GRAPEVINE, TX — Risk stratification of patients with chronic heart failure and a defibrillating cardiac resynchronization therapy (CRT-D) device may be improved by combining natriuretic peptide levels with a proprietary index composed of six common physiologic markers picked up at remote monitoring, suggests an analysis based on a 900-patient cohort[1]. Importantly, the strategy has been tested only for feasibility, not yet in a prospective trial.
Still, investigators hope that the HeartLogic (HL) score, based on an algorithm developed by Boston Scientific for use in implantable defibrillators (ICDs) and pacemakers, can help guide medical therapy in heart-failure patients with devices.
The HL score by itself independently predicted HF risk, but not as well as the score added to natriuretic peptide levels. Combined with N-terminal pro-BNP (NT-proBNP) levels, it identified a large subgroup of patients “at exceptionally low risk of events,” according to Dr John P Boehmer (Milton S Hershey Medical Center and Penn State College of Medicine, Hershey, PA).
That subgroup amounted to 53% of the cohort in the post hoc analysis based on patients from the prospective Multisensor Chronic Evaluation in ambulatory Heart Failure Patients (MultiSENSE) study[2].
In practice, clinicians remotely monitoring their implanted-device patients would decide whether to use the HL score alone or in conjunction with other parameters, Boehmer told theheart.org | Medscape Cardiology.
It might be used selectively—for example, in “patients that you are more concerned about and want to monitor more closely,” according to Boehmer, who presented the biomarker analysis from the previously published MultiSENSE study here at the Heart Failure Society of America 2017 Scientific Meeting.
The algorithm was recently approved in Europe and the US for use in
Resonate ICDs and cardiac resynchronization therapy (CRT) devices and will be commercially available later this year, Boston Scientific said.
In the MultiSENSE study, all patients had been implanted with the company’s Cognis defibrillating CRT devices and were followed on the company’s LATITUDE remote monitoring system. But it can be used in standard pacemakers as well, Boehmer noted.
Six Familiar Components
The HL score is derived from data on the first and third heart sounds, thoracic impedance, respiration rate, ratio of respiration rate to tidal volume, heart rate, and patient activity level as sensed by the implanted device.
Clinicians decide on a specific score, a level between 6 and 40, as the cut point that defines the patient’s risk status, Boehmer explained in an interview. In MultiSENSE, that threshold was set at 16, which was seen to show a good balance between sensitivity and specificity, he said.
When the score rose to 16 in continuous monitoring, patients entered an “alert state.” The score had to recover to 10 or less before the patient was considered to be out of the alert state, Boehmer said.
In practice, data for creating the score would be uploaded routinely along with other remote-monitoring indicators. For patients entering an alert state, there’s a large time window in which to respond with possible intensification of medical therapy, Boehmer said, and so it usually wouldn’t be considered urgent.
The upload should be performed “maybe twice a week,” he recommends. “The way we envision it, this is something that will be taken care of during business hours.”
In the primary MultiSENSE analysis, the median time between the beginning of alert states and any later HF event was 34 days; 89% of HF events occurred at least 2 weeks after alert onset.
“It’s Easier to Predict Than Prevent”
As assigned discussant after Boehmer’s presentation, Dr Lynn Warner Stevenson (Brigham and Women’s Hospital, Boston, MA), observed that in medicine “it’s easier to predict than prevent,” and that it isn’t clear how one responds to a changed HL score.
“In this case, the signal is quite complex,” even though its components are well recognized signs with relevance to heart failure. And the signal “has to have kinetics, so that you can quickly tell if your action has in fact moved you out of risk,” she said.
“The truth is, we don’t know what happens when we try to treat a HeartLogic index score. Because we haven’t tried it,” said Boehmer. The MultiSENSE trial’s purpose was to derive and validate the score and retrospectively determine its ability to predict HF events.
Still, he said, “We do know observationally from looking at heart-failure hospitalizations within MultiSENSE that when patients are treated, their HeartLogic index does drop, and it drops fairly abruptly and in many cases goes back to baseline.”
