Article Text

Download PDFPDF
Chronic heart failure management and remote haemodynamic monitoring
  1. Aaron M Wolfson,
  2. Michael Fong,
  3. Luanda Grazette,
  4. Joseph E Rahman,
  5. David M Shavelle
  1. Division of Cardiovascular Medicine, University of Southern California, Los Angeles, California, USA
  1. Correspondence to Dr David M Shavelle, Division of Cardiovascular Medicine, University of Southern California, Los Angeles, CA 90007, USA; shavelle{at}


Heart failure (HF) has a large societal and economic burden and is expected to increase in magnitude and complexity over the ensuing years. A number of telemonitoring strategies exploring remote monitoring and management of clinical signs and symptoms of congestion in HF have had equivocal results. Early studies of remote haemodynamic monitoring showed promise, but issues with device integrity and implantation-associated adverse events hindered progress. Nonetheless, these early studies established that haemodynamic congestion precedes clinical congestion by several weeks and that remote monitoring of intracardiac pressures may be a viable and practical management strategy. Recently, the safety and efficacy of remote pulmonary artery pressure-guided HF management was established in a prospective, single-blind trial where randomisation to active pressure-guided HF management reduced future HF hospitalisations. Subsequent commercial use studies reinforced the utility of this technology and post hoc analyses suggest that tight haemodynamic management of patients with HF may be an additional pillar of therapy alongside established guideline-directed medical and device therapy. Currently, there is active exploration into utilisation of this technology and management paradigm for the timing of implantation of durable left ventricular assist devices (LVAD) and even optimisation of LVAD therapy. Several ongoing clinical trials will help clarify the extent and utility of this strategy along the spectrum of patient with HF from individuals with chronic, stable HF to those with more advanced disease requiring heart replacement therapy.

  • heart failure

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


The societal and economic burden of heart failure (HF) continues to increase with an estimated 26 million people living with HF throughout the world.1 Costs are staggering with a majority coming from hospitalisation-related expenditures.2 Among patients with HF with reduced ejection fraction (HFrEF) there is an array of guideline-directed medical therapy (GDMT) and device therapy proven to improve quality of life (QOL) and longevity. Those with HF with preserved ejection fraction (HFpEF) have limited treatment options but nearly identical morbidity and mortality as those with HFrEF.3

Traditional strategies for avoiding HF hospitalisations (HFH) rely on targeting clinical signs and symptoms of congestion. Many telemonitoring strategies focusing on the remote management of congestion have been explored, but despite sophisticated approaches, success has been limited.4 In part, these approaches focus on clinical variables that lag behind the initial rise in intracardiac pressures.5–7 An alternative approach targets haemodynamic congestion before symptoms manifest. Insights gained from studies using the Chronicle device, an implantable single-lead system with a right ventricular pressure sensor, highlighted a rise in cardiac filling pressures weeks before an HFH.5

This review will focus on the current evidence regarding utilisation of the Federal Drug Agency (FDA)-approved implantable pulmonary artery (PA) pressure sensor, CardioMEMS HF System (Abbott Laboratories, Abbott Park, IL). Herein we will review the CardioMEMS Heart Sensor Allows Monitoring of Pressure to Improve Outcomes in New York Heart Association (NYHA) Class III Heart Failure Patients (CHAMPION) trial and its post hoc analyses, the subsequent ‘open access’ study and several investigator-directed studies. We will also discuss other considerations and future directions, and provide our recommendations regarding this recently developed HF management strategy.

Remote haemodynamic monitoring

The era of remote haemodynamic monitoring (RHM) began with the Chronicle device (described above) which allowed for continuous haemodynamic monitoring.8 The device calculated an estimated PA diastolic (ePAD) pressure that served as a surrogate for left-sided filling pressures. The feasibility study was a prospective, observational, historical control (each patient served as their own control using data collected at two separate times) study that included 32 NYHA class II–III patients implanted with the device.8 Compared with historical controls, there was a 57% reduction in HFH once clinicians used RHM data following the initial 9-month observational and validation period. This study provided the basis for the pivotal Controlled Trial of an Implantable Continuous Hemodynamic Monitor in Patients with Advanced Heart Failure (COMPASS-HF),9 which was a prospective, single-blind, randomised controlled trial of 274 NYHA class III–IV patients. Although COMPASS-HF did not meet its primary endpoint, a retrospective analysis demonstrated a 36% reduction in HFH.9 Several observations from the COMPASS-HF trial provide the basis for the concept of pressure-guided therapy (figure 1).5 First, HF-related events are characterised by a gradual increase in cardiac filling pressures. Second, following an HF-related event, pressures are ‘reset’ to a lower level (figure 1). Third, patients with HFpEF reside at slightly lower filling pressures (figure 1) than patients with HFrEF. Fourth, the rise in filling pressures antecedent to an HF event occurs over several weeks. Fifth, at an ePAD cut-off of 18 mm Hg, the risk for HFH significantly increases (figure 2).10 Reducing Decompensation Events Utilizing Intracardiac Pressures in Patients with Chronic Heart Failure trial included NYHA class II–III patients using the Chronicle device and planned to enrol 1300 patients, but was terminated after only 400 patients had been enrolled due to pressure-sensing lead failure.11

