LVEF) categories but was slightly stronger in HFrEF [30]. Therefore, evidence-based therapies must be implemented to overcome the risk attributable to comorbid HF and CKD [1]. In patients with type 2 DM and CKD, sodium/glucose cotransporter 2 inhibitors are recommended to reduce the risk of HF hospitalization and cardiovascular death, while finerenone is recommended to avoid HF hospitalization [26].

Iron deficiency

Iron deficiency, defined as a serum ferritin concentration < 100 ng/mL or 100–299 ng/mL with transferrin saturation < 20%, is present in up to 55% of chronic HF cases and 80% of acute HF cases. Iron-deficient HF patients have an elevated risk of recurrent HF hospitalizations and all-cause mortality [1]. A meta-analysis of 10 trials with 3373 participants revealed that intravenous iron administration in HF patients with iron deficiency reduces the rates of a composite of total HF hospitalizations or cardiovascular death (risk ratio 0.75, p < 0.01) as well as the first HF hospitalization or cardiovascular death (OR 0.72, p = 0.04), as compared to standard of care or placebo [31]. Therefore, in symptomatic patients with HFrEF or HFmrEF and iron deficiency, intravenous iron supplementation is recommended to alleviate HF symptoms and to improve quality of life, and it should be considered to reduce the risk of HF hospitalization [26].

Cancer

Cancer patients might develop HF due to the cardiotoxicity of anticancer therapy or cancer itself, requiring a multidisciplinary approach [32]. A retrospective analysis of 12,867 patients with prior cancer hospitalized for acute HF demonstrated poor survival (1.6 years in prior cancer patients vs. 2.6 years in non-cancer patients). In addition, those with prior cancer were less frequently hospitalized in the cardiology ward and less frequently received the guideline-recommended HF medications [33]. Among the PARADIGM-HF and ATMOSPHERE trial participants, those with HFrEF and cancer history had a higher risk of HF hospitalization and non-cardiovascular death than those without prior cancer. However, cancer history had no impact on outcomes in HFpEF patients [34].

Pulmonary disorders

Chronic obstructive pulmonary disease (COPD) is present in about 20% of HF patients [1]. In the PARADIGM-HF trial, COPD was associated with a higher risk of HF hospitalization (HR 1.32), but not with cardiovascular or all-cause mortality in HFrEF patients [35]. Accordingly, COPD independently predicted HF and all-cause hospitalizations but not mortality in the TOPCAT study participants with HFpEF [36]. Furthermore, among myocardial infarction survivors, HF hospitalizations at 6 months were more common (9.4 vs. 4.6%, p < 0.001) and resulted in death more often in those with comorbid COPD (5.7 vs. 4.2%, p < 0.001) [37]. Because HF treatment is well tolerated in COPD patients and optimal COPD management can improve cardiac function, the guideline-recommended approach should be followed in this population [1].

In addition, HF patients can be screened for sleep-disordered breathing [1]. Sleep apnea occurs in as many as half of HF cases and correlates with an increased risk of hospitalizations, mortality, and aggravation of HF symptoms [38, 39]. Available data suggest that positive airway pressure treatment may improve outcomes in HF patients with obstructive sleep apnea, but further research is warranted [40].

Thyroid dysfunction

Thyroid hormones contribute to hemodynamic alterations and thus may affect HF decompensation. In a Danish cohort of 7380 HF patients, those with overt hypothyroidism or overt hyperthyroidism had an increased risk of HF hospitalization (one-year adjusted HR 1.51 and 1.26, respectively) and mortality (one-year adjusted HR 1.81 and 1.46, respectively) [41]. In addition, analysis of the T.O.S.CA. registry confirmed that subclinical hypothyroidism is an independent predictor of cardiovascular death in HFrEF patients (HR 2.96, p = 0.03) [42]. Therefore, thyroid function assessment and treatment of thyroid disorders by general endocrine guidelines is recommended in all HF patients [1].

