Article Text
PDF
Abstract
Heart failure (HF) represents a major health and economic burden worldwide. In spite of best current therapy, HF progresses with unpredictable episodes of deterioration that often require hospitalisation. These episodes are often preceded by accumulation or redistribution of fluid causing haemodynamic overload on the heart. Remote and telemonitoring of the HF patient, assessing symptoms and signs, thoracic impedance derived fluid status follow-up or direct haemodynamic measurements with chronic implanted devices are presently under investigation for the potential to detect impending HF decompensation early. The current evidence for volume status monitoring in HF using those novel management strategies is reviewed.
Statistics from Altmetric.com
Introduction
Heart failure (HF) represents a major health and economic burden which is increasing with the ageing of populations around the world. In the USA, over 5.7 million people are currently estimated to live with HF.1 In Europe, over 15 million people are estimated to have HF, and with a similar prevalence of asymptomatic left ventricular (LV) dysfunction, approximately 4% of the European population has either HF or LV dysfunction.2 Despite advances in pharmacological and other therapies, rates for HF related hospital admission have not substantially decreased and represent a major driver for healthcare expenditure.1 Recent data indicate that inhospital care accounts for approximately 60% of total HF costs.3 Rehospitalisation for worsening HF predicts adverse prognosis, especially in the elderly, and is often initiated by intrathoracic fluid overload leading to symptomatic pulmonary congestion.4 5 The vast majority of patients with acute decompensated HF (ADHF) has underlying chronic HF. Our current understanding of mechanisms contributing in ADHF is still insufficient but altered LV loading conditions and hypervolaemia are likely important contributing factors. Intrathoracic fluid accumulation frequently precedes hospital admission. Conceptually, continuous monitoring of fluid status in HF patients could aid identification of volume overload, thus providing an opportunity to intervene at an early stage and possibly avert hospital admission for ADHF. However, early clinical detection of ADHF is challenging.6–8 Haemodynamic disturbances underlying ADHF may start weeks before the actual onset of typical HF symptoms such as fatigue, body weight gain or shortness of breath. Moreover, these are common, especially in the elderly without HF, and may be overlooked both by doctors and patients themselves. Diagnostic tools widely used in HF workup such as chest x-ray, cardiac catheterisation and conventional echocardiography are of limited use in determining the individual patient's fluid status.7 9–11
Biomarkers in the assessment of clinical status of HF have emerged over the past two decades and are now routinely measured in various clinical settings. While the role of B type natriuretic peptide (BNP) in diagnosis as well as prognostification of HF is well established, there has been ongoing debate regarding its role as a guide to monitoring and adjustment of HF therapy. Recent meta-analyses of major randomised controlled trials (RCTs) in the field have suggested a mortality benefit in patients with monitored BNP, presumably due to enhanced use of drugs such as angiotensin converting enzyme inhibitors (ACEI) and β blockers in the cohort exhibiting biomarker increases.12 13 Another report concluded that N terminal BNP guided HF specialist care in addition to home based nurse care was cost effective and cheaper than standard care.14 There are conflicting data as to whether BNP guided HF care reduces rehospitalisation rates.13 15 BNPs may not be sensitive enough tools to detect rapidly decompensating HF. In ADHF, acute changes in LV filling pressures will likely not be reflected by simultaneous changes in NPs due to their long half-lives, thus limiting their clinical utility in that setting. Furthermore, patient characteristics (ie, age, gender, body weight) may influence plasma levels of BNP and other NPs, making interpretation even more difficult.9 10
Therefore, novel strategies to more precisely assess and monitor fluid status in HF have been explored over recent years. Some of those developments seem to hold promise in improving early detection of which patients will likely be readmitted for ADHF, with the potential to intervene early. Bringing down HF hospitalisation rates may not only improve patient quality of life but also reduce longer term clinical outcomes and alleviate the enormous HF related cost to society.
This review seeks to summarise current knowledge on integrating fluid status monitoring into the overall management of HF patients.