Figure 1

Trend of estimated pulmonary artery diastolic (ePAD) pressure over time in patients enrolled in COMPASS-HF study. Blue indicates patients with systolic heart failure, heart failure with reduced ejection fraction; red indicates patients with diastolic heart failure, heart failure with preserved ejection fraction. (A) Patients with heart failure-related event have a trend up in ePAD pressures prior to the heart failure-related event. (B) Patients with no heart failure-related event have stable ePAD pressures over time. (From figure 2, Bourge et al 9 Permission required).

Figure 2

Probability of heart failure event within 6 months in patients enrolled in COMPASS-HF study. The risk of a heart failure event increases above an estimated pulmonary artery diastolic pressure cut-off of 18 mm Hg. HF, heart failure. (From figure 2, Stevenson et al 10 Permission required).

The next approach to measure intracardiac pressures involved a sensor (HeartPod) implanted within the interatrial septum that directly measured left atrial pressure.12 Correlation between left atrial pressure and pulmonary capillary wedge pressure was excellent.12 Left atrial pressure measurements were available to the patient using a patient advisory module that provided physician-specified, tailored dosing of medications based on pressure readings. The Hemodynamically Guided Home Self-Therapy in Severe Heart Failure Patients trial was a prospective, observational, open-label study evaluating the safety and efficacy of the HeartPod system in a cohort of 40 NYHA class III–IV patients.13 Compared with historical controls, there were significant reductions in HF-related events and filling pressures. The subsequent randomised Left Atrial Pressure Monitoring to Optimize Heart Failure Therapy (LAPTOP-HF)14 trial was stopped prematurely by the data and safety monitoring board due to implant-related trans-septal complications requiring pericardiocentesis or surgical repair.

The current approach to RHM involves the CardioMEMS HF System (figure 3) that uses a small, battery-free, microelectromechanical system sensor mounted on two nitinol coils. The device is implanted into a small branch of the left PA via the femoral vein in a 30 min procedure. Patients transmit PA pressures from home by lying down on a pillow containing a radiofrequency antenna that powers the device. The PA pressure readings are sent on a daily basis to a secure website ( for healthcare provider review. Knowledge of the absolute PA pressure, trends in PA pressures and established pressure goals allows for titration of medications prior to the onset of HF symptoms.

Figure 3

CardioMEMS HF System. (A) CardioMEMS sensor. (B) CardioMEMS sensor is implanted into a distal branch of the left pulmonary artery. (C) Patient is instructed to take daily pressure readings from home using the home electronics unit. (D) Pressure waveforms transmitted from the monitoring system to the database are immediately available for healthcare provider review. (E) Transmitted information consists of pressure trend information and individual pulmonary artery pressure waveforms. (From figure 1, Abraham et al 15 Permission required).


The CHAMPION trial15 was a prospective, single-blind, randomised controlled trial that enrolled 550 NYHA class III patients, irrespective of ejection fraction (EF), with an HFH within the preceding 12 months. The treatment arm was managed using daily PA pressure data plus standard care versus standard care alone in the control arm (ie, treating physicians were not given access to daily PA pressure data). The device was shown to be safe with no observed pressure sensor failures. Device-related or system-related complications occurred in only eight patients (1%). All patients remained in their randomisation group until the last patient completed 6 months of randomised follow-up, thus allowing for a longer randomisation period for additional analysis. The trial met its primary endpoint with a 28% reduction in HFH at 6 months and 37% reduction in HFH for the entire randomised follow-up (table 1).