Infections

Heart failure patients are at increased risk of infections, which might exacerbate HF symptoms, leading to HF decompensation [1]. The incidence rate of pneumonia in HFrEF and HFpEF patients enrolled in the PARADIGM-HF and PARAGON-HF trials was 29 and 39 per 1000 patient-years, respectively [43]. The Kyoto Congestive Heart Failure registry demonstrated that patients hospitalized for acute HF with newly diagnosed infection have higher rates of in-hospital mortality (16.3 vs. 3.2%, p < 0.001) and 1-year post-discharge mortality (19.3 vs. 13.6%, p < 0.001) compared to a non-infection group [44]. Analysis of 310,485 HF patients revealed that prior hospitalization for infection is associated with an increased risk of major adverse cardiac events, all-cause mortality, myocardial infarction, stroke, and HF hospitalization (HR 1.760, 1.587, 1.332, 1.769, and 1.993, respectively) [45]. Therefore, influenza, COVID-19, and pneumococcal vaccinations should be considered in all HF patients to avoid infections [1].

Association between baseline characteristics and the risk of heart failure decompensation

Body mass

Obesity predisposes to HF, particularly HFpEF, due to metabolic and hemodynamic alterations [1]. The risk of developing HF increases by 7% in women and 5% in men per unit of body mass index (BMI) [46]. However, according to the obesity-survival paradox, HF patients with a BMI ≥ 25 kg/m² may have a lower risk of all-cause mortality than normal-weight patients [47]. Nonetheless, in a cohort of 6142 patients with acute decompensated HF, the protective effect was confined among those aged > 75 years, with LVEF < 50% or new-onset HF [48]. No survival benefit was reported in obese HF patients with type 2 DM [49] and in those with preserved functional capacity defined as peak oxygen uptake > 14 mL/kg/min [50]. Moreover, BMI provides no information about body composition or fat accumulation, unlike other anthropometric measures such as body surface area [47]. A post-hoc analysis of the PARADIGM-HF trial revealed that the obesity-survival paradox disappears after adjustment for other prognostic factors and is less evident for the waist-to-height ratio compared to BMI. In addition, adiposity reflected by both indices increased the risk of HF hospitalization [51]. Accordingly, body surface area positively correlates with total and cardiovascular mortality, but not HF hospitalizations [52]. Consequently, maintenance of normal weight through caloric restriction and exercise training seems beneficial in HF patients [1].

Of note, bioimpedance analysis has emerged as a promising method to assess volume status in HF patients. Among 49 patients hospitalized for acute decompensated HF, a unit increase in the difference between total body water at admission and discharge measured by bioimpedance was associated with a 23% risk reduction of HF rehospitalization or cardiovascular death over 6 months (HR 0.776, p = 0.0006) [53]. Moreover, a study including 100 stable HFrEF patients showed that fluid overload assessed by bioimpedance is an independent predictor of HF decompensation at 3 months (adjusted HR 2.03, p < 0.001) [54]. Similarly, bioimpedance vector length was associated with clinical heart failure (adjusted HR 1.28) in the Chronic Renal Insufficiency Cohort comprising 3751 individuals [55]. Therefore, implementation of bioimpedance analysis might improve the clinical management of HF patients, but it requires further optimization and standardization [56]. The usefulness of impedance in the prediction of HF decompensation is supported in numerous trials, where it has been proven that this parameter used along with a combination of remaining measures of cardiac status (i.e., heart rate, blood pressure, heart rate variability) can be used successfully in domestic settings [57]

Age

A retrospective Danish nationwide cohort study reported a decline in mortality rate in 194,997 HF patients from 1996–2000 to 2016–2020. Although the results were similar in all age groups, the difference was least robust in those aged ≥ 80 years [58]. Among 2331 HFrEF patients included in the HF-ACTION trial, age modified the effect of BMI and depressive symptoms on all-cause mortality, with a significantly increased risk in patients aged ≥ 70 years. The rates of primary (all-cause mortality or all-cause hospitalization) and secondary (all-cause mortality, and cardiovascular mortality or HF hospitalization) endpoints were the highest in those aged ≥ 70 years, in contrast to those aged < 60 years [59]. In a study with 2228 hospitalized HFmrEF patients followed up for a median of 30 months, those aged > 80 years (HR 3.874, p = 0.001) and > 60 to ≤ 80 years (HR 2.211, p = 0.044) had a higher risk of all-cause mortality compared to those aged ≤ 40 years, after multivariable adjustment. Although the highest rates of HF rehospitalization were observed in those aged >80 years (4.9 vs. 14.6 vs. 16.1% for ≤ 40, > 60to ≤ 80, and > 80 years, respectively), age was nota predictor of HF rehospitalization in multivariableanalysis [60]. Correspondingly, the CHART-2 studywith 2824 HFpEF patients concluded that thoseaged ≥ 65 years are more likely to present with comorbidities that increase the risk of HF-related mortality [61].