Emerging strategies to monitor fluid status in HF
Home and telemonitoring
Given the importance of hypervolaemia in HF related events, monitoring of weight and HF specific symptoms as a surrogate for fluid status has received considerable attention in recent years. Efforts have been made to systematically and continuously assess fluid status associated variables either at clinical follow-ups or through structured telephone calls. However, it has been unclear whether those strategies translate into clinical benefit. Several recent studies have sought to establish evidence for such a benefit (table 1).
Overview of important studies of fluid monitoring in heart failure
The Weight Monitoring in HF (WHARF) trial was a large multicentre RCT of a technology based daily weight and symptom monitoring system.16 It included HF patients in New York Heart Association (NYHA) class III or IV. The trial failed to meet its primary endpoint of reduced 6 month rehospitalisation rates but demonstrated a substantial reduction in the secondary endpoint of mortality.
The Trans-European Network-Home Care Management System (TEN-HMS) study was a large scale RCT comparing home based telemonitoring services or nurse based telephone support to usual care.19 20 In TEN-HMS, telemonitoring failed to meet its primary endpoints of days lost to death or hospitalisation improvements of patient quality of life, but both interventions led to lower 1 year mortality than usual care.
A recent report by the Cochrane Review Group compared structured telephone interview and telemonitoring to standard care.23 That meta-analysis comprised over 8000 patients and included 11 studies (all published before the end of 2008) which evaluated telemonitoring (total of 2710 subjects) and 16 which evaluated structured telephone support (5613 subjects). Telemonitoring reduced all-cause mortality while structured telephone support showed a non-significant trend. Both interventions reduced HF hospitalisations. Heterogenous protocols and the small sample size of most of the trials included in that report warrant caution when interpreting the ascribed benefits.
Further illustrating the limitations of pooled efficacy data, two very recent large RCTs (not included in the aforementioned Cochrane review) have raised doubts as to the benefits of telemonitoring. First, the Telemedicine to Improve Mortality in Heart Failure (TIM-HF) study evaluated 710 patients with NYHA class II or III HF, LV ejection fraction (EF) ≤35% and on optimal medical therapy.22 Using portable devices, ECG, blood pressure and body weight measurements of the telemonitored cohort (n=354) were reviewed daily by telemedical centres. After a mean follow-up of 26 months, telemonitoring had no significant effect on all-cause mortality, cardiovascular death or HF hospitalisation compared with patients receiving usual care (n=356). In the even larger telemonitoring for HF (TELE-HF) trial, 826 patients recently hospitalised for HF were randomised to daily telemonitoring by means of a telephone based interactive voice response system collecting data on weight and symptoms, and compared with 827 patients on standard care.21 Data in the telemonitored cohort were reviewed by the patients' clinicians. The primary endpoint was readmission for any reason or death from any cause within 180 days after enrolment. Secondary endpoints included hospitalisation for HF, number of days in the hospital and number of hospitalisations. Again, telemonitoring in TELE-HF did not improve any of these outcomes. Moreover, no subgroup (age, gender, EF, etc) could be identified that benefitted from the intervention. Importantly, adherence to the intervention decreased from an initial 90.2% to only 55.1% by 6 months, and almost 15% of patients never actually used the device. There was no per protocol analysis to allow conclusions on potential benefits in those study subjects that adhered to the intervention. TELE-HF did not report information on medication changes or on how clinicians used information gained from telemonitoring.
In an effort to refine monitoring of fluid status associated parameters, a simple rule of thumb algorithm was retrospectively compared with a sophisticated moving average convergence divergence algorithm to detect abnormal weight gain in telemonitored HF patients.20 While the moving average convergence divergence algorithm was much more specific than the rule of thumb algorithm in detecting weight gain, overall sensitivity was rather poor. As a significant number of episodes of worsening HF in that cohort were not associated with weight gain at all, the authors concluded that telemonitoring of weight gain alone may be of limited use for HF management.
Together, current evidence on home based telemonitoring strategies does not definitely point to consistent additional benefits above standard care for HF patients.