Table 1

Summary of the CHAMPION trial and open-access registry evaluating the CardioMEMS HF System

Similar to the LAPTOP-HF, CHAMPION included a detailed haemodynamic-guided care algorithm with a suggested medication titration scheme based on PA pressure ranges that stratified a patient’s status as hypovolemic, optivolemic or hypervolemic16 (figure 4). Standardised management consisted of a hierarchical strategy focusing first on adjustment of diuretic therapy when patients were categorised, according to prespecific pressures goals, as hypervolemic or hypovolemic. If pressures did not return to an optivolemic value, changes to vasodilators were made. Once stable in the optivolemic range, providers were instructed to maximise GDMT as tolerated. To quantify cumulative haemodynamic changes, baseline to 6-month follow-up change in mean PA pressure was quantified as a pressure-time product (mm Hg-days) and referred to as the area under the curve (AUC) relative to the baseline mean PA pressure. AUC was significantly lower in the treatment versus control group (table 1).

Figure 4

Treatment guidelines based on low (hypovolemic), normal (optivolemic) and elevated (hypervolemic) pulmonary artery pressures. ACC, American College of Cardiology; AHA, American Heart Association; IV, intravenous. (From figure 1, Costanzo et al 23 Permission required).

Open access: sustained efficacy

A subgroup of patients participating in the CHAMPION trial (n=347; 63%) were followed beyond the original study period. During this ‘open-access’ period, the former randomised treatment group maintained an HFH rate that was essentially unchanged compared with HFH rates seen during the randomised portion of the trial (table 1).17 Notably, the former control group had significant reductions in HFH rates, similar in magnitude to that seen in the original randomised treatment group, once daily pressures became available to providers (table 1). In other words, outside of the clinical trial setting, prior treatment effects were maintained in the former treatment arm while the former control group gained benefits similar to those seen in the original treatment group. These findings emphasise the results of the CHAMPION trial.

Subgroup analyses of CHAMPION trial

A number of subgroup analyses of the CHAMPION trial data set provide a nuanced profile of different patient phenotypes enrolled in the study18–22 (table 2). In an analysis looking at EF, HFH rates were significantly lower in the treatment arms of all subgroups with a greater numerical benefit seen in the EF>40% subgroup (table 2).19 Benza et al 21 evaluated 314 patients (57%) with pulmonary hypertension (PH) associated with left heart disease and found that those with PH had higher HFH rates than non-PH patients (0.77/patient-years vs 0.37/patient-years; HR 0.49, 95% CI 0.39 to 0.61, p<0.001). Additionally, within different PH subgroups, a haemodynamic-guided strategy still reduced HFH compared with usual care (table 2). In another study, patients with chronic obstructive pulmonary disease (COPD) (n=187) had both lower HF and respiratory-related hospitalisations (table 2),18 but no difference in respiratory-related hospitalisations in the non-COPD group.18

Table 2

Summary of subgroup analyses from the CHAMPION trial

Costanzo et al 23 analysed pressure-linked medication changes made throughout the randomisation period of the CHAMPION trial. Within the treatment group, investigators were advised to follow a pressure-driven medication titration algorithm defined by specified PA pressure goals as discussed earlier (figure 4). There were more than double the medication changes in the RHM group (table 2). Additionally, there were more frequent increases and decreases in diuretic doses in the RHM group compared with the control arm, but no deleterious effect on renal function.23 Similarly, hydralazine and nitrate therapy changes were more frequent in the RHM group. Doses of neurohormonal antagonist therapy were significantly higher in the treatment arm at the end of the study period but unchanged in the control arm. Lastly, the frequency of diuretic and vasodilator changes correlated with baseline PA diastolic pressures.

Givertz et al 20 analysed the subset of patients with HFrEF (n=456 patients) on either ACE inhibitors (ACEI)/angiotensin receptor blocker (ARB) or beta-blocker (BB) or dual ACEI/ARB and BB therapy. Treatment groups on either ACEI/ARB or BB or both agents had a significant reduction in HFH (table 2). Unadjusted all-cause mortality was lower in the treatment group (table 2) and remained significant after adjustment for either vasodilator or neurohormonal doses. Numerically, the magnitude of mortality reduction for patients on both ACEI/ARB and BB was greater than being on a single agent; whether this finding suggests an undertreated patient subgroup or is instead a marker of more advanced disease due to hypotension and/or advanced renal disease was not established. While mortality was not a prespecified endpoint in CHAMPION, these findings track the observations of Zile et al 24 demonstrating that greater short-term reductions in filling pressures are associated with lower mortality. In patients with HFrEF, doses of GDMT were higher in the treatment group at the end of the study period.20 As the authors point out, this analysis highlights a potential synergy between combined management of intracardiac pressures and neurohormonal blockade leading to a further reduction in morbidity and mortality.20

Real-world clinical experience

Desai et al 25 performed an analysis of Medicare beneficiaries with HF, not included in CHAMPION, comparing HFH rates before and after CardioMEMS sensor implantation with each patient serving as their own historical control. The study included 1114 patients with 6 months of continuous data and 480 patients with 12 months of continuous data. In both cohorts, HFH rates were significantly lower after sensor implantation (table 3). Data pertaining to medication changes were not linked to PA pressure trends. Nevertheless, the aforementioned study by Costanzo et al 23 established a link between an algorithm-based medication titration scheme, intracardiac pressure reduction and a decrease in HFH rates. Inpatient implantation of the sensor did not impact these findings.25 These findings support those of the CHAMPION trial but do not replace the need for additional clinical trials. A small, single-centre experience describing outcomes in a private practice, hospital-based setting demonstrated a significant reduction in HFH and NYHA class with a concomitant increase in 6 min walk times and QOL measures26 (table 3).

Table 3

Summary of ‘real-world’ studies evaluating the CardioMEMS HF System

Heywood et al looked at the first 2000 patients following FDA approval, referred to as the ‘general-use cohort’, implanted with the sensor and compared cumulative pressure changes to the CHAMPION cohort (table 3).27 The AUC was significantly lower in the general use cohort compared with the CHAMPION treatment group; neither sex nor baseline EF impacted this finding. Overall, patients with higher baseline pressures had larger, more negative AUC values; not surprising as the magnitude of pressure reduction has been shown to be positively correlated with baseline pressures.28 While pressure reduction is achievable outside of the original clinical trial, the AUC method is a relative metric and may not be the most appropriate surrogate endpoint to track. Other metrics are needed, perhaps looking at cumulative mm Hg-days, without respect to a baseline value, or the percentage of time spent within a goal pressure range. Lastly, pressure transmission adherence, outside of a clinical trial setting, remained high in this study with a mean time of 1.27 days between pressure transmission and a >98% weekly adherence rate.

In a small retrospective study (table 2) of 27 patients from the CHAMPION trial, Feldman et al 29 found that while still within the randomised period of the trial, patients in the treatment arm had a trend towards earlier left ventricular assist device (LVAD) implantation and less renal insufficiency than those in the control arm. In this light, the authors discuss the concept that haemodynamic monitoring may help identify when ‘medical futility’ has been reached as patients progress to more advanced HF. Additionally, the magnitude of pressure reduction after LVAD was greater for patients in the treatment arm compared with the control arm. These findings are provocative and worthy of prospective clinical trials.

At the 2018 American College of Cardiology Scientific Sessions, Abraham et al 30 presented a retrospective study of 1087 patients implanted with the CardioMEMS device outside of the CHAMPION trial with a propensity-matched control group obtained from Centers for Medicare and Medicaid Service (CMS)-derived administrative claims data (table 3). The primary endpoint was all-cause mortality and HFH at 12 months. The RHM group had a 30% reduction in all-cause mortality and 24% reduction in HFH. While these findings are retrospective, it is another example suggesting a synergistic effect between neurohormonal blockade and chronic, aggressive control of intracardiac filling pressures.

Practice guidelines

In European Society of Cardiology practice guidelines, RHM has a class IIb recommendation in patients with symptomatic HF and a recent HFH.31 There is no mention of RHM in the current American College of Cardiology/American Heart Association/Heart Failure Society of America guidelines.

Device safety

Data from the CHAMPION trial demonstrate a low and tolerable rate of adverse outcomes (table 3). Vaduganathan et al 32 described the reported postmarket safety outcomes in 5500 patients implanted with the CardioMEMS HF System using reports from the Manufacturer and User Facility Device Experience database that displayed trends similar to those seen in the CHAMPION trial (table 3). More safety data will become available with the completion of several ongoing clinical trials.

Ongoing clinical trials

The Hemodynamic-GUIDEd Management of Heart Failure (GUIDE-HF) trial33 is a prospective North American trial at 140 sites with a planned enrolment of 3600 NYHA class II, III and IV patients with HF, regardless of EF with either a history of an HFH in the preceding 12 months and/or an elevated N-terminal pro b-type natriuretic peptide (NT-proBNP) or BNP (table 4). The primary composite endpoint will include HFH, mortality and non-HFH evaluations requiring intravenous diuretic therapy. The Investigation to Optimize Hemodynamic Management of Left Ventricular Assist Devices Using the CardioMEMS Pulmonary Artery Pressure Sensor in Advanced Heart Failure is a small, observational, multicentre study with an anticipated enrolment of 100 patients.34 The study will evaluate the impact of RHM on functional status, QOL and hospital readmissions in patients with LVADs. It will also explore the impact of medication and pump speed changes on intracardiac pressures. The 1200 patient Post-Approval Study (USA) with similar inclusion and exclusion criteria to the CHAMPION trial recently completed enrolment and will provide additional real-world experience with an estimated study completion date in June 2020.35

Table 4

Summary of ongoing clinical studies evaluating the CardioMEMS HF System

Additional considerations

The current strategy for RHM is based on targeting intracardiac pressures. Undoubtedly, limited clinical resources will impede uptake of this technology and future iterations will need to incorporate other cardiovascular variables to add precision to patient management with the goal of developing a reliable automated alert system. Using the COMPASS-HF trial data set, Adamson et al tested an algorithm to predict future HF-related events based on serial pressure increases in ePAD.6 Their algorithm had a sensitivity of 83.1% and an adjusted false-positive rate of 1.4/patient-year.6 More sophisticated management approaches should be pursued. For example, heart rate obtained from RHM has yet to be studied. A post hoc analysis of the Ivabradine and Outcomes in Chronic Heart Failure trial36 found that time-updated heart rate improved risk stratification for adverse cardiovascular events compared with baseline heart rate alone.37 Additionally, novel waveform analytical methods of aortic pressure tracings quantifying ventricular-vascular coupling38 39 may identify actionable changes in cardiovascular dynamics that precede a rise in filling pressures. Application of such methods to remotely obtained PA waveforms may have added clinical value. Similarly, a recent study found that right ventricular pulsatile load quantified as pulmonary artery elastance (Ea) was a stronger predictor of mortality compared with conventional markers of pulmonary vascular disease.40 Given the importance of preserved right-sided ventricular and vascular function in heart replacement therapy, remotely tracking Ea may prove useful for determining when to pursue a durable LVAD and/or heart transplantation. Should RHM platforms provide a reliable cardiac output, calculation of Ea remotely would be possible. Additionally, remote optimisation of LVAD settings may become feasible with knowledge of intracardiac pressures. While purely speculative, research exploring these ideas may augment the uptake and utility of an RHM strategy in both the short and long-term care of the patient with HF. Lastly, studies of cost-effectiveness are encouraging.41–43

Summary of recommendations

Based on current evidence, we recommend consideration of RHM in NYHA class III patients with (1) an unstable clinical course despite best attempts at optimal medical management, (2) adequate renal and blood pressure reserve to permit frequent medication changes, (3) the ability to reliably transmit pressure data and (4) the ability to readily respond to provider-initiated alterations in medical therapy and laboratory evaluations. Ideally, patients would be cared for at experienced HF centres having adequate infrastructure to implement, respond to and coordinate the iterative patient care needed for an effective RHM strategy. Our recommendations are consistent with the approach suggested by Ollendorf et al.44


HF is a clinical syndrome marked by both neurohormonal disarray and pathophysiologic haemodynamic stress. While neurohormonal blockade therapy is well established, the current available data regarding RHM strategies suggest a synergistic effect with aggressive reduction of intracardiac pressures to physiologic ranges. The GUIDE-HF study will help answer the question whether RHM holds value as another pillar of HF therapy.



  • Contributors The contributions of the authors are as follows: DMS: conception and design, drafting of the manuscript, revising the manuscript critically for important intellectual content and final approval of the submitted version. AMW: drafting of the manuscript, revising the manuscript critically for important intellectual content and final approval of the submitted version. LG: conception and design, revising the manuscript critically for important intellectual content and final approval of the submitted version. MF and JR: revising the manuscript critically for important intellectual content and final approval of the submitted version. None of the authors had any writing assistance.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests DMS, LG, JR and AMW are on the speakers' bureau for Abbott Vascular, the manufacturer of the CardioMEMS HF System. DMS, LG, JR and MF receive research support from Abbott Vascular.

  • Patient consent Not required.

  • Provenance and peer review Commissioned; externally peer reviewed.