Race

Analysis of 137,582 HF hospitalizations in the United States found 5.9-fold higher rates of HF hospitalization in non-Hispanic Afro-Americans compared to non-Hispanic Whites, but no differences between non-Hispanic Whites and Hispanic adults [62]. In addition, a recent analysis of the CDC WONDER registry revealed that HF-related, age-adjusted mortality rates in the years 1999–2020 decreased in Whites, Hispanics, and Asians//Pacific Islanders (annual percent change –0.35,–0.58 and –0.73, respectively, p < 0.01 for all),but remained stable in American Indians/Alaska Natives and Afro-Americans/African Americans. Accordingly, the highest HF-related, age-adjusted mortality rates were observed in Afro-Americans, followed by non-Hispanic Whites [63]. Therefore, optimal HF management and inspection for other socioeconomic confounders in Afro-American patients is of utmost importance.

Sex

Although the overall lifetime risk for HF is comparable between women and men (23.3 vs. 24.5%, respectively), the lifetime risk for HFpEF is higher in women (11.5 vs. 6.4%, respectively), while the lifetime risk for HFrEF is greater in men (18.1% vs. 11.9%, respectively). In the PREVEND study with 8558 participants, the mean age at onset of HF was 74.2 years in women compared to 72.1 years in men [64]. Indeed, the overall cardiovascular risk in women increases following menopause and may be altered by female-specific risk factors [47]. First, acute HF may arise due to pre-eclampsia, peripartum cardiomyopathy, and amniotic fluid embolism [65]. Hypertensive disorders of pregnancy are the most frequent risk factor for pregnancy-related HFpEF, which accounts for 7 per 100,000 pregnancy-related hospitalizations [66]. Notably, women with pre-existing HFrEF are at increased risk of pregnancy complications. Thus, pregnant HF patients require multidisciplinary care with adjustment of HF medication to protect the mother and avoid fetal harm [67]. Second, radiotherapy, as well as common breast cancer therapeutics — anthracyclines and trastuzumab — increase the risk of HF [68]. In women with incident invasive breast cancer or incident hospitalized HF, prevalent HF or prevalent breast cancer, respectively, is associated with increased mortality [69]. Third, evidence for the efficacy and safety of HF treatment in women may be limited due to female underrepresentation in clinical trials [70]. However, sex is not a predictor of guideline-directed medical therapy use or adverse events in HF patients [71]. One-year outcomes following hospitalization for acute decompensated HF are comparable between women and men [72]. Notwithstanding, given varying HF etiology, older age, and female-specific comorbidities, effective HF management in women may be challenging.

Laboratory assessment to predict heart failure decompensation

Biomarkers

Measurement of N-terminal pro-B type natriuretic peptide (NT-proBNP) and BNP concentrations remains the initial step of the diagnostic algorithm for HF. Moreover, both NT-proBNP and BNP, along with mid-regional pro-atrial natriuretic peptide, are used to confirm or rule out a new-onset acute HF [1]. In the GUIDE-IT trial with 894 HFrEF patients, doubling NT-proBNP levels was associated with an elevated risk of cardiovascular death or HF hospitalization (HR 1.17, p = 0.0003). An increase in NT-proBNP levels preceded the onset of adjudicated events, suggesting that routine NT-proBNP monitoring might support the clinical decision-making process [73]. Interestingly, women achieving NT-proBNP levels ≤ 1000 pg/mL experienced 82% risk reduction of the primary endpoint, compared to only a 59% reduction observed in men [74]. Conversely, in the NICE trial including 157 HFpEF patients, NT-proBNP monitoring did not allow for prediction of HF rehospitalization or HF decompensation on a populational level [75], but the individual variations of this compound remain of unclear importance. In addition, natriuretic peptides might be elevated due to other cardiac causes, such as acute coronary syndrome or AF, and non-cardiac conditions such as CKD or COPD. In contrast, lower concentrations are observed in obesity and overweight [1]. Although natriuretic peptides are good prognostic markers, routine measurement to guide HF therapy is not supported by the current ESC guidelines [1]. Other biomarkers that might allow for prediction of worsening HF are listed in Table 1.