Thoracic impedance monitoring with novel ICD and CRT devices
Since the very first pacemaker was implanted into a patient in 1958, the use and complexity of cardiovascular implantable electronic devices has been ever expanding. Nowadays, these include conventional pacemakers, implantable cardioverter defibrillators (ICD) to treat life threatening arrhythmia and cardiac resynchronisation therapy devices (CRT-D) to restore interventricular synchrony. The latest generations of CRT-D and ICD devices are capable of monitoring thoracic impedance which has been shown to correlate well with pulmonary fluid status, and may provide an early warning of deteriorating HF (table 1). The correlation between LV filling pressures and intrathoracic impedance is inverse—that is, it decreases when there is evolving fluid accumulation within the thoracic cage. In HF patients, serial measurements of thoracic impedance have been demonstrated to reflect pulmonary fluid status and, importantly, predict HF decompensation even before the onset of symptoms.39 40 In a recent registry study, intrathoracic impedance was significantly correlated with N terminal proBNP and with mitral E wave deceleration time, but not with clinical HF score.28 In a large animal model of rapid pacing induced chronic HF, serial measurement of intrathoracic impedance with an implantable system effectively revealed changes in pulmonary congestion which were reflected by elevated LV end diastolic pressure.41 To facilitate interpretation of impedance data, algorithms based on impedance measurements are usually applied to compute a fluid index (FI). As impedance is influenced by the actual volume status and may vary substantially within HF cohorts, FI needs to be individualised in each HF patient, with baseline FI values determined during a period of clinical stability. Surpassing of a predefined FI threshold would indicate fluid overload and impending HF decompensation, and allow for swift intervention by HF physicians. In a case control study, patient management using an algorithm based on intrathoracic impedance monitoring (OptiVol; Medtronic Inc, Minneapolis, Minnesota, USA) has been shown to reduce hospital admissions for HF.25
The Medtronic Impedance in Diagnostics in HF Trial (MIDHeFT) in patients with NYHA classes III and IV HF showed a sensitivity of 77% for FI algorithms to detect hospitalisation for fluid overload.24 The Fluid Accumulation Status Trial (FAST) compared serial measurements of thoracic impedance with weight changes in 156 NYHA class II or III HF patients and implantable ICD or CRT-D, with a mean follow-up of 537 days.27 FAST demonstrated that impedance data were more sensitive than weight gain in predicting HF decompensation (76 vs 23%). The relatively low specificity improved when impedance data were combined with weight monitoring. The Program to Access and Review Trending Information and Evaluate Correlation to Symptoms in Patients With HF (PARTNERS-HF) study prospectively evaluated the utility of combined diagnostic algorithm including impedance data to predict HF hospitalisations in patients with NYHA classes III and IV HF, reduced LV EF, broad QRS and who had a CRT-D (Medtronic Inc). A total of 694 patients were followed for almost 12 months in this unblinded observational study. The impedance based algorithm identified a cohort at high risk of experiencing a HF event within the subsequent month.26 Importantly, there seems to be a link between patient reported HF self-care and the likelihood of an FI threshold crossing event.42
In most of these trials, the predefined FI algorithm led to a considerable number of false positive alerts and likely increased healthcare utilisation. This lack of specificity may present an obstacle to broader implementation of the technology into clinical practice. Current efforts to develop improved FI based algorithms demonstrated lower false positive alerts at similar sensitivity.43
The recent Sensitivity of the InSync Sentry OptiVol Feature for the Prediction of HF (SENSE-HF) study was a large prospective, multicentre, double blind study that evaluated an impedance based algorithm, OptiVol, in 501 NYHA class II and class III HF patients with CRT-D.29 Using OptiVol, the trial showed a low sensitivity of 42% and low positive predictive value of only 38% for future HF events. The Diagnostic Outcome Trial in HF (DOT-HF) was a large prospective phase IV RCT designed to test whether monitoring of intrathoracic impedance (OptiVol) could reduce morbidity and mortality in patients with chronic NYHA classes II–IV HF.44 All study subjects were implanted with an ICD or CRT-D capable of monitoring impedance (Medtronic Inc), and randomised to have all device based information (including audible alerts for preset fluid threshold crossings) available to patients and doctors (access group) or to a control group without that information.30 The primary endpoint was a composite of all-cause mortality and HF hospitalisation, and occurred in 48 of 168 (29%) patients in the access arm versus 33 of 167 (20%) in the control arm (p=0.063). Even if the trial was terminated early due to to low enrolment rates (only 336 of intended 2400 subjects were included), post hoc futility analysis deemed it unlikely that better recruitment would have changed overall outcome. The currently ongoing OptiLink-HF Study is another substantial study in the field. Approximately 1000 patients will be required to demonstrate a 30% reduction in the primary outcome (composite of all-cause death or cardiovascular hospitalisation).31