Sometimes it would take a while before the score receded to baseline, but its trajectory would reverse “relatively early,” he added.
“We have that from the observational data, but we don’t yet have it from a true real-time prospective intervention, and that’s going to be the first phase of MANAGE-HF.” That’s the trial, now getting under way, that will randomize an estimated 2700 remotely monitored device patients to have the HL function on or off and follow them for mortality and HF hospitalization.
“I’m convinced that you are identifying a group at high risk,” added Stevenson about using the HL score, but “I’m not sure it’s totally fair to compare it with [NT-proBNP], because you only measured [NT-proBNP] at baseline, and you measured HeartLogic daily.”
In practice, Boehmer explained in an interview, natriuretic peptides are not measured daily. They might be measured more often than once a year, or they may not, but “at whatever rate you choose to get it, it won’t be as frequent as you are with HeartLogic. So we thought the comparison was still fair from a clinical perspective, to look at a baseline NT-proBNP and compare it with a HeartLogic score that’s changing day by day.”
Both Markers Independently Predictive
MultiSENSE followed 900 patients with CRT-D devices and their HL function turned on, mostly in NYHA class 2 to 4 (94% were NYHA class 2–3), for up to a year for the end point of admission with HF as the primary cause or unscheduled administration of IV diuretics or ultrafiltration. Their mean baseline LVEF was 29.6%.
In multivariate analysis, entering the alert state with an HL score >16 independently predicted HF events more strongly than other independent predictors, including baseline NT-proBNP, NYHA class, and history of atrial fibrillation or renal disease.
Predictive Value (Event Rate-Ratio*) of Separate Heart Failure Risk Markers in MultiSENSE Post Hoc Analysis
| Parameter | ERR (95% CI) | P |
|---|---|---|
| HeartLogic score >16 | 5.91 (3.63-9.62) | <0.0001 |
| NT-proBNP >1000 | 2.40 (1.29-4.46) | 0.0055 |
| History of AF/AFl | 2.34 (1.32-4.14) | 0.0036 |
| History of renal disease | 2.30 (1.40-3.79) | 0.0010 |
| NYHA class 3-4 | 2.17 (1.21-3.88) | 0.0094 |
AF/AFl=atrial fibrillation or atrial flutter
ERR=Event rate ratio
*Adjustment included age, gender, BMI, systolic and diastolic BP, total hemoglobin, history of MI, potassium, creatinine, and blood urea nitrogen
Using 16 and 1000 pg/mL as risk cut points for the HL score and NT-proBNP levels, respectively, a high HL score was seen to identify a higher-risk subset of patients even among those with low natriuretic peptides. Those with a higher HL score and higher NT-proBNP levels showed 50 times the risk of HF events compared with patients with lower scores and levels.
Risk Stratification by Natriuretic Peptides with HeartLogic Score: Events/Patient-Year (Event-Rate Ratio)
| Risk Indicator | HeartLogic Score <16 | HeartLogic Score >16 |
|---|---|---|
| NT-proBNP <1000 | 0.02 (1.00)* | 0.47 (23.5) |
| NT-proBNP >1000 | 0.16 (8.0) | 1.00 (50.0) |
*Comparator
Boehmer thinks the HL index, although it might in practice be preferentially used in higher-risk patients, “is going to be predictive regardless of initial risk going in.”
Although its predictive power was not formally analyzed by baseline risk, for example by NYHA class, his observation was it “performed very well” on the study’s subset of patient with an indirect marker of high risk: a heart-failure hospitalization within a year of study entry.
MultiSENSE was funded by Boston Scientific. Boehmer discloses receiving research funding from Abbott, Merck, and Amgen and consulting or serving on an advisory board for Boston Scientific, Abbott, and Medtronic. Stevenson discloses receiving research funding from Novartis and consulting or serving on an advisory board for Abbott.
Follow Steve Stiles on Twitter: @SteveStiles2. For more from theheart.org | Medscape Cardiology, follow us on Twitter and Facebook.
Tidak ada komentar:
Posting Komentar