Table 1.

Potential biomarkers to predict heart failure (HF) decompensation [3]

Biomarker	Congestion, inflammation, or prognosis
NPs	Prognosis
cTn	Prognosis, inflammation
CA-125	Prognosis, congestion, inflammation
Bio-ADM	Prognosis, congestion
Kidney markers (e.g., eGFR, NGAL, miRNA, cystatin C)	Prognosis, congestion
Albuminuria	Prognosis, congestion
Procalcitonin	Inflammation
sST2, GDF-15, galectin-3	Prognosis, inflammation

Bio-ADM — biologically active adrenomedullin; CA-125 — carbohydrate antigen 125; cTn — cardiac troponin; eGFR — estimated glomerular filtration rate; GDF-15 — growth differentiation factor-15; miRNA — microRNA; NGAL — neutrophil gelatinase-associated lipocalin; NPs — natriuretic peptides; sST2 — soluble suppression of tumorigenicity 2

Laboratory abnormalities

Heart failure medications, such as renin-angiotensin system inhibitors and diuretics, might lead to electrolyte disturbances. Severe hypokalemia, hyperkalemia, and hyponatremia are life-threatening conditions, while hypochloremia independently predicts mortality in HF patients. Hence, blood tests for electrolytes are recommended as a part of routine HF follow-up. Furthermore, diuretic treatment may result in hyperuricemia, which is prevalent in about 50% of HF patients [1]. A study including 1943 patients with chronic HF showed a J-shaped correlation between serum uric acid and a composite of mortality and 6-month readmission [76]. A meta-analysis of 33 studies demonstrated that for every 1 mg/dL elevation in uric acid levels, the risk of all-cause mortality and the composite endpoint in HF patients increased by 4 and 28%, respectively [77]. Although allopurinol and febuxostat reduce uric acid levels, their effect on HF outcomes remains uncertain. Of note, non-steroidal anti-inflammatory drugs used in gout attacks and arthritis might precipitate HF decompensation and thus should be avoided [1]. In addition, rheumatoid arthritis increases the risk of HF and might be associated with an increased risk of all-cause mortality in HFrEF patients; hence, they should be addressed in HF management [78].

Medical imaging to predict heart failure decompensation

Echocardiography

Echocardiography is a recommended method to assess cardiac function, confirm or rule out HF diagnosis, and define HF phenotype [1]. A 12-month increase in LVEF of ≥ 5 percentage units was associated with a reduced risk of all-cause mortality, cardiovascular mortality, and HF hospitalization (HR 0.62, 0.54, and 0.66, respectively) over a 3-year follow-up of 2484 HFrEF patients [79]. A study with 1082 patients initially diagnosed with HFpEF demonstrated that LVEF < 40% at 6-month follow-up, which occurred in 138 participants, is an independent predictor of cardiac events as compared to sustained LVEF > > 50% (HR 1.424, p = 0.039) [80]. A meta-analysis of 46 studies comprising 20,056 HFpEF patients identified three independent echocardiographic predictors of the composite of cardiovascular death and HF hospitalization: left ventricular global longitudinal strain (HR 1.24 per 5% decrease), left atrial reservoir strain (HR 1.30 per 5% decrease), and tricuspid annular plane systolic excursion (TAPSE) to pulmonary artery systolic pressure (PASP) ratio (HR 1.17 per 0.1-unit decrease). Increasing E/e’, left ventricular mass index, left atrial ejection fraction, and moderate-to-severe tricuspid regurgitation appeared independent but less robust predictors [81]. In a study with 176 HFrEF patients, higher TAPSE and lower left atrial volume index were associated with improved LVEF, and hence with better survival [82]. Other echocardiographic parameters relevant to HFrEF prognosis include global longitudinal strain, E/e’, and left atrial area [83]. In addition, tricuspid regurgitation jet velocity was identified as an independent predictor of acute decompensated HF among 188 patients with end-stage renal disease (OR 8.356, p = 0.007) [84]. Although serial echocardiography is not necessary, the current ESC guidelines advocate repeated testing if the clinical status has deteriorated, as well as 3–6 months after HFrEF therapy optimization [1]. Of note, ultrasound might also be used to assess other signs of congestion associated with worsening HF, such as lung B-lines, jugular vein diameter, intra-renal venous flow, or inferior vena cava diameter, as well as to detect changes in pulmonary and residual congestion [3].