HF management based on invasive haemodynamic monitoring
Supranormal LV filling pressures are a hallmark and one of the principal haemodynamic abnormalities in HF decompensation. The relationship between cardiac pressures and HF events has therefore been the subject of longstanding interest and research. Pulmonary artery catheterisation (PAC) using thermodilution/Swan Ganz catheters has been the undisputed gold standard for invasive haemodynamic assessment. Early observational studies and registry data including patients with ADHF or cardiogenic shock after acute myocardial infarction have not been able to demonstrate beneficial effects of the use of PAC.45–47 However, most of those reports stem from the percutaneous coronary intervention (pre-PCI) (and pre-thrombolysis) era, and the lack of randomisation usually meant that the most seriously ill patients (with the worst prognosis) were more likely to undergo PAC. Accordingly, the importance of PAC in a contemporary HF setting is unclear. The Evaluation Study of Congestive HF and Pulmonary Artery Catheterisation Effectiveness (ESCAPE) study randomised 433 patients hospitalised with severe symptomatic HF to receive therapy guided by PAC derived haemodynamic data and clinical assessment versus therapy based on clinical assessment alone. ESCAPE showed that addition of PAC to clinical assessment did not affect overall mortality and hospitalisation.48 Significantly more patients in the PAC group (21.9 vs 11.5%) experienced an inhospital adverse event, but inhospital and 30 day mortality was not affected by the use of PAC. In contrast with the apparent lack of benefit of PAC guided therapy in ADHF, the relevance in chronic HF is unclear.
Implantable continuous haemodynamic monitoring devices
During the past decade, permanently implantable devices have emerged that provide accurate and timely long term haemodynamic data (table 1). Among these implantable continuous haemodynamic monitoring (ICHM) devices are those that chronically assess pressures in the right ventricle (RV), pulmonary artery and left atrium.32 49–51
Right ventricular pressure monitoring
In a feasibility study, 32 patients with HF received a permanent RV ICHM system (Chronicle; Medtronic Inc) similar to a single lead RV pacemaker. The ICHM delivered accurate RV pressure data over time that correlated well with LV filling pressures obtained from conventional PAC.49 In this cohort, hospitalisations before using ICHM data for clinical management averaged 1.08 per patient year and decreased to 0.47 per patient year (57% reduction; p<0.01) after integration of RV pressure data into the follow-up.32 The subsequent landmark Chronicle Offers Management to Patients with Advanced Signs and Symptoms of HF (COMPASS) trial sought to establish whether integration of RV ICHM derived pressures would reduce HF morbidity.33 COMPASS was a prospective, multicentre, randomised, single blind, parallel controlled trial and included 274 NYHA class III/IV HF patients with a previous HF hospitalisation, all of whom were implanted with the same ICHM as above. Subjects were randomised to an ICHM guided HF management strategy or control group follow-up without ICHM data available. ICHM guided HF management in COMPASS did not reduce HF related events compared with standard care which was probably the reason why the Food and Drug Administration has not thus far approved the technology.33 52 This surprising lack of efficacy deserves further discussion. Sample size calculations were based on an event rate of at least 1.2 per 6 patient months in the control group to show a 30% reduction in HF related events with 80% power. The trial, however, reported an event rate as low as 0.85 per 6 patient months in the control group, being further (non-significantly) reduced by 21% to 0.67 in the intervention group. It is noteworthy that the HF event rate in the control group decreased from 1.8 per 6 patient months (ie, by over 50%) after enrolment, probably driven by the very tight follow-up (at almost weekly intervals) which seems unrealistic to achieve in daily clinical practice.53 Even if technically underpowered to meet its efficacy endpoints, COMPASS provided novel important insights into the pathophysiological changes during decompensation in patients with HF with reduced and preserved EF.52 Pressure increases preceded HF related events by 3–4 weeks, and interestingly, no significant body weight changes were found in relation to HF events. Data on medication changes in relation to ICHM data are yet to be published and will further our understanding of HF management guided by RV haemodynamics. Very recently, the Reducing Decompensation Events Utilising Intracardiac Pressures in Patients with Chronic HF (REDUCE-HF) trial was halted with only 400 of the planned 1300 patients enrolled, due to problems with the pressure sensor leads seen in earlier studies.34 The HF event rate was even lower than in COMPASS, probably due to a healthier patient cohort (table 1), and the device had not led to reduced hospitalisation or other HF events when it was stopped.35