Cardiac magnetic resonance

Cardiac magnetic resonance (CMR) is currently recommended for the assessment of myocardial structure and function in HF patients with poor echocardiogram acoustic windows and those with suspected myocarditis, amyloidosis, sarcoidosis, Chagas disease, Fabry disease, left ventricular non-compaction, or hemochromatosis. Cardiac magnetic resonance imaging with late gadolinium enhancement should also be considered to distinguish between dilated cardiomyopathy and ischemic heart disease [1]. The Society for Cardiovascular Magnetic Resonance Registry, including 3837 HF patients, showed that a diagnosis confirmed with CMR differs from the pre-test indication in 49% of cases [85]. Another retrospective analysis revealed that a new, post-CMR diagnosis was established in 38.7% of 243 HF patients, altering treatment strategy in 16.9% of the cases [86]. A meta-analysis of 9 studies demonstrated that scar detected by late gadolinium enhancement (HR 1.6, p = 0.008), fibrosis on T1-mapping (HR 1.25, p < 0.001), myocardial ischemia on stress CMR, and right ventricular dysfunction (HR 3.19, p = 0.03) are associated with adverse outcomes, including HF hospitalization, cardiac transplantation, and death, in HFpEF patients [87]. Accordingly, a cardiovascular magnetic resonance study of 120 participants with new-onset HFrEF followed-up for a median of 8.9 years showed that the fibrosis pattern was predictive for a composite of all-cause mortality, HF hospitalization, or aborted sudden cardiac death (HR 2.69, p = 0.011 for infarct, and HR 2.97, p = 0.006 for midwall fibrosis) [88].

Telemonitoring to predict heart failure decompensation

According to the current ESC guidelines, non-invasive home monitoring may be considered in HF patients to reduce the risk of recurrent cardiovascular and HF hospitalizations and cardiovascular death [1]. The TIM-HF2 trial with 1571 HF patients proved that remote patient management comprising telemonitoring, telephone interviews, and cooperation between the treating physician and telemedical center is associated with a lower percentage of days lost due to unplanned cardiovascular hospitalization (4.88 vs. 6.64%, p = 0.046) and a lower rate of all-cause death (7.86 vs. 11.34 per 100 person-years of follow-up, p = 0.028) compared to the usual care [89]. In a German study including 6065 HF patients at high risk for rehospitalization, telemonitoring reduced the number of HF rehospitalizations (17.9 vs. 21.8 per 100 patient-years, p < 0.001) and all-cause hospitalizations (129.0 vs. 133.2 per 100 patient-years, p = 0.015) [90]. In the LINK-HF study investigating wearable technology, machine learning-based analysis of remotely collected physiological data of 100 HF patients predicted rehospitalization for HF decompensation with 76–88% sensitivity and 85% specificity. Notably, initial alert preceded readmission by a median of 6.5 days, allowing for medical intervention to prevent hospitalization [91]. A meta-analysis of 65 non-invasive telemonitoring studies and 27 invasive telemonitoring studies with a total of 36,549 HF patients followed-up for a mean of 11.5 months demonstrated an association between telemonitoring and a significantly reduced risk of all-cause mortality (pooled OR 0.84, I² 24%), first HF hospitalization (OR 0.81, I² 22%), and total HF hospitalizations (OR 0.85, I² 70%) [92]. Remote monitoring of implantable devices used in the treatment of heart failure, such as cardiac resynchronization therapy devices or implantable cardioverter-defibrillators, allows changes in cardio-neural changes to be tracked by continuously tracking physiological parameters. This approach enables early identification of at-risk patients, allowing for timely interventions and potentially reducing hospitalizations. Numerous studies have been investigating this issue, but due to the heterogenous methods used for remote monitoring, the sole effect of this medical intervention remains unclear [57]. Moreover, an increasing role of data analysis shows that the predictive value can be achieved by observing a set of parameters and the interplay between them [93–95]. Importantly, given the methodological diversity, effective telemonitoring requires standardization, personalization, and integration with an engaging, informing, and empowering system of care [96], as summarized in Figure 2.