Left atrial pressure monitoring
A different approach to assess cardiac filling pressures is by implantation of a left atrial pressure (LAP) sensing system. HeartPOD (St Jude Medical Inc, Minneapolis, Minnesota, USA) was the first implantable LAP sensor to be reported.51 Similar to the RV ICHM system, HeartPOD consists of a small, pulse generator-like coil antenna and a lead carrying a septal anchor fixation system with a distal sensing diaphragm. The lead is implanted percutaneously and advanced across the atrial septum with the sensor depicting LAP signals. HeartPOD was previously shown to provide accurate and stable measurements in keeping with simultaneously obtained pulmonary capillary wedge pressure.50 In the recently published Haemodynamically Guided Home Self-Therapy in Severe HF Patients (HOMEOSTASIS) trial, 40 ambulatory patients in HF NYHA classes III and IV and a HF hospitalisation requiring intravenous therapy during the past 12 months underwent percutaneous implantation of the HeartPOD system.36 The study design was observational and prospective, with a follow-up of 25±19 (range 1–63) months. LAP was read twice daily, and both patients and clinicians were blinded to the LAP data the first 3 months after implantation. HF therapy was thereafter guided by LAP readings. HeartPOD derived LAP correlated highly with pulmonary capillary wedge pressure measured at 3 and 12 months (r=0.98, average difference of 0.8±4.0 mm Hg) under various loading conditions, and no important device related safety issues were raised. HOMEOSTASIS demonstrated encouraging significant reductions of LAP together with improvements in NYHA class and EF. Importantly, LAP guided management led to significant increases in β blocker and ACEI/angiotensin receptor blocker (ARB) use, as well as reduced use of diuretics. Subsequently, an additional 44 patients were implanted with HeartPOD. Recently published 48 month follow-up data in a total of 84 patients witnessed good long term sensor performance.54 The ongoing LAP Monitoring to Optimise HF Therapy trial (LAPTOP-HF; http://www.clinicaltrials.gov, NCT01121107; planned enrolment 730 patients) using HeartPOD or a similar LAP sensor combined with CRT-D (‘Promote LAP’) will evaluate whether HF related events are reduced in patients who are managed with the LAP management system versus those who receive the current standard of care.
Pulmonary artery pressure monitoring
A different device making use of ambulatory haemodynamic parameters is an implantable pulmonary artery sensor (CardioMEMS, Atlanta, Georgia, USA). The CardioMEMS sensor is a small yet ingenious device that is deployed in a distal pulmonary artery branch during routine right heart catheterisation, and delivers continuous pulmonary artery pressure (PAP) data.55 An apparent advantage over other ICHM devices is its small size and the lack of need for batteries or leads. The device was evaluated in the CardioMEMS Heart Sensor Allows Monitoring of Pressure to Improve Outcomes in NYHA Class III HF Patients (CHAMPION) trial.37 CHAMPION was a prospective, randomised, single blinded trial in patients with NYHA class III HF, irrespective of LV EF, and a previous hospital admission for HF. All patients were implanted with the ICHM device and then randomised to PAP guided therapy (n=270) or standard care (n=280). The primary efficacy endpoint was HF related events at 6 months, with pressure sensor failure and ICHM related complications as safety endpoints. After a mean follow-up of 15 months, in spite of a very low HF event rate (0.44 per 6 patient months in the standard care cohort), haemodynamic guided HF therapy substantially reduced HF related hospitalisations (to 0.31 per 6 patient months), significantly reduced PAP and improved quality of life. Integration of PAP data also led to significantly greater medication use. It is remarkable that background medical therapy at baseline was very good with over 90% and almost 80% of patients using β blockers and ACEI/ARBs, respectively, and furthermore, that patients with reduced versus preserved EF benefitted equally. The specific medication changes by which the encouraging results of the CHAMPION trial were achieved deserve further discussion.56 PAP guided HF therapy led to significantly greater utilisation of nitrates, ACEI/ARBs and β blockers.38 Diuretics were frequently adjusted, but not differently between groups. Extending positive signals from previous smaller, mostly observational, studies, CHAMPION was the first randomised trial sufficiently powered to detect and demonstrate effects on clinically meaningful endpoints.
Discussion
With the advent of technology allowing continuous monitoring of fluid status signals, early identification of pulmonary fluid accumulation in HF patients has moved within reach. Several devices have provided evidence that integration of fluid status is clinically feasible, with some encouraging results regarding endpoints.
First, remote or telemonitoring of HF symptoms integrating changes in body weight as a surrogate of fluid status has been extensively studied in recent trials. Even if some of the trials have suffered from low adherence to intervention, overall results have not demonstrated substantial benefit over and above standard HF care. Newer data indicate that body weight changes in HF patients are likely not sensitive (nor specific) enough signals to permit early identification of impending HF decompensation. This may be partly explained by fluid redistribution (not retention) which has been recently proposed as an important contributory mechanism.57
A different fluid monitoring concept is based on serial measurements of intrathoracic impedance, exploiting its inverse correlation with lung water content. A number of currently available CRT-D and ICD devices are capable of providing valid impedance derived fluid indexes. As indications for CRT-D and ICD devices in clinical HF care are ever expanding, additional fluid status signals could be obtained at ‘no extra cost’. Ongoing large scale clinical trials seek to establish whether HF management incorporating impedance data is superior to standard care.
Non-invasive impedance monitoring using impedance cardiography (ICG) may be suitable for patients who would not otherwise be considered for receiving an implantable device but more definitive outcome data are required to support their use in HF management.
Directly measured haemodynamic parameters as markers of intracardiac filling pressures constitute another promising avenue in fluid status monitoring, and a number of different devices are the subject of ongoing investigation.
Recent data support the potential for this approach in reducing HF related events even in cohorts with low event rates that already receive state of the art care. Few studies have included HF patients with preserved EF, which account for approximately half of ADHF hospitalisations.58 Data from CHAMPION and COMPASS studies point to a similar benefit for HF patients with preserved versus reduced EF.33 37 38
While the field advances rapidly, a number of issues remain to be resolved. Obviously, fluid status monitoring in HF by itself does not alter outcomes. In the clinic, decompensated HF and hypervolaemia are most frequently treated by increasing use of diuretics and/or vasodilators. Diuretic overuse might induce postural symptoms and azotaemia, and may be harmful in the long term.59 We still do not know from several published trials whether knowledge of fluid status data actually led to medication changes; specifically, to enhanced use of drugs known to reduce morbidity and mortality in HF. The medical community needs to learn what specific medication changes produced results superior to standard care, as recently reported for the CHAMPION study.38 56 Next, exactly how were the fluid status data translated into treatment decisions? Given the multitude of monitoring devices, unifying guidelines for intervention thresholds need to be established. We also need to learn more about managing ADHF presenting without concomitant weight gain, where volume redistribution rather than overload may be the pathophysiological abnormality. Finally, perhaps previous expectations of the devices to reduce risk in the range 20–30% have simply been too optimistic, given the very low event rates in some of the reported HF cohorts. Nevertheless, despite these ongoing issues, device based fluid status monitoring appears to represent a novel and promising tool in the management of HF.
References
Footnotes
Funding TGVL is supported by a post-doctoral research grant from South-Eastern Norwegian Health Authorities.
Competing interests None.
Provenance and peer review Commissioned; not externally peer reviewed.