Figure 2.

Telemonitoring as a part of heart failure (HF) management; BP — blood pressure; HR — heart rate

Emerging role of artificial intelligence

Machine learning has been used to create prediction models for HF decompensation. For instance, a model applied across the LVEF spectrum displayed an area under the curve (AUC) of 0.76 for worsening HF events and 0.83 for mortality [97]. A model developed based on data from the TIM-HF2 trial outperformed the conventional algorithm (AUC 0.855 vs. 0.727, p < 0.001) and showed a continuously increasing risk for HF hospitalization in the three weeks preceding the admission [98]. A risk score dedicated to HFrEF patients, incorporating age, COPD, eGFR, NT--proBNP, uric acid, end-diastolic mid-right ventricular diameter, mitral annular plane systolicexcursion (MAPSE), systolic pulmonary arterypressure (sPAP), and moderate-to-severe mitralregurgitation, demonstrated an AUC of 0.678 forcardiovascular events and 0.740 for cardiovascularmortality [99]. Correspondingly, a model to predicta composite of HF hospitalization or cardiovascular death in HFpEF patients comprising age, AF, BMI, BNP, sodium, uric acid, triglyceride, blood urea nitrogen, glycated hemoglobin, left atrial size, mitral regurgitation, left ventricular posterior wall thickness, intraventricular septum thickness, and previous HF hospitalizations displayed an 86.9% AUC [100]. Therefore, the implementation of artificial intelligence might improve the prediction of HF decompensation.

Conclusions

The risk of HF decompensation, along with the associated hospitalizations and mortality, is influenced by both cardiac and non-cardiac comorbidities. Coronary artery disease, AF, DM, CKD, COPD, arterial hypertension, valvular heart disease, particularly severe aortic stenosis, and secondary mitral regurgitation, hypo- and hyperthyroidism, as well as iron deficiency and sleep apnea, require timely diagnosis and evidence-based co-management in HF patients. Vaccinations are recommended to prevent infections as potential triggers of HF decompensation. Comprehensive, guideline-recommended approaches may be necessary in women and cancer patients with HF, while special attention should be paid to Afro-Americans and those in advanced age to provide accurate treatment and avoid adverse outcomes. Maintenance of normal weight seems beneficial in HF patients, but waist-to-height ratio or body surface area might exhibit better prognostic relevance compared to BMI alone. Risk stratification may be further improved by analysis of fluid overload through bioimpedance. Natriuretic peptides are established diagnostic and prognostic markers, but laboratory assessment should also detect electrolyte disturbances and hyperuricemia. Echocardiography assesses not only diagnostically essential LVEF, but also other prognostically relevant parameters, such as PASP, TAPSE, E/e’, global longitudinal strain, or tricuspid regurgitation. Ultrasound may also be useful to detect subclinical congestion associated with adverse events. The use of CMR in uncertain cases might enhance diagnostic accuracy and allow for patient-tailored therapy. Finally, the incorporation of artificial intelligence and telemonitoring into a dedicated system of care offering regular follow-ups and patient education may be the key to personalized HF management and reduced HF burden. The clinical variables predisposing to HF decompensation and the non-invasive monitoring tools allowing it prediction are summarized in Central illustration.

Central illustration.

Clinical variables predisposing to heart failure decompensation and the non-invasive monitoring tools to